New Posts

Does Physx Support "Culling collision"?

Thu, 03 Nov 2011 02:18:04 -0700

Hi,

I created thousands of objects in the scene which was exported to .nxb file by using 3dmax plug-in tools.

These objects will never collide with each other, they only collide with my character model(pulling or pushing .etc ).

I found if i load the nxb file,it will be very very slow .yes ,the objects are too many.

so i wondering can physx only calculate collision detection between character model and nearby objects?just like Frustum Culling in OpenGL.

how about separated these objects into sevral nxb files,loaded the file when is near the character model.

or filter collision?

Thanks.

Showing one left and right frame at display

Wed, 02 Nov 2011 12:07:48 -0700

Hi everybody,

I am very new at devoloping with Nvidia 3d vision. I must show my own left and rigth frame, Could you please explain how can I do this.
I appreciate vey much if you provide an answer as detailed as possible.

createActor returns NULL

Wed, 02 Nov 2011 09:06:10 -0700

I don't think the fix is as obvious as most solutions I've read.

I'm working in a rather large codebase, using a custom viewer instead of physXViewer, and having trouble getting physX to work. The problem seems to come down to createActor returning a null value.

As a sanity check I used the following example code:
http://mmmovania.blogspot.com/2011/04/phys...e-bouncing.html

I can use the source code for this by itself, moving the appropriate dll's to where the solution file is. Using VS2008 I create a new instance of SimplBox and everything works fine. I put the relevant global vars into InitializePhysX(). I placed this entire method into my code, as a debug flag.

Although nothing has changed as far as physX content, createActor fails to return a non-NULL value. The hierarchy looks a bit like this

Solution 'runtime'
- Visualization (physX content here)

Solution 'tools'
- Viewer (Calls Visualization; PhysXCore, cudart, etc dll's placed in folder here)

Any suggestions here as to what is different when my .exe is in a different solution than the physX method itself?

build PhysX 3.X library for iOS

Tue, 01 Nov 2011 03:42:39 -0700

hi，is there any example build Mac OSX SDK for iOS support ?

i saw the udk have declared that they've port the physX for iOS platform
http://physxinfo.com/news/4707/january-udk...-physx-for-ios/

is it possible uses the 3.X SDK to build lib for iOS platform ?

thanks for any advise!

nVidia OpenGL Lost Connection

Tue, 25 Oct 2011 15:37:33 -0700

Hi!
Im running Windows 7 32 bit and when i play games Truck and Trailers or F1 2011 codemasters the game crashes and nVidia openGL lost connections....error code 8 pop up window! I try to install older or new drivers but nothing helps.I try to uninstall drivers in control panel add/remove and with driversweeper befor to install any drivers but nothing helps.Before i was play T and T and F1 2011 without any problems but right now can`t play anymore.Is my card nVidia GeForce 8800GT damaged or is problem with drivers?With older drivers 260.99 just crashes the game and freeze white screen but no pop up OpenGL lost connection...

Motherboard is MS 7369
Processor AMD 64x2 Dual Core 3800+
Graphic Card NVidia GeForce 8800GT
Memory 2GB DDR2

Sorry about my english!

Softbody Contact Report

Tue, 25 Oct 2011 01:25:09 -0700

Hi I am using PhysX V.2.8.1. In this version,

Is there any way to get contact report for Softbody collided with rigid body?
And can we get force vector applied to vertex of the softbody?

Thank you.

rigid objects immersed in moving water

Sun, 23 Oct 2011 13:29:45 -0700

I'm trying to simulate a physical process involving small rigid objects immersed in moving water.
The capacity of the water tank is one cubic meter and the rigid objects size varies between 5 mm and 20 mm.

For some unknown reason, probably due to my lack of experience using PhysX, I'm struggling to set the parameters required to obtain the realistic simulation of water through the SPH.
As for the interaction between rigid objects and water, the corresponding shape flag as well as the fluid flag were both set.

Furthermore, after adjusting the value of the "collisionResponseCoefficient" to a value other than 0, rigid objects in contact with the fluid are spat at high speed out of the tank.
Can anybody give me a hand in here�

Are those problems related to the difference in value between the water density (1000 kg/m3) and rigid objects (2000-7000 kg/m3)?
What is the most appropriate way to include the interaction forces between the fluid and rigid objects (drag force and buoyancy)?

Thank you for any help you can give me.

VRAM size incorrect

Fri, 21 Oct 2011 00:03:42 -0700

Problem = get VRAM size is not correct, get in here is 14MBYTE, but actually should be 512MBYTE (VbeInfoBlock.TotalMemory)
CARD OF VGA = GT440
M/B = INTEL P45
SYSTEM = DOS

CODE (VbeInfoBlock.TotalMemory):

/* SuperVGA information block */
struct
{
uchar VESASignature[4]; /* 'VESA' 4 byte signature */
uint VESAVersion; /* VBE version number */
uchar far *OEMStringPtr; /* 4byte , Pointer to OEM string , bankSwitch(bx,dx) */
ulong Capabilities; /* 4byte , Capabilities of video card */
ulong VideoModePtr; //unsigned far *VideoModePtr; /* 4byte , Pointer to supported modes */
uint TotalMemory; /* 2byte , Number of 64kb memory blocks */

//VBE 2.0
short OemSoftwareRev;
char *OemVendorNamePtr;
char *OemProductNamePtr;
char *OemProductRevPtr;
char reserved[222];
char OemData[256];

} VbeInfoBlock;

void main (void)
{
uchar bb1,bb2;
uchar bank1;
time_t start_time;
ulong data;
union REGS regs;
struct SREGS sregs;
unsigned far *p;
uint index;

clrscr();

// check vesa version
if (!getVbeInfo())
{
printf("No VESA VBE detected\n");
exit(1);
}
else
{
printf("\n name = %s " , VbeInfoBlock.VESASignature ); // VESA
printf("\n version = %x h " , VbeInfoBlock.VESAVersion ); // 300 h
ll1 = VbeInfoBlock.TotalMemory ;
vga_total_dram = (ll1*64L)/1024L;
printf("\n total MEM = %ld MB " , vga_total_dram); // 14 MBYTE
};
}

// -------------------------------------------------------------------------
/* Get SuperVGA information, returning true if VBE found */
int getVbeInfo()
{
union REGS in,out;
struct SREGS segs;
char far *VbeInfo = (char far *)&VbeInfoBlock;
in.x.ax = 0x4F00;
in.x.di = FP_OFF(VbeInfo);
segs.es = FP_SEG(VbeInfo);
int86x(0x10, &in, &out, &segs);
return (out.x.ax == 0x4F); // 4f is success
}

Non-Standard Mode 1366x238x32

Wed, 19 Oct 2011 10:55:00 -0700

Hi,
We have got this new Touch-Panel from 3M that has 1366x238 as native resolution. The NvCpl API returns 1366x768 as the native resolution and can't set it to 1366x238. This implies that I can't calibrate the Touch-Panel. My question is: How to force the 1366x238 non-standard resolution?

OS: WinXp
Drivers: v280.26
Card: GT 430

I have tried modifying the nv4_disp.inf to include 1366x238x32 in NV_Modes without success.

Any help would be appreciated.

Thanks,
MathMad

Traingle Mesh of a plane collision problem

Wed, 19 Oct 2011 04:22:44 -0700

Hi
I have created a triangle mesh and it's a plane.

But in visual debugger it visually display only the lower surface of the plane. If I move camera up and see the down side it won't display the upper surface.
please see the image

I have noticed this is same for the collisions too.

When I drop a object it won't collide the plane. Instead it go through the plane and collide the floor bounce back and collide the back face of the plane.

Could you please provide a hint to overcome this issue?

Thank you

Softbody Interpenetration Problem

Sun, 16 Oct 2011 18:31:09 -0700

I am new to this forum. I am currently facing some problem with Softbody.
I create softbody and rigid body in a scene. When the rigid body is trying to touch the softbody, interpenetration problem occurs.
As my understanding, PhysX detect collision between Softbody and Rigid Bodies with Vertex to Face (Point to Triangle) Collision Detection method.
So, if the distance between each vertices of the softbody is too far and Faces of the rigid body are quite small, it miss collision.
Am I right? Any solution to this problem? Kindly discuss with me, please.
Thank you.

GPU acceleration of fluid

Sat, 15 Oct 2011 15:22:52 -0700

Hello,

How to enable gpu acceleration of fluid?

When I try to do following I get error "attempt to create physx hardware scene when physx hardware not accessible"

CODE

fluidDesc.flags |= NX_FF_HARDWARE;

I have GeForce 540M and intel something (optimus).
Is this a problem with two different GPUs

cheers

Working with XNA

Sat, 15 Oct 2011 06:50:32 -0700

As the description tells, this is a complete shot in the dark so I do apologise if this is in the wrong area of the forums but it is something I can find little to no information on elsewhere. I am beginning a small project within a university course within which we must create a game within XNA. I am curious if Nvidia PhysX and APEX (Cloth, particles etc), will work within XNA and if so how one might begin implimenting such things.

Any help your able to provide, be it link or just helpful tips would be greatly appreciated. Thanks in advance.

why I Can't createFluid

Fri, 14 Oct 2011 06:16:31 -0700

Hi, everybody, I'd like to simulate debris disaster using Fluid., but now I met a problem that I can't create fluid.
I'm sure every parameter of the fluid is right, because it is the same with the sample. I just copy the sampleFluidParticle project in the SDK
But when come to this line: mFluid = scene->createFluid(desc); the point of mFluid is always 0X000000
Does anybody met this situation before? Who knows the reason?

Offtopic: GPUDirect

Thu, 13 Oct 2011 04:57:36 -0700

Hallo to everyone,
I've recently got to know about GPUDirect here on developer.nvidia.com. There in the presention (http://developer.download.nvidia.com/compute/cuda/4_0/docs/GPUDirect_Technology_Overview.pdf) i've found that p2p bandwidth strongly depends on motherboard archticture. So does anybody know what switches mentioned there in the presentation? I supose it is something like asus p6t7 solutions witn two NF200 attached to X58 chipset. am i right?

starting this topic, i suugest not only to discuss GPUDirect questions but also to share your practical achievements in host - device transaction performance.

As for me i have GTX 285 on intel 946GZ/PL/G chipset with Q6600 CPU with 667MHz DDR2. In quda bandwidth test i got 2Gbyte/s in both directions with pinned memory allocation method. In nvidia-settings i see PCIe x16 gen 1. Maybe it is the reason why i get so weak bandwidth.
with standart allocation i got 1.3Gb/s and 2 Gb/s.
...i do use cudatoolkit_3.2.16 and 285.05.09 driver.

SDK 3.1 Linked Libraries Issue

Wed, 12 Oct 2011 10:28:24 -0700

Hello,

I am having an issue linking the PhysX libraries to a base program. I have uploaded the link errors to error.txt. The libraries are built in Ubuntu 11.04 64bit. The program is compiled using Codeblocks 10.05, following the included instructions provided in the documentation for linking the header documents and libraries. Any insight into the cause of these errors would be greatly appreciated.

multiple traingle meshes into multiple actors instead single actor

Wed, 12 Oct 2011 05:54:34 -0700

Hi
I have multiple traingle meshes in a vector.
std::vector triangleMeshes;

my code is this and it work fine.

CODE

NxActorDesc ActorDesc;
for (int i=0; i < triangleMeshes.size(); i++)
{
    NxTriangleMeshShapeDesc traingleShapeDesc;
    traingleShapeDesc.meshData  = triangleMeshes[i];

    ActorDesc.shapes.pushBack(&traingleShapeDesc);

    nxActor = phyScene->mNxScene->createActor(ActorDesc);
}

Instead of creating a actor for each mesh can I create a single actor using all the meshes I have?
If I do this actor is created using only the last mesh.

CODE

NxActorDesc ActorDesc;
for (int i=0; i < triangleMeshes.size(); i++)
{
    NxTriangleMeshShapeDesc traingleShapeDesc;
    traingleShapeDesc.meshData  = triangleMeshes[i];

    ActorDesc.shapes.pushBack(&traingleShapeDesc);
}
nxActor = phyScene->mNxScene->createActor(ActorDesc);

Why I cannot do above. Please help

Thanks

NxCapsuleController and upDirection

Thu, 06 Oct 2011 00:08:11 -0700

I am currently trying to implement something similar to the 'SampleCharacterController' example in the PhysX SDK with a capsule controller.

Background:

I have multiple PhysX scenes in my spherical world, spaced out via a patch system (sections of the world). These scenes do not use fixed axis systems (i.e. have arbitrary up vectors which are not axis aligned).
I have created terrain PhysX meshes for all of the terrain in my scenes, and I have a vehicle controller which works quite well using them.

The problem:

I noticed there was an 'upDirection' required to be specified in the descriptor of the capsule controller.
When i set the upDirection to a specific aligned axis (NX_X/ NX_Y/ NX_Z), I notice that my controller will hit the terrain tri-meshes and get resolved strangely (trying to move along the real Y axis I assume?).
The capsule itself when visually displayed in debug mode (and using one of the fixed up axis enum's) will be on an undesirable angle.
The gravity vector for each scene, is resolved to be the scenes origin on the surface, heading towards the centre of the planet.

Questions:

Why can't the controller resolve terrain penetration by just using the scenes current gravity vector?
Is there a way I can use this controller with this arbitrary no axis aligned system?

Thanks.

visual debugger not working with SDK 3.1?

Wed, 05 Oct 2011 06:11:38 -0700

when can we expect fixed version?

sdk 3.0.1 libraries broken ?

Tue, 04 Oct 2011 07:01:15 -0700

Hello,

I'm trying to test a simple source using physx but the sdk seems to be not working correctly
i searched a bit through the forum, it's seems a lot of people having trouble with linking physx libraries but i didnt find someone missing such a basic function as PxCreatePhysics()

i took this sample code : http://about.something.pl/wp-content/uploads/main.cc (from http://about.something.pl/art/getting-star...-on-linux.html)

CODE

g++ -DNDEBUG -I$PHYSX/SDKs/PhysXAPI -I$PHYSX/SDKs/PhysXAPI/extensions -I$PHYSX/PhysXFoundationSDK -I$PHYSX/pxtask/include -I$PHYSX/SDKs/PhysXAPI/deprecated -c samplebox.cc samplebox.cc: In function �void DrawAxes()�:
samplebox.cc:94:64: warning: deprecated conversion from string constant to �char*�
samplebox.cc:105:64: warning: deprecated conversion from string constant to �char*�
samplebox.cc:116:64: warning: deprecated conversion from string constant to �char*�

warnings seems to be acceptable, they are not related to PxCreatePhysics() proto

CODE

g++ -L$PHYSX/SDKs/lib/linux64 -lrt -lpthread -lPhysX3 -lPhysX3PROFILE -lPhysX3CHECKED -lPhysX3Extensions -lPhysX3ExtensionsPROFILE -lPhysX3ExtensionsCHECKED -lPhysX3MetaDataCHECKED -lRepX3CHECKED -lPhysX3CharacterKinematicCHECKED -lGL -lGLU -lglut samplebox.o
samplebox.o: In function `InitializePhysX()':
samplebox.cc:(.text+0x4d4): undefined reference to `PxCreatePhysics'
samplebox.cc:(.text+0x538): undefined reference to `PxInitExtensions'
samplebox.cc:(.text+0x5da): undefined reference to `PxDefaultCpuDispatcherCreate(unsigned int, unsigned int*)'
samplebox.cc:(.text+0x94f): undefined reference to `PxCreateDynamic'
samplebox.o: In function `__static_initialization_and_destruction_0(int, int)':
samplebox.cc:(.text+0x134f): undefined reference to `PxDefaultErrorCallback::PxDefaultErrorCallback()'
samplebox.cc:(.text+0x1354): undefined reference to `PxDefaultErrorCallback::~PxDefaultErrorCallback()'
samplebox.o:(.data+0x0): undefined reference to `PxDefaultSimulationFilterShader(unsigned int, PxFilterData, unsigned int, PxFilterData, physx::pubfnd3::PxFlags<PxPairFlag::Enum, unsigned short>&, void const*, unsigned int)'
collect2: ld returned 1 exit status

(i linked with all libraries i found using grep containing references to used functions)
it seems there's something really wrong with physx libraries
changing order seems to have no effect

any idea how to manage to link this simple exemple ?

thx in advance,

StereoIssues problem : can't find DXUT10d.lib

Tue, 04 Oct 2011 02:09:10 -0700

Hello, I am a newcomer for using Nvidia Graphics SDK.

I have just setup VS10 and Nvidia Graphics SDK environment but there are some problems.

I installed the latest Nvidia Graphics SDK 11, Microsoft DirectX SDK, Visual Studio 2010 and set the corresponding path for include and library files, with Windows 7 64bit as my OS.

However, when I compiled the sample code "StereoIssues" of Nvidia Graphics SDK 11, it came with a error message "can't find DXUT10d.lib".

I checked all the lib files of the installed SDK but only found DXUT10.lib, without a "b", in NV Direct3D SDK11\Lib.

I googled this problem but only found a similar problem, however, without a proper solution.

And I also wonder if there is any tutorial websites or websites for Nvidia Graphics SDK programming, including the environment setup.

Thanks for any reply

Physx 2.8x for Linux SDK

Mon, 03 Oct 2011 22:29:39 -0700

Dear Sir,

Would you please tell me where is physx 2.8x for linux 64bit SDK, we need it in a hurry~
Please let me know please~~

Best regards!!

Mac OS 64bit Libraries

Mon, 03 Oct 2011 11:54:15 -0700

Is there any plan to release Mac 64bit binaries or (universal libs) of the PhysX Libraries?

Was a little bit of a disappointment to see that they are only i386.

Thanks! Martin

Compile errors when adding PVD calls/classes

Sat, 01 Oct 2011 15:37:48 -0700

I get the following compile error whenever I try to use anything related to PVD (PvdConnection or PvdConnectionManager):

2>f:\physx-3.0.2_pc_sdk_core\sdks\physxapi\foundation\internal\include\pshashmap.h(90): error C2976: 'std::tr1::unordered_map' : too few template arguments
2> f:\program files (x86)\microsoft visual studio 10.0\vc\include\unordered_map(83) : see declaration of 'std::tr1::unordered_map'

I'm probably missing an include somewhere but it's proven hard to track down. The line in question is:

CODE

template <class Key,
              class Value,
              class HashFn = Hash<Key>,
              class Allocator = Allocator >
    class HashMap: public internal::HashMapBase<Key, Value, HashFn, Allocator>

I figured Hash or Allocator were missing, so I added PsHash.h and PsAllocator.h to my includes with no success. All of the samples compile fine so it's definitely something I'm missing, but whatever it is it isn't obvious even after I've gone through the sample (specifically SampleBase.h/.cpp) code a couple of times.

I'm using VS2010 on Windows 7 (with 32 bit compiler).

Any ideas?

reverse normals of a rigid body? (or how to make round fluid container)

Thu, 29 Sep 2011 22:23:34 -0700

Hey all,

I'm trying to figure the best way to simulate a round container to hold fluids without resorting to triangle mesh (due to speed issues). If a capsule could have it's normals reversed maybe it would work as a container?

Does anyone have any ideas for a round fluid container? thanks

Character Controller getting stuck

Thu, 29 Sep 2011 13:52:12 -0700

Hey all.

I've been having a problem with my character controller that I just cannot seem to fix. When I jump my character has a somewhat smallish chance of getting stuck if I am jumping into an object such as a tree or even a hill in the terrain. This happens around 5% of the time but never when I am running into an object. I just cannot work out for the life of me what could be causing it.

Here's my move request code (in C# - I'm using the PhysX candy wrapper):

CODE

public void PlayerMoveRequest(Entity mobEntity, Vector3 desiredDisplacement, ref Vector3 pos, bool followTerrain, float timeSinceLastFrame)
        {
            if (playerController != null && terrainLoaded && physicsEnabled)
            {
                ControllerFlags playerCollisionFlags;
                if (mobControllers[mobEntity].Jump.IsJumping)
                {
                    if (mobControllers[mobEntity].ForwardMomentum == Vector3.Zero && desiredDisplacement != Vector3.Zero)
                    {
                        mobControllers[mobEntity].ForwardMomentum = desiredDisplacement / timeSinceLastFrame;
                    }
                    desiredDisplacement = mobControllers[mobEntity].ForwardMomentum * timeSinceLastFrame;
                    desiredDisplacement.y += mobControllers[mobEntity].Jump.GetHeight(timeSinceLastFrame, gravity) / 1000;
                }
                else if (followTerrain)
                {
                    // do gravity check
                    if (mobControllers[mobEntity].IsFalling)
                        mobControllers[mobEntity].FallTime += timeSinceLastFrame;
                    else
                        mobControllers[mobEntity].FallTime = timeSinceLastFrame;
                    desiredDisplacement.y += gravity * mobControllers[mobEntity].FallTime / 1000;
                }
                playerController.Move(desiredDisplacement, collisionMask, 25, out playerCollisionFlags);
                if (playerCollisionFlags == ControllerFlags.Down)
                {
                    mobControllers[mobEntity].Jump.StopJump();
                    mobControllers[mobEntity].IsFalling = false;
                    mobControllers[mobEntity].FallTime = 0.0f;
                    mobControllers[mobEntity].ForwardMomentum = Vector3.Zero;
                }
                else
                {
                    mobControllers[mobEntity].IsFalling = true;
                }
                phscene.Simulate(timeSinceLastFrame);
                phscene.FlushStream();
                phscene.FetchResults(SimulationStatuses.AllFinished, true);
                // Get the players current position
                pos = playerController.Actor.GlobalPosition;
                pos.y -= 1130; // We offset the player a bit as the physics seem to give us an incorrect y value
            }
        }

and my character controller creation class:

CODE

public void LoadPlayerController(Entity playerEntity, SceneNode characterNode)
        {
            float modelHeight = characterNode.WorldAABB.Maximum.y - characterNode.WorldAABB.Minimum.y;
            float radius = 150;
            CapsuleControllerDesc playerCapsuleDesc = new CapsuleControllerDesc();
            playerCapsuleDesc.Height = modelHeight - radius * 2;
            playerCapsuleDesc.Radius = radius;
            playerCapsuleDesc.ClimbingMode = CapsuleClimbingModes.Constrained;
            playerCapsuleDesc.Position = new Vector3(characterNode.Position.x, characterNode.Position.y.Ceiling(), characterNode.Position.z);
            playerCapsuleDesc.SkinWidth = 50f; // 5cm
            playerCapsuleDesc.StepOffset = 500; // Player can step over objects upto 50cm high
            playerCapsuleDesc.UpDirection = HeightFieldAxes.Y;
            playerController = physics.ControllerManager.CreateController(phscene, playerCapsuleDesc);
            playerController.Actor.Name = characterNode.Name;
            ControllerFlags playerCollisionFlags;
            playerController.Move(new Vector3(0, 3000, 0), collisionMask, 25, out playerCollisionFlags);
            characterToLoad = null;
            MobMovementState mobMovementState = new MobMovementState(playerController);
            mobControllers.Add(playerEntity, mobMovementState);
        }

The world uses millimeters as the unit which explains the rather large numbers used. Another note is that it almost looks like the character controller could be sliding a tiny bit before getting stuck if that helps anyone. Also the visual debugger never shows any objects so no use asking me to use that

Thanks,
Andrew.

cgCreateEffectFromFile does not detected all errors

Wed, 28 Sep 2011 16:46:36 -0700

Hi,

I believer that there's a significant bug in cgCreateEffectFromFile(). In essence, it does not detect all errors present in the shader file. Let's take the following simple code as an example:

CODE

struct App_input {
    float4 position : POSITION;
    float4 normal   : NORMAL;
    float2 texcoords   : TEXCOORD0;
};

struct Vertex_to_fragment {
    float4 position    : POSITION;
    float2 texcoords   : TEXCOORD0;
};

///// program /////////////////////////////

Vertex_to_fragment vs_main(App_input input)
{
    Vertex_to_fragment output;

    output.position = mul(glstate.matrix.mvp, input.position);
    output.texcoords = input.texcoords;

    return out; // <<<< TYPE-O HERE
}

float4 fs_main(Vertex_to_fragment input) : COLOR
{
    float4 output;

    output = float4(1,0,0,1);

    return output;
}

///// TECHNIQUES /////////////////////////////

technique default_tech
{
    pass p0
    {
        VertexProgram = compile vp40 vs_main();
        FragmentProgram = compile fp40 fs_main();
    }
}

In the vertex function fs_main(), there is a type-o. Instead of "return out;" I should have written "return output;". Technically, if you call cgCreateEffectFromFile() with the cgfx shader above, you would expect an error. But unfortunately, no errors are detected!

CODE

CGeffect ef = cgCreateEffectFromFile( ctx, mFXFile.c_str(), NULL );
if ( !ef || cgGetError()!=CG_NO_ERROR )
{
// print errors...
}

If I comment out the fragment program in the technique section, then the errors are detected:

CODE

technique default_tech
{
    pass p0
    {
        VertexProgram = compile vp40 vs_main();
        //FragmentProgram = compile fp40 fs_main();
    }
}

Also, if I swap the programs in the technique section, then the error are detected:

CODE

technique default_tech
{
    pass p0
    {
        FragmentProgram = compile fp40 fs_main();
        VertexProgram = compile vp40 vs_main();
    }
}

Can someone shed some light on this issue? Is this a known bug? Or is it by design? Or a limitation?

Thanks,

Direction of a body - an actor

Tue, 27 Sep 2011 22:59:33 -0700

Hi together,

i try to set the direction of an actor towards the ground. It should be picked and moved in all directions, only the direction should be static. Does anyone know, how i may solve it?

Thanks

onShapeHit being invoked with unexpected worldNormals

Tue, 27 Sep 2011 17:42:36 -0700

I'm attempting to make our character slide when it is on surfaces that are too steep for it to stand on. In the onShapeHit callback of the character controller when the impact direction is straight down, I'm calculating the surface angle from worldNormal. However, the angles I'm seeing don't always make sense.

The calculation to determine an angle from the world normal is pretty simple. (UP is the vector 0, 1, 0)

CODE

const Ogre::Radian angle = Ogre::Math::ACos(hit.worldNormal.dot(UP))

When the character is transitioning from terrain to a small kinematic box, the normal, and hence the angle I calculate do not match my expectations. The box in this case is .1m high, resting on, and parallel to the terrain (I've verified this in the PhysX debugger). When the terrain is flat, the calculated angle can be as high as 20 degrees. If the terrain approaching the box is not flat (in my case a 7 degree slope), the resulting angle will as high as 51 degrees.

What's unclear to me is, why when the character controller sweeps the capsule straight down onto a box, the impact normal would not be (0, 1, 0). If I don't attempt to slide the character, the controller has no problem moving the capsule onto the box (even though the normal would indicate it exceeds the slope limit).

Does anyone have any insight as to why I'm seeing these normals, or a better way to determine if the character is on a surface that is too steep to stand on?

NxController & NxGroupsMask

Tue, 27 Sep 2011 07:46:24 -0700

Hi, I'm currently using NxGroupsMask to filter my Shape collisions and trying to setup NxController to works the same -with no real Luck-.

Setup is as follow :

For all Shapes :
- setGroup(0)
- setGroupsMask(NewMask) with NewMask = (1<
For my scene :
- setFilterOps(NX_FILTEROP_OR, NX_FILTEROP_OR, NX_FILTEROP_SWAP_AND);
- setFilterBool(true);
- NxGroupsMask zeroMask; ZeroMask.bits0=zeroMask.bits1=zeroMask.bits2=zeroMask.bits3=0;
- setFilterConstant0(zeroMask);
- setFilterConstant1(zeroMask);

Everything works fine, shapes from different channel won't collide if BitMask is not correctly set, and the same is valid for same CustomChannel.

And then I want to make my NxCharacter work.

I'm setting the Desc to interactionFlag = NXIF_INTERACTION_USE_FILTER
My NxController shape is set with a custom GroupsMask (same as my Rigid bodies shapes)

and when I move it, I use : move(vDeltaPos, 1<<0, minDist, collisionFlags, sharpness, &NewMask)

NewMask being the one I set on my shape.

It correctly stop colliding with my rigid bodies, while still colliding with the world (wich is what my setup needs), but unflagging my CharacterController Channel won't prevent the Controller from colliding into each other.

Did I miss something with my ::move() call, or does the NxController implements the NxGroupsMask in sweeping differently from rigid bodies ?

Don't hesitate to ask if it's unclear.
Thank for any answer.

PhysX SDK for MacOSX & iOS Platforms Please

Sun, 25 Sep 2011 14:47:28 -0700

I would like to request the creation of a PhysX SDK > PhysX Platforms > MacOS X and iOS subforum.

Is a code license required to use PhysX on iOS devices, or are the binaries available privately?

Screen Coordinates in Fragment Shader

Sun, 25 Sep 2011 05:56:39 -0700

Hello,

what is the easiest way to get screen coordinates (xy position of the pixel) in the fragment shader?

Thanks

Calculate the pixel color using parameters from 4 vertices

Sun, 25 Sep 2011 04:21:37 -0700

I'm using OpenGL and CG. I need calculate the pixel color using the color or other parameters from all vertices of GL_QUADS which contains that pixel. I know that i must use fragment shader but i dont know how. To my calculations i must haw distance from pixel to all vericles and color of that vertices. I know the basic of using CG fragment and vertex shaders in OpenGL but i don't know how to get the proper values. I nead that for my study, please help me.

Fluid does not interact with Terrain

Sat, 24 Sep 2011 06:12:45 -0700

Now I need a Fluid interact with the Terrain, which Was Created from thiangle mesh. but the fluid don't interact with the terrain , it just flow through the terrain. how can I solve the problem?
Can fluid interact with trianglemesh? And any parameters need to be set?

not all actors are visible in new versions of the debugger

Fri, 23 Sep 2011 12:18:06 -0700

With the new DirectX renderer in the PhysX Visual Debugger (I've tried 2.0000.9942 and 2.0001.0850.4481), I'm not seeing all actors (new.PNG), though they are still selectable and their bounding boxes are visible (new_sel.PNG).

Does anyone else have this problem?

I didn't have this problem with an older version (2.0000.8251.2581) of the debugger.

Here's my DxDiag log:

CODE

------------------
System Information
------------------
Time of this report: 9/23/2011, 14:49:39
       Machine name:
   Operating System: Windows 7 Professional 64-bit (6.1, Build 7601) Service Pack 1 (7601.win7sp1_gdr.110622-1506)
           Language: English (Regional Setting: English)
System Manufacturer: INTEL_
       System Model: WBIBX10J
               BIOS: Default System BIOS
          Processor: Intel(R) Core(TM) i5 CPU         750  @ 2.67GHz (4 CPUs), ~2.7GHz
             Memory: 4096MB RAM
Available OS Memory: 4086MB RAM
          Page File: 2982MB used, 5188MB available
        Windows Dir: C:\Windows
    DirectX Version: DirectX 11
DX Setup Parameters: Not found
   User DPI Setting: Using System DPI
System DPI Setting: 96 DPI (100 percent)
    DWM DPI Scaling: Disabled
     DxDiag Version: 6.01.7601.17514 32bit Unicode

------------
DxDiag Notes
------------
      Display Tab 1: No problems found.
        Sound Tab 1: No problems found.
        Sound Tab 2: No problems found.
          Input Tab: No problems found.

--------------------
DirectX Debug Levels
--------------------
Direct3D:    0/4 (retail)
DirectDraw:  0/4 (retail)
DirectInput: 0/5 (retail)
DirectMusic: 0/5 (retail)
DirectPlay:  0/9 (retail)
DirectSound: 0/5 (retail)
DirectShow:  0/6 (retail)

---------------
Display Devices
---------------
          Card name: ATI Radeon HD 4300/4500 Series
       Manufacturer: Advanced Micro Devices, Inc.
          Chip type: ATI display adapter (0x954F)
           DAC type: Internal DAC(400MHz)
         Device Key: Enum\PCI\VEN_1002&DEV_954F&SUBSYS_16181462&REV_00
     Display Memory: 2292 MB
   Dedicated Memory: 504 MB
      Shared Memory: 1787 MB
       Current Mode: 1920 x 1080 (32 bit) (60Hz)
       Monitor Name: Generic PnP Monitor
      Monitor Model: ASUS VH242H
         Monitor Id: ACI24F3
        Native Mode: 1920 x 1080(p) (60.000Hz)
        Output Type: DVI
        Driver Name: aticfx64.dll,aticfx64.dll,aticfx32,aticfx32,atiumd64.dll,atidxx64.dll,atiumdag,a
tidxx32,atiumdva,atiumd6a.cap,atitmm64.dll
Driver File Version: 8.17.0010.1091 (English)
     Driver Version: 8.881.0.0
        DDI Version: 10.1
       Driver Model: WDDM 1.1
  Driver Attributes: Final Retail
   Driver Date/Size: 7/28/2011 17:39:14, 852992 bytes
        WHQL Logo'd: Yes
    WHQL Date Stamp:
  Device Identifier: {D7B71EE2-D60F-11CF-9173-1236BEC2C535}
          Vendor ID: 0x1002
          Device ID: 0x954F
          SubSys ID: 0x16181462
        Revision ID: 0x0000
Driver Strong Name: oem41.inf:ATI.Mfg.NTamd64.6.1:ati2mtag_R7X:8.881.0.0:pci\ven_1002&dev_954f
     Rank Of Driver: 00E62001
        Video Accel: ModeMPEG2_A ModeMPEG2_C
   Deinterlace Caps: {6E8329FF-B642-418B-BCF0-BCB6591E255F}: Format(In/Out)=(YUY2,YUY2) Frames(Prev/Fwd/Back)=(0,0,1) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY DeinterlaceTech_PixelAdaptive
                     {335AA36E-7884-43A4-9C91-7F87FAF3E37E}: Format(In/Out)=(YUY2,YUY2) Frames(Prev/Fwd/Back)=(0,0,0) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY DeinterlaceTech_BOBVerticalStretch
                     {5A54A0C9-C7EC-4BD9-8EDE-F3C75DC4393B}: Format(In/Out)=(YUY2,YUY2) Frames(Prev/Fwd/Back)=(0,0,0) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY
                     {6E8329FF-B642-418B-BCF0-BCB6591E255F}: Format(In/Out)=(UYVY,UYVY) Frames(Prev/Fwd/Back)=(0,0,1) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY DeinterlaceTech_PixelAdaptive
                     {335AA36E-7884-43A4-9C91-7F87FAF3E37E}: Format(In/Out)=(UYVY,UYVY) Frames(Prev/Fwd/Back)=(0,0,0) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY DeinterlaceTech_BOBVerticalStretch
                     {5A54A0C9-C7EC-4BD9-8EDE-F3C75DC4393B}: Format(In/Out)=(UYVY,UYVY) Frames(Prev/Fwd/Back)=(0,0,0) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY
                     {5A54A0C9-C7EC-4BD9-8EDE-F3C75DC4393B}: Format(In/Out)=(YV12,0x32315659) Frames(Prev/Fwd/Back)=(0,0,0) Caps=
                     {3C5323C1-6FB7-44F5-9081-056BF2EE449D}: Format(In/Out)=(NV12,0x3231564e) Frames(Prev/Fwd/Back)=(0,0,2) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY DeinterlaceTech_PixelAdaptive
                     {552C0DAD-CCBC-420B-83C8-74943CF9F1A6}: Format(In/Out)=(NV12,0x3231564e) Frames(Prev/Fwd/Back)=(0,0,2) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY DeinterlaceTech_PixelAdaptive
                     {6E8329FF-B642-418B-BCF0-BCB6591E255F}: Format(In/Out)=(NV12,0x3231564e) Frames(Prev/Fwd/Back)=(0,0,1) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY DeinterlaceTech_PixelAdaptive
                     {335AA36E-7884-43A4-9C91-7F87FAF3E37E}: Format(In/Out)=(NV12,0x3231564e) Frames(Prev/Fwd/Back)=(0,0,0) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY DeinterlaceTech_BOBVerticalStretch
                     {5A54A0C9-C7EC-4BD9-8EDE-F3C75DC4393B}: Format(In/Out)=(NV12,0x3231564e) Frames(Prev/Fwd/Back)=(0,0,0) Caps=VideoProcess_YUV2RGB VideoProcess_StretchX VideoProcess_StretchY
                     {5A54A0C9-C7EC-4BD9-8EDE-F3C75DC4393B}: Format(In/Out)=(IMC1,UNKNOWN) Frames(Prev/Fwd/Back)=(0,0,0) Caps=
                     {5A54A0C9-C7EC-4BD9-8EDE-F3C75DC4393B}: Format(In/Out)=(IMC2,UNKNOWN) Frames(Prev/Fwd/Back)=(0,0,0) Caps=
                     {5A54A0C9-C7EC-4BD9-8EDE-F3C75DC4393B}: Format(In/Out)=(IMC3,UNKNOWN) Frames(Prev/Fwd/Back)=(0,0,0) Caps=
                     {5A54A0C9-C7EC-4BD9-8EDE-F3C75DC4393B}: Format(In/Out)=(IMC4,UNKNOWN) Frames(Prev/Fwd/Back)=(0,0,0) Caps=
                     {5A54A0C9-C7EC-4BD9-8EDE-F3C75DC4393B}: Format(In/Out)=(S340,UNKNOWN) Frames(Prev/Fwd/Back)=(0,0,0) Caps=
                     {5A54A0C9-C7EC-4BD9-8EDE-F3C75DC4393B}: Format(In/Out)=(S342,UNKNOWN) Frames(Prev/Fwd/Back)=(0,0,0) Caps=
       D3D9 Overlay: Not Supported
            DXVA-HD: Not Supported
       DDraw Status: Enabled
         D3D Status: Enabled
         AGP Status: Enabled

Import .obj-files as Rigid Body

Fri, 23 Sep 2011 04:08:16 -0700

Hi there,

i try to create a softbody and a rigidbody which have to interact in a specific way. Creating the softBody isnt�t a problem but if i want do create a rigidBody in the same way, i don�t find a solution. The models are made with blender and the .tet files are created by physXViewer. Does anyone have an idea how i could get it working?

Thanks

linking errors of GLEW

Thu, 22 Sep 2011 14:45:03 -0700

Hello,

I have a problem when I started to compile the examples of CG tutorials of chapter 5 of lighting: http://developer.nvidia.com/node/88
these are the errors:

Error 1 error LNK2001: unresolved external symbol ___GLEW_VERSION_1_1 09_vertex_lighting.obj
Error 2 error LNK2019: unresolved external symbol _glewInit referenced in function _main 09_vertex_lighting.obj
Error 3 error LNK2001: unresolved external symbol ___wglewSwapIntervalEXT 09_vertex_lighting.obj
Error 4 fatal error LNK1120: 3 unresolved externals Win32\Debug\09_vertex_lighting.exe

any idea why?
thanks

Basic question in CG tutorial regarding semantics

Thu, 22 Sep 2011 10:28:39 -0700

Hello everyone,
First, I have a simple question, when I start to write a shading program, is this totally inhibits the normal pipeline? (I am asking this questions bec. I found some examples in the cg tutorial that implement a vertex program that has only one function, to receive position and color and then out them to be used by fragment program, while some other examples has only fragment program without a vertex program and vise versa, or even why should I write a program that computes modelviewproj matrix as in example:Example 4-1. The C4E1v_transform Vertex Program in http://developer.nvidia.com/node/89 while openGL can do this automatically).

secondly, I have some basic questions regarding semantics, as in "unified" which states that a variable is unifies when its initial value comes from an environment that is external to the specified Cg program, while in a lot of the cg tutorial examples' position and color are not unifiedwhile they are initially coming from the openGL main file, so what is the difference?

another issue, here in the tutorial:http://developer.nvidia.com/node/88 , in Example 5-2. The C5E2v_fragmentLighting Vertex Program, the semantics for Out Float3 objectPos was TEXCOORD0 and Out Float3 oNormal was TEXCOORD1. there is no good explanation why they are not of POSITION type, so can anybody help?

Thanks a lot for your time.

Custom Uniforms

Wed, 21 Sep 2011 06:55:03 -0700

I am new to the fxcomposer platform. What i want to achieve is to pass the previous frame Mouse Position along with the Current Mouse Position inside the Shader.

I cannot find the right spot to put my custom code. Maybe i cannot do that at all.

If anyone can direct me to the right place i would appreciate it.

Is PerfHUD still being supported/maintained?

Tue, 20 Sep 2011 22:58:12 -0700

I recently purchased an Nvidia card to learn their development tools.

However, the latest PerfHUD does not work... it just says: "Failed to Initialize NVIDIA PerfSDK"

OS: Win7 64-bit
GPU: Geforce 550 Ti
Nvidia.com driver: 280.26
PerfHUD: 6.70.1201.2315

Nvidia writes great development tools but I'm a bit disappointed to not see this one working.

Can anyone offer a bit of help?

Thanks,
- Vashren

Vertex Buffer Objects and glDrawElements problem on Mac OSX

Tue, 20 Sep 2011 14:56:26 -0700

I've recently started going through the Cg Tutorial book and want to switch from using immediate mode to the non-deprecated VBO method before I get too far into it.

I have a simple vertex shader which doesn't do much at all yet. It just takes a position input and spits out a position and colour. I'm just trying to render a simple triangle and if I pass the verts to the shader using immediate mode or vertex arrays (both deprecated) it works fine. However, I can't seem to get any output if I use VBOs. I'm sure it'll be something simple that I'm missing but I can't see what I'm doing wrong.

I know that using glDrawElements is a bit overkill for just a single triangle but I'm wanting to ensure that everything's working before I start loading models and whatnot.

CODE

        GLfloat triVerts[9] = 
        {
            0.0f, 1.0f, 0.0f,
            -1.0f, -1.0f, 0.0f,
            1.0f, -1.0f, 0.0f
        };
        
        GLuint triIndices[3] = 
        {
            0, 1, 2
        };
        
        glEnableVertexAttribArray(0);
        
        GLuint vertBuffer;
        GLuint elementBuffer;

        glGenBuffers(1, &vertBuffer);
        glBindBuffer(GL_ARRAY_BUFFER, vertBuffer);
        glBufferData(GL_ARRAY_BUFFER, sizeof(GLfloat) * 9, triVerts, GL_STATIC_DRAW);

        glGenBuffers(1, &elementBuffer);
        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, elementBuffer);
        glBufferData(GL_ELEMENT_ARRAY_BUFFER, sizeof(GLuint) * 3, triIndices, GL_STATIC_DRAW);
    
       	glBindBuffer(GL_ARRAY_BUFFER, vertBuffer);
        glVertexAttribPointer(0, 3, GL_FLOAT, GL_FALSE, 0, 0);

        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, elementBuffer);
        glDrawElements(GL_TRIANGLES, 3, GL_UNSIGNED_INT, 0);
    
    	glDeleteBuffers(1, &vertBuffer);
        glDeleteBuffers(1, &elementBuffer);
        
        glDisableVertexAttribArray(0);

As mentioned above, if I use vertex arrays like below the triangle renders fine:

CODE

        GLfloat triVerts[9] = 
        {
            0.0f, 1.0f, 0.0f,
            -1.0f, -1.0f, 0.0f,
            1.0f, -1.0f, 0.0f
        };
        
        GLuint triIndices[3] = 
        {
            0, 1, 2
        };
        
        glEnableClientState(GL_VERTEX_ARRAY);
        glVertexPointer(3, GL_FLOAT, 0, triVerts);
        glDrawElements(GL_TRIANGLES, 3, GL_UNSIGNED_INT, triIndices);
        glDisableClientState(GL_VERTEX_ARRAY);

Any ideas?

NX_PHYSICS_SDK_VERSION returns NULL

Tue, 20 Sep 2011 06:24:15 -0700

Hi there,

i have a problem with the basic examples. I try to create a scene but
gPhysicsSDK = NxCreatePhysicsSDK(NX_PHYSICS_SDK_VERSION);
seems to return NULL. I dont know why.

CODE

void InitializePhysX(){

gPhysicsSDK = NxCreatePhysicsSDK(2.84);
if(gPhysicsSDK == NULL){
    cerr<<"Error creating PhysX device."<<endl;
    cerr<<"Exiting..."<<endl;
    exit(1);
}

Does anyone know something about this problem?
Thanks

I did it. How may i close this Thread?

Where can I get the references?

Tue, 20 Sep 2011 06:00:44 -0700

Hi guys,

I have been using PhysX for a while for my research. I have two questions:
1 Is there any references for the implementation of features of PhysX? For example, the cloth simulation algorithm.
2 If the algorithm is confidential, is there any instructions I can follow if I want to publish something and use PhysX as a reference?

Cheers,

He

query 'dedicated video memory' on windows 7

Sun, 18 Sep 2011 05:54:48 -0700

I have dell latitude E6410 with NVidia NVS 3100M card - with Dedicated video memory of 512 MB.
Running the attached C++ code I get that the card has only 485632 KB of dedicatrd video memory.

Can you point me to the cause of the difference ?

Thanks.

The C++ code:

IDXGIAdapter * pAdapter;
IDXGIFactory* pFactory = NULL;

// Create a DXGIFactory object.
if(FAILED(CreateDXGIFactory(__uuidof(IDXGIFactory) ,(void**)&pFactory))) {
return 0;
}

int memMax = 0;
for (UINT i = 0; pFactory->EnumAdapters(i, &pAdapter) != DXGI_ERROR_NOT_FOUND; ++i) {
DXGI_ADAPTER_DESC adapterDesc;
pAdapter->GetDesc(&adapterDesc);
int mem = adapterDesc.DedicatedVideoMemory/1024; // convert bytes to KBytes
}

The card specification:

NVIDIA System Information report created on: 09/18/2011 15:23:08
System name: APOPC11

[Display]
Processor: Intel® Core™ i5 CPU M 560 @ 2.67GHz (2661 MHz)
Operating System: Windows 7 Professional, 64-bit (Service Pack 1)
DirectX version: 11.0
GPU processor: NVS 3100M
Driver version: 189.74
CUDA Cores: 16
Core clock: 606 MHz
Shader clock: 1468 MHz
Memory clock: 790 MHz (1580 MHz data rate)
Memory interface: 64-bit
Total available graphics memory: 4094 MB
Dedicated video memory: 512 MB
System video memory: 0 MB
Shared system memory: 3582 MB
Video BIOS version: 70.18.53.00.04
IRQ: 16
Bus: PCI Express x16

[Components]

nvCpl.cpl 2.5.630.20 NVIDIA Control Panel Applet
nvCplUI.exe 2.5.630.20 NVIDIA Control Panel
nvViTvS.dll 8.16.11.8974 NVIDIA Video and TV Server
NVMCTRAY.DLL 8.16.11.8974 NVIDIA Media Center Library
nvDispS.dll 8.16.11.8974 NVIDIA Display Server
NVCPL.DLL 8.16.11.8974 NVIDIA Compatible Windows7 Display driver, Version 189.74
NVCUDA.DLL 8.16.11.8974 NVIDIA CUDA 2.2 driver
nvGameS.dll 8.16.11.8974 NVIDIA 3D Settings Server

Processor: Intel® Core™ i5 CPU M 560 @ 2.67GHz (2661 MHz)
Operating System: Windows 7 Professional, 64-bit (Service Pack 1)
DirectX version: 11.0
GPU processor: NVS 3100M
Driver version: 189.74
CUDA Cores: 16
Core clock: 606 MHz
Shader clock: 1468 MHz
Memory clock: 790 MHz (1580 MHz data rate)
Memory interface: 64-bit
Total available graphics memory: 4094 MB
Dedicated video memory: 512 MB
System video memory: 0 MB
Shared system memory: 3582 MB
Video BIOS version: 70.18.53.00.04
IRQ: 16
Bus: PCI Express x16

Actors floating above ground

Thu, 15 Sep 2011 14:02:03 -0700

I have a problem: My dynamic actors, cubes, are all floating a certain offset above the ground, and I simply don't have any idea anymore what could be causing it.
The ground is a regular heightfield. My player, a capsule charactercontroller, has never had this problem.
I've tried playing around with skin widths, with zero difference.
I'm using metric scale, the cube is 0.25 per side, and I currently have a global skin width of 0.0125. I'm not sure if that's a good value, but changing it hasn't fixed the issue anyway.
I don't know what else could be causing this.
Any input would be very much appreciated..!

Also, I can't use the visual remote debugger: connecting to it freezes everything until I close it. No idea why.

I'm using the following visualizations in this image:

CODE

pnxSDK->setParameter( NX_VISUALIZATION_SCALE, 0.5f );
pnxSDK->setParameter( NX_VISUALIZE_WORLD_AXES, 1.0f );
pnxSDK->setParameter( NX_VISUALIZE_ACTOR_AXES, 1.0f );
pnxSDK->setParameter( NX_VISUALIZE_BODY_MASS_AXES, 1.0f );
pnxSDK->setParameter( NX_VISUALIZE_CONTACT_POINT, 1.0f );
pnxSDK->setParameter( NX_VISUALIZE_CONTACT_FORCE, 1.0f );
pnxSDK->setParameter( NX_VISUALIZE_BODY_LIN_VELOCITY, 1.0f );
pnxSDK->setParameter( NX_VISUALIZE_BODY_ANG_VELOCITY, 1.0f );

Stereoscopin rendering with Open GL and GeForce

Thu, 15 Sep 2011 08:27:10 -0700

Hi there.

I'm new here and i want to say sorry in advance if this subject was discussed already, but i would like to ask you for a clear answer.

Is there ANY possibility to use 3d Vision stereoscopic technology with OpenGL not DX library and GeForce not Quadro chip?

I'm a member of developers' team, and we're working on application for home users, using stereoscope rendering. We need to render two frames, separately for left and right eye, and then display it in stereoscopic mode. Is there any possibility to do it like in case of DX? Render both images, create a special texture for both, add special header etc?

Greetings and thanks in advance,
Zegar

PhysX Runtimes

Wed, 14 Sep 2011 06:25:40 -0700

Unity3D employs PhysX for physical simulation and yet does not require the user to install PhysX drivers/runtime... either it does this automatically or it has some bundled special version of PhysX... my guess is the latter.

Does anyone know if developers are able to do this, or did Unity3D get a special deal? For possible use in a 3D browser plugin, avoiding additional install dependencies is a big usability bonus.

Thankyou.

How to Clothing?

Tue, 13 Sep 2011 23:32:32 -0700

Hi,

I have to figure bones in the fabric department, when cloth collision with rigid skeleton fabric collapse. I read physXSKD clothing, but little information, do not know how to do:unsure:

PhysX is C or c++ ?

Tue, 13 Sep 2011 15:53:19 -0700

Hi all

PhysX is written in C or C++ ?

kiko

NV_STEREO_IMAGE_SIGNATURE and DX10/11

Sun, 11 Sep 2011 04:02:55 -0700

Hi there,
as some of you may be aware there is a low-level hack through which you can control the stereoization process and render the left and right eye images manually, then you write a special value in the last row. Upon rendering this new scene, the nVidia driver picks the value up and activates the shutter glasses. I've been able to test it under DX9 so I know that it definitely can work but due to some methods not being available anymore in DX10/11 I've been unable to make it work.

The algorithm goes like this:
Render left eye image
Render right eye image
Create a texture able to contain them both PLUS an extra row (so the texture size would be 2 * width, height + 1)
Write this NV_STEREO_IMAGE_SIGNATURE value
Render this texture on the screen

My test code skips the first two steps, as I already have a stereo texture. It was a former .JPS file, specifically one of those included in the sample pictures coming with the nvidia 3D kit. Step number 5 uses a full screen quad and a shader to render the stereoized texture onto it through an ortho-projection matrix. The sample code I've seen for DX9 doesn't need this and simply calls the StretchRect(...) method to copy the texture back onto the backbuffer. So maybe it is for this reason that is not working? Is there a similar method to accomplish this in DX10? I thought that rendering onto the backbuffer would theoretically be the same than copying (or StretchRecting) a texture onto it, but maybe it is not?

Here follows my code (slimdx):
Stereoization procedure

CODE

static Texture2D Make3D(Texture2D stereoTexture)
{
    // stereoTexture contains a stereo image with the left eye image on the left half
    // and the right eye image on the right half
    // this staging texture will have an extra row to contain the stereo signature
    Texture2DDescription stagingDesc = new Texture2DDescription()
    {
        ArraySize = 1,
        Width = 3840,
        Height = 1081,
        BindFlags = BindFlags.None,
        CpuAccessFlags = CpuAccessFlags.Write,
        Format = SlimDX.DXGI.Format.R8G8B8A8_UNorm,
        OptionFlags = ResourceOptionFlags.None,
        Usage = ResourceUsage.Staging,
        MipLevels = 1,
        SampleDescription = new SampleDescription(1, 0)
    };
    Texture2D staging = new Texture2D(device, stagingDesc);

    // Identify the source texture region to copy (all of it)
    ResourceRegion stereoSrcBox = new ResourceRegion { Front = 0, Back = 1, Top = 0, Bottom = 1080, Left = 0, Right = 3840 };
    // Copy it to the staging texture
    device.CopySubresourceRegion(stereoTexture, 0, stereoSrcBox, staging, 0, 0, 0, 0);

    // Open the staging texture for reading
    DataRectangle box = staging.Map(0, MapMode.Write, SlimDX.Direct3D10.MapFlags.None);
    // Go to the last row
    box.Data.Seek(stereoTexture.Description.Width * stereoTexture.Description.Height * 4, System.IO.SeekOrigin.Begin);
    // Write the NVSTEREO header
    box.Data.Write(data, 0, data.Length);
    staging.Unmap(0);

    // Create the final stereoized texture
    Texture2DDescription finalDesc = new Texture2DDescription()
    {
        ArraySize = 1,
        Width = 3840,
        Height = 1081,
        BindFlags = BindFlags.ShaderResource,
        CpuAccessFlags = CpuAccessFlags.Write,
        Format = SlimDX.DXGI.Format.R8G8B8A8_UNorm,
        OptionFlags = ResourceOptionFlags.None,
        Usage = ResourceUsage.Dynamic,
        MipLevels = 1,
        SampleDescription = new SampleDescription(1, 0)
    };

    // Copy the staging texture on a new texture to be used as a shader resource
    Texture2D final = new Texture2D(device, finalDesc);
    device.CopyResource(staging, final);
    staging.Dispose();
    return final;
}

NV_STEREO_IMAGE_SIGNATURE data

CODE

        // The NVSTEREO header.
        static byte[] data = new byte[] {0x4e, 0x56, 0x33, 0x44,   //NVSTEREO_IMAGE_SIGNATURE         = 0x4433564e;
            0x00, 0x0F, 0x00, 0x00,   //Screen width * 2 = 1920*2 = 3840 = 0x00000F00;
            0x38, 0x04, 0x00, 0x00,   //Screen height = 1080             = 0x00000438;
            0x20, 0x00, 0x00, 0x00,   //dwBPP = 32                       = 0x00000020;
            0x02, 0x00, 0x00, 0x00};  //dwFlags = SIH_SCALE_TO_FIT       = 0x00000002

Main

CODE

private static Device device;

[STAThread]
static void Main()
{
    // Device creation
    var form = new RenderForm("Stereo test") {ClientSize = new Size(1920, 1080)};
    var desc = new SwapChainDescription()
                   {
                       BufferCount = 1,
                       ModeDescription = new ModeDescription(1920, 1080, new Rational(120, 1), Format.R8G8B8A8_UNorm),
                       IsWindowed = true,
                       OutputHandle = form.Handle,
                       SampleDescription = new SampleDescription(1, 0),
                       SwapEffect = SwapEffect.Discard,
                       Usage = Usage.RenderTargetOutput
                   };

    SwapChain swapChain;
    Device.CreateWithSwapChain(null, DriverType.Hardware, DeviceCreationFlags.Debug, desc, out device, out swapChain);
    //Stops Alt+enter from causing fullscreen skrewiness.
    Factory factory = swapChain.GetParent<Factory>();
    factory.SetWindowAssociation(form.Handle, WindowAssociationFlags.IgnoreAll);

    Texture2D backBuffer = Resource.FromSwapChain<Texture2D>(swapChain, 0);
    RenderTargetView renderView = new RenderTargetView(device, backBuffer);

    ImageLoadInformation info = new ImageLoadInformation()
                                    {
                                        BindFlags = BindFlags.None,
                                        CpuAccessFlags = CpuAccessFlags.Read,
                                        FilterFlags = FilterFlags.None,
                                        Format = SlimDX.DXGI.Format.R8G8B8A8_UNorm,
                                        MipFilterFlags = FilterFlags.None,
                                        OptionFlags = ResourceOptionFlags.None,
                                        Usage = ResourceUsage.Staging,
                                        MipLevels = 1
                                    };

    // Make texture 3D
    Texture2D sourceTexture = Texture2D.FromFile(device, "medusa.jpg", info);
    Texture2D stereoizedTexture = Make3D(sourceTexture);
    ShaderResourceView srv = new ShaderResourceView(device, stereoizedTexture);

    // Create a quad that fills the whole screen
    ushort[] idx;
    TexturedVertex[] quad = CreateTexturedQuad(Vector3.Zero, 1920, 1080, out idx);

    // fill vertex and index buffers
    DataStream stream = new DataStream(4*24, true, true);
    stream.WriteRange(quad);
    stream.Position = 0;

    Buffer vertices = new SlimDX.Direct3D10.Buffer(device, stream, new BufferDescription()
                                                                       {
                                                                           BindFlags = BindFlags.VertexBuffer,
                                                                           CpuAccessFlags = CpuAccessFlags.None,
                                                                           OptionFlags = ResourceOptionFlags.None,
                                                                           SizeInBytes = 4*24,
                                                                           Usage = ResourceUsage.Default
                                                                       });
    stream.Close();

    stream = new DataStream(6*sizeof (ushort), true, true);
    stream.WriteRange(idx);
    stream.Position = 0;
    Buffer indices = new SlimDX.Direct3D10.Buffer(device, stream, new BufferDescription()
                                                                      {
                                                                          BindFlags = BindFlags.IndexBuffer,
                                                                          CpuAccessFlags = CpuAccessFlags.None,
                                                                          OptionFlags = ResourceOptionFlags.None,
                                                                          SizeInBytes = 6*sizeof (ushort),
                                                                          Usage = ResourceUsage.Default
                                                                      });

    // Create world view (ortho) projection matrices
    QuaternionCam qCam = new QuaternionCam();

    // Load effect from file. It is a basic effect that renders a full screen quad through
    // an ortho projectio=n matrix
    Effect effect = Effect.FromFile(device, "Texture.fx", "fx_4_0", ShaderFlags.Debug, EffectFlags.None);
    EffectTechnique technique = effect.GetTechniqueByIndex(0);
    EffectPass pass = technique.GetPassByIndex(0);
    InputLayout layout = new InputLayout(device, pass.Description.Signature, new[]
                                                                                 {
                                                                                     new InputElement("POSITION", 0,
                                                                                                      Format.
                                                                                                          R32G32B32A32_Float,
                                                                                                      0, 0),
                                                                                     new InputElement("TEXCOORD", 0,
                                                                                                      Format.
                                                                                                          R32G32_Float,
                                                                                                      16, 0)
                                                                                 });
    effect.GetVariableByName("mWorld").AsMatrix().SetMatrix(
        Matrix.Translation(Layout.OrthographicTransform(Vector2.Zero, 90, new Size(1920, 1080))));
    effect.GetVariableByName("mView").AsMatrix().SetMatrix(qCam.View);
    effect.GetVariableByName("mProjection").AsMatrix().SetMatrix(qCam.OrthoProjection);
    effect.GetVariableByName("tDiffuse").AsResource().SetResource(srv);

    // Set RT and Viewports
    device.OutputMerger.SetTargets(renderView);
    device.Rasterizer.SetViewports(new Viewport(0, 0, form.ClientSize.Width, form.ClientSize.Height, 0.0f, 1.0f));

    // Create solid rasterizer state
    RasterizerStateDescription rDesc = new RasterizerStateDescription()
                                           {
                                               CullMode = CullMode.None,
                                               IsDepthClipEnabled = true,
                                               FillMode = FillMode.Solid,
                                               IsAntialiasedLineEnabled = true,
                                               IsFrontCounterclockwise = true,
                                               IsMultisampleEnabled = true
                                           };
    RasterizerState rState = RasterizerState.FromDescription(device, rDesc);
    device.Rasterizer.State = rState;

    // Main Loop
    MessagePump.Run(form, () =>
        {
            device.ClearRenderTargetView(renderView, Color.Cyan);

            device.InputAssembler.SetInputLayout(layout);
            device.InputAssembler.SetPrimitiveTopology(PrimitiveTopology.TriangleList);
            device.InputAssembler.SetVertexBuffers(0, new VertexBufferBinding(vertices, 24, 0));
            device.InputAssembler.SetIndexBuffer(indices, Format.R16_UInt, 0);

            for (int i = 0; i < technique.Description.PassCount; ++i)
            {
                // Render the full screen quad
                pass.Apply();
                device.DrawIndexed(6, 0, 0);
            }

            swapChain.Present(0, PresentFlags.None);
        });

    // Dispose resources
    vertices.Dispose();
    layout.Dispose();
    effect.Dispose();
    renderView.Dispose();
    backBuffer.Dispose();
    device.Dispose();
    swapChain.Dispose();

    rState.Dispose();
    stereoizedTexture.Dispose();
    sourceTexture.Dispose();
    indices.Dispose();
    srv.Dispose();
}

Thanks in advance!

Problem with shader or i dont know what :P

Fri, 09 Sep 2011 14:32:17 -0700

I have this strange problem with my simple shader program:

CODE

uniform int timer;  // (1)

void VSMain(float3 objectSpacePos : POSITION, float2 inTexC : TEXCOORD0,// float3 normal : NORMAL,
    out float4 position : POSITION, out float2 texC : TEXCOORD0)
{
    //float displacement = 0.5f * sin(objectSpacePos.y * 1.0f * timer) + 1.0f;
    //normal = normalize(normal);
    //float4 displacementDirection = float4(normal.x, normal.y, normal.z, 0);
    //float4 newPosition = position + displacement * displacementDirection;
    position = float4(objectSpacePos, 1.0f);
    texC = inTexC;
}

In the application program i create and setup everything needed by the program (im using dx11 and write in c++). The program is very simple and only displays one textured triangle. But when i compile and run it, i get an error:

I used cgIsProgramCompiled(), cgGetLastListing() and cgD3D11GetProgramErrors() to check, if anything went wrong but according to these functions it didint. The debugger shows error at line (2):

CODE

        oContext = cgCreateContext();
        oProgramVS = cgCreateProgramFromFile(oContext, CG_SOURCE, "shaders/vs.cg", CG_PROFILE_VS_5_0, "VSMain", NULL);
        bool compiled = cgIsProgramCompiled(oProgramVS);
        const char * pCompilerError = cgGetLastListing( oContext );
        oProgramFS = cgCreateProgramFromFile(oContext, CG_SOURCE, "shaders/fs.cg", CG_PROFILE_PS_5_0, "FSMain", NULL);
        compiled = cgIsProgramCompiled(oProgramVS);
        pCompilerError = cgGetLastListing( oContext );
        oResult = cgD3D11SetDevice(oContext, poDevice);
        pCompilerError = cgGetLastListing( oContext);
        if (FAILED(oResult))
        {
            return oResult;
        }
        ID3DBlob * err = cgD3D11GetProgramErrors(oProgramVS);
        oResult = cgD3D11LoadProgram(oProgramVS, D3DCOMPILE_OPTIMIZATION_LEVEL1);   // (2)

(Sry, code is a lil messy now due to those error checking functions. ).
What is weird is that when i comment out the line (1) in shader, the program runs correctly. Its weird cuz shader compiler shows no errors in shader code. Anyway, i provide the value for the "timer" uniform value, every frame, like this:

CODE

        DWORD time = GetTickCount();
        CGparameter timeParam = cgGetNamedParameter(oProgramVS, "timer");
        cgSetParameter1i(timeParam, (int)time);

(Ok, this code doesnt make much sense now but i just want to make it run now so it doesnt matter. )

Do you see what i did wrong so that i got this error?

trouble with writing function which creates an clone of NxActor...

Thu, 08 Sep 2011 14:11:07 -0700

Hello anybody!

I'm writing function which clones actor. But first I need to create some method which copies shapes of actor A to actor B.

By NxActor::saveToDesc(...) we don't have an information about shapes in NxActorDesc. So, there is need to write by using NxActor class.

Here code is:
#define NX_ALL_SHAPES ((1<
bool copyActorShapes(NxActor &destActor, const NxActor &sourceActor, NxU32 shapeFlags, bool updateCenterMass)
{
NxActorDesc dstActorDesc;
bool result = false;
NxU32 num_shapes = sourceActor.getNbShapes();
NxShape *const *shapes = sourceActor.getShapes();
destActor.saveToDesc(dstActorDesc);
for(NxU32 i = 0; i {
NxShape *shape = shapes[i];
NxU32 shapesCount = destActor.getNbShapes();
if(shape->isConvexMesh() && (shapeFlags & (1< {
NxConvexShapeDesc convexShapeDesc;
shape->isConvexMesh()->saveToDesc(convexShapeDesc);
destActor.createShape(convexShapeDesc);
}
else
if(shape->isTriangleMesh() && (shapeFlags & (1< {
NxTriangleMeshShapeDesc triangleMeshShapeDesc;
shape->isTriangleMesh()->saveToDesc(triangleMeshShapeDesc);
destActor.createShape(triangleMeshShapeDesc);
}
else
if(shape->isBox() && (shapeFlags & (1< {
NxBoxShapeDesc boxShapeDesc;
shape->isBox()->saveToDesc(boxShapeDesc);
destActor.createShape(boxShapeDesc);
}
else
{

.................................................................

copy other shapes

................................................................
}
if(shapesCount != destActor.getNbShapes()) // actor currently updated
{
result = true;
}
}
if(updateCenterMass && dstActorDesc.density != 0.0)
destActor.updateMassFromShapes(dstActorDesc.density,0.0);
return result;
}

The problem is: after copying convex shapes they can collide with another actor (which has the same data of shapes) N (?) times and then no collision between them PhysX doesn't detect.
SDK doesn't report errors! In visual remote debugger, these shapes are drawing and no flag which disables collision between these shapes set (as VRD shows).

Should I load and cook mesh again if I want to create actor with the same shapes?
Is there any CPU friendly way to clone objects in SDK?

Any help/opinion needed.

-me

Leave Physx Data on GPU for rendering

Wed, 07 Sep 2011 03:52:12 -0700

Hey..
we are going to start a student project this winter term and we are planning to use physics for fluid simulation.
performance is very important for this project so it's an exclusion criterion if its not possible to leave the simulated particles (or spheres) on gpu and get their position in form of vbo or something like that to use it as input for a geometry shader or computate shader ... without copy the particles to main memory and back to the graphics device

is there such a possibility?

thanks!!

OpenGL PBO

Mon, 05 Sep 2011 02:27:38 -0700

Hi,

I�m in charge to improve our software performance allowing support of multiple HD videos playing.
Our hardware target is NVidia Quadro 4000 - 5000 - 6000 with quick solid state disk.
Our software target is OpenGL 3.2 or greater on Windows 7 64bits with vsync activated.

The previous version of our software have a quite simple and slow architecture:
- A buffering thread per video file in charge to read from solid state disk and decode video stream to CPU memory.
- An unique drawing thread in charge to upload current frame of each video from CPU memory to a corresponding GPU texture.
This was done by glTexSubImage calls from an unique thread owning the unique drawing OpenGL context.

After a lot of reading and some run-time test we found that PBO orphaning is one of the quickest PBO technique.
Thankfully it�s the easiest technique to implement (double buffering or sync fences are more tricky and do not seem to give us better results).

So we just implement the following architecture for further release of our software:

- An additional �uploading� OpenGL context was created to perform PBO filling.
wglShareLists function call allow this context to share display lists of drawing context.
This OpenGL context activation is shared sequentially between each video buffering thread.
A lock system avoid multiple activation of this context in more than one buffering thread at a time.

- Each buffering thread have their own PBO objects and fill them thru this new �uploading� OpenGL context using glBufferData(NULL), glMapBuffer and glUnmapBuffer function calls.

- PBO content are unpacked to GPU texture with the former drawing thread and it�s drawing OpenGL context using glBindBuffer and glTexSubImage function calls.

- A lock per PBO was used avoiding concurrent access between draw and �uploading� thread/context.
This lock ensure that glBindBuffer was not called between glMapBuffer and glUnmapBuffer.

This architecture seem to work correctly on Win7 64bits with one GTX295 or two GTX480 (driver version 280.26), or with one Quadro 4000 (driver version 275.89).
We disable Aero to avoid some texture refreshing problems:
I guest that updating PBO content in another context with another thread can introduce some invalidation issues for Aero.

But more probably, something was missing in our implementation:
Sporadically, on Windows XP 32bit with 280.26 driver on two GTX480,
wglMakeCurrent call on �uploading� context return false with ERROR_TRANSFORM_NOT_SUPPORTED value.

Is our design is correct according to OpenGL and NVidia driver implementation on Windows ?

To resolve refreshing issue, is some glFinish call must be done to complete �uploading� context before SwapBuffer on drawing context ?

Is it an issue to create OpenGL context with old wglCreateContext API instead of new wglCreateContextAttribsARB specifying profile and versions explicitly?

Thanks in advance for your answer.

Maximum pre-rendered frames overriden?

Fri, 02 Sep 2011 12:14:23 -0700

Hi we have a game that is running solid 60 hz in DirectX 9 on a PC with a GTX 480 and Win 7 x64.

We are experience latency between input and display of about 3 frames. We can see the time input is polled to the time there is a screen response drawn.

We don't think the latency is in the game, since it always three frames no matter how long or short the frame draw time is. We can detect the 3 frame latency easily by running in debug and bypassing the windows message queue.
But we also notice (small ~ 50 ms) latency in release.

Driving the game higher then 60 hz seems undesirable because we then have tearing so we use Sync to Vertical Retrace as well.

Since we are manually setting "Maximum pre-rendered frames" to 0 in the NVidia Control panel we were wondering is there some other issue?

Could we some how be ignoring or overriding the control panel in the way we set the card up?

Any advice would be a help.
Thanks
Karl

Shader arrays not working

Fri, 02 Sep 2011 03:36:06 -0700

Hello,

I have a problem compiling effect that contains an array of vertex or pixel shaders, getting a message box saying "Object reference not set to an instance of object".

The array code is for example this one:

VertexShader VertexShaders[3] =
{
compile vs_3_0 VertexShader_PosNorm(),
compile vs_3_0 VertexShader_PosNormTex(),
compile vs_3_0 VertexShader_PosNormTex_Dir_Shadow(),
};

Has anyone seen this problem before? Do I have to give up shader arrays in FX Composer?

Thanks in advance.

Syncing OpenGL app with a stereo DLP 3D Projector

Fri, 02 Sep 2011 01:10:00 -0700

I am developing a 3D Stereo application in OpenGL and I want to use a 3D DLP Ready projector (just as this one) and active shutter glasses (link).

I found that I cannot use nVIDIA 3D Kit, since it doesnt let me control what left/right eyes are displaying in each frame.
nVIDIA driver does control all the process, and I did not find the proper API [3] to tune/sync it as needed.

This webpage from Texas Instrument explains how all DLP 3D compatible projectors syncs left/right frames with compatible DLP active shutter glasses.

Now, my question is:
How can my app tell the projector what the left/right frames are?
Is there a similar approach as the one used in [1] and [2] for the new HDMI standard?
I am using a GeForce card. Do I need a Quadro solution?

Thanks!

References:
[1] HDMI 1.4 Full HD 3D Output via OpenGL Rendering, Is this output format possible with current nvidia hardware ...
[2] What is Full HD 3D (FHD3D)?
[3] nvAPI

DDS 8.8 VU 16 bpp signed format

Thu, 01 Sep 2011 02:47:22 -0700

I have a question about saving a DuDv map (RGB/8) in the DDS 8.8 VU 16 bpp signed format.

What happens is that the saved DDS is much brighter than the original.
For instance the default color (128, 128, 255) becomes (192, 192, 255).

Does anyone know how to avoid this ?

Original DuDv map:

DuDv map saved as DDS 8.8 VU 16 bpp signed:

How to set D3D11 Rasterizer States when using CGFX?

Wed, 31 Aug 2011 05:27:13 -0700

Hi

I've just been trying to change the cullmode of my D3D app using the good old ID3D11RasterizerState interface. That did not work so I tried to set the states for the pass in the CG effect file.

The outcome was a crash of my application because now the ID3DBlob returned by cgD3D11GetIASignatureByPass is NULL. I investigated a little bit further and found the following:

If a place a render state setting in a CGFX pass the technique validation seems to check the state name to be valid but nothing more. So if I provide a typo as state value, such as "Wireframes" instead of "Wireframe" for the "FillMode" state my technique is still considered valid. But the blob I get back from CG to create my input layout is NULL then.

The same applies for states which are missing for Direct3D. As it turns out the CG reference tells me, that certain states, such as everything dealing with the cull mode and polygon facing, is OpenGL only. *grml*

So even placing a usually valid "FrontFace = CW;" line in my CGFX pass I get a NULL blob from cgD3D11GetIASignatureByPass which I assume is due to lack of support for this render state setting in Direct3Dn.

Am I doing something wrong here? If not, how would I change the backface culling in Direct3D when rendering with a CG effect? My own ID3D11RasterizerState seems to be ignored for rendering and it looks as if CG pushes some default rasterizer state onto my D3D device?

regards,
Stefan

When to release a ID3DBlob retrieved by cgD3D11GetIASignatureByPass

Wed, 31 Aug 2011 02:24:34 -0700

Hi

I am about to play a little bit with CG effect files. To create the D3D input layout for the IA stage I do the following in my effect pass activation method:

CODE

[...]
   if (blabla_needs_initialization_blabla)
   {
      uint num = sizeof(layout) / sizeof(layout[0]);
      ID3D10Blob *blob = cgD3D11GetIASignatureByPass(pass);
      checkAndPrintCgError("cgD3D11GetIASignatureByPass");

      HRESULT hr = Direct3D11RenderContext::getD3D11Device ()->CreateInputLayout (layout, num,
                                                                                  blob->GetBufferPointer(),
                                                                                  blob->GetBufferSize(),
                                                                                  &_input_layout);

      blob->Release ();

      if (FAILED (hr))
        {
          printError("creating input layout failed");
          return false;
        }
   }

   Direct3D11RenderContext::getD3D11Context ()->IASetInputLayout (_input_layout);
}

Just as I used to do when working with plain D3D11 where I get the blob fom the D3DX11CompileFromFile function.

But when terminating my app it crashes when I do this ...

CODE

cgD3D11SetDevice(_cg_context, 0);

... which I now boiled down to the release of the blob. The backtrace looks kinda broken going from the cgD3D11SetDevice call to cgD3D11IsProfileSupported twice. But since setting a NULL device to CG frees all internal D3D11 resources cg allocated I stumble across the blob mentioned above. If I do not release it my app seems to work (and terminate) fine in the way it is supposed to do.

So my question is: Am I not allowed to release this blob because CG keeps the blob internally and releases it on its own?

Thanx.

regards,
Stefan

8800GT Video card damage

Mon, 29 Aug 2011 09:26:41 -0700

Hello,

My motherboard is MSI P55-GD65. Video card is evga GeForce-8800GT. Yestarday when I switched on computer I have noticed some color signs on a monitor during BIOS loading, then in a monitor all BIOS symbols were mixed up. After the system has not been loaded at all.

Please suggest does this mean that the Video card is damaged. If yes, what can be the source of the problem: driver, motherboard or something else. And fianlly what you will suggest me to do?

Now I haven't assurance that if I will buy new card it will not be damaged as well.

Please response.

Thanks in advance,
Vardan

How to identify which 2 shapes take place in a collision, the normal reaction between them and the type of the collision?

Sat, 27 Aug 2011 10:43:04 -0700

Hello, I had my question post in a wrong section previously so I'm re-posting it again here.
I'm making a car game. My car is a single actor with a box shape and 4 wheel shapes. I want to check whether the body of the vehicle (the BoxShape) is colliding with any objects in the terrain so i can play the sounds. But the problem I'm having is how can i figure out which shape is colliding? I tried UserContactReport but it doesn't give me any information about which shape hits the terrain. (Wheels also trigger user contact reports and i dont want to make sounds for those collisions).

So does anyone know an answer for my problem.
Is their a method to figure out which shape is had taken place in collision?

PS-I tried separating the wheels into a different actor and then making a fixed joint between them but then the system is not stable. I also try ContactModifyReports but they don't provide me any information about the strength of the collision or whether the two shapes are sliding.

physx and ogre integration

Fri, 26 Aug 2011 15:02:45 -0700

hi all

anyone has an idea about ogre and physx integration ??

I got an idea about NxOgre .. but it has some issues ... and it seems it hasn't support clothing " yet "
and I need clothing

so ... what is the best way to integrate between both

is NxOgre the only way .. or there is another way ...

plz help

thx

tex2DARRAY dx dy

Fri, 26 Aug 2011 12:45:35 -0700

Hello!

float4 tex2D(sampler2D samp, float2 s, float2 dx, float2 dy);
float4 tex2DARRAY(sampler2DARRAY samp, float3 s, float dx, float dy);

In tex2D dx and dy are float2, but in tex2DARRAY dx and dy are float.
But Why?
SampleGrad in DirectX for ddx ddy requires float2, for example.

Tensile test

Fri, 26 Aug 2011 00:25:23 -0700

HI, I've picked softbody as a theme for my bachelor's thesis. I'd like to check physx softbody capabilities and I've decided to perform tensile tests to do so. I know it's more of an "effect feature" then realistic physics feature. Problem is that there is no way to measure force in physx. I've tried several approaches but none of them gave me good results. The only way is to hang SB to a static actor and pull it down with known force, then measure strain. More realistic would be pulling the SB by a kinematic actor and measure force that SB applies to the static actor. I've tried to use a distanceJoint and hook's law, and I found out that it works if I pulled the joint with a force (although there is a strange rate of 1.5 between the force and spirngLength*springConstant), but when I pull the spring with softbody, the spring stops lengthening at some point.
Is there a better way to do this?

how to change the vdpau colorkey color

Thu, 25 Aug 2011 07:00:30 -0700

We are developing a GUI application on RHEL 6.1 that displays a movie in a window. We are using Mplayer-vdpau backend for movie playback. We configured Mplayer to display the movie on a hardware overlay surface so that we can used colorkeying to draw transparent symbology over the video window. The default color of the vdpau color key is green. Anybody knows if it is possible to change the colorkey color for the vdpau video output (-vo vdpau) by a user defined color. Is there a vdpau API or setting to change the color key?

The "-colorkey" setting in Mplayer works for the directx video output. Unfortunately, it is not supported for the vdpau output.

Raycasting usage of PhysX

Tue, 23 Aug 2011 23:59:03 -0700

Hi!
In our project we use PhysX just for raycasting and trigger activation. No rigid body physics are used. Just alot (very alot) number of kinematic/static actors with 30+ number of shapes (convex) and many (very many) raycasts.
The question is, is there anything particular i should enable/disable to increase performance of just these operations? I mean (for example) disable usual collisions between rigid bodies, which we don't use. Things like that.

P.S. We are using batched queries and this is not a raytracer.

Cg Tutorial examples, OpenGL ATI issue

Tue, 23 Aug 2011 20:41:14 -0700

Hi I'm new to Cg, exploring the examples in CG Toolkit and book to see how to convert a matlab code over (part of a thesis).

Video card is ATI HD2300 (very old) in a laptop.
I notice all DX examples work, even DX11, probably using emulation?
All OpenGL examples open an empty window. I need to copy glut32.dll into every folder with exe to run, is that normal?

I programmed openGL a long time ago. And I vaguely remember some GL functions especially how buffers are ordered/setup, worked on this laptop while it shows empty window on another machine, and vice versa when switching the code.

By empty window I mean the console window appears, and GL window appears but content is the desktop.

Chapters 1-2 mention it's possible the same code can work on both DX/GL. Does that mean the host app has no GL/DX specific commands and sends shaders that only use the core Cg runtime (no DX/GL runtime)? I'm curious since I am more familiar with GL than DX, but DX seems to provide more debug info.

3D Vision Streaming with HTML5

Tue, 23 Aug 2011 12:18:49 -0700

I'm trying to follow the Nvidia 3D Streaming with HTML5 pdf instructions: http://www.3dvisionlive.com/3dapps (Download the HTML-5 instructions)

to streaming 3D webm videos in HTML5 with 3DVision on my web-site which is hosted on a linux server running apache,

but I cannot get the video to stream in stereo 3D, only 2D.

Nvidia support was not able to offer any help as soon as linux was mentioned, but pointed me here.

I've uploaded the video.js, video-js.css, and NvStereoUtils.js to the root directory of the server and placed the following code on the new page:

The video plays fine in stereo 3D with the 3DVision video player, side by side half, left right 16:9, and I'm also able to view 3D videos on 3Dvisionlive.com,

but I cannot get it to work on my web-site.

Thanks for your help.

Problem with nVidia FX Composer 2.51 - no objects in Render window

Tue, 23 Aug 2011 09:25:21 -0700

I have problem with nVidia FX Composer 2.5 (2.51.0827.1525). I can't see any objects in Render window, so I can't view my shader result. I can select object, but object is transparent or not rendered. This is happens also when I load existing sample projects.

PhysX math precision

Mon, 22 Aug 2011 01:48:52 -0700

Hi!
I have a problem with maths in PhysX. My math library uses SSE for vectors and computes normalization almost correctly, with just a small difference in 5-6 digit compared to PhysX math.
As an example, this vector was normalized with my maths:
{x=0.96988481 y=-0.17314638 z=-0.17124110 }
And this one is the result of normalization with PhysX:
{x=0.96989465 y=-0.17314814 z=-0.17124283 }
As you can see the difference is very small. However, when trying to pass my vector to ray casting functions as direction, PhysX reports that my vector is not of unit length. Indeed there is a small prescision issue with it, but generally it should be unit length. Vectors that are normalized with PhysX pass OK. But why is this small prescision matters?
I tried "PhysX::NxGetUtilLib()->NxSetFPUPrecision24()" but it doesn't help.
Also tried setting different floating point prescision modes in project's properties - no luck.
Is there anything special that might be done for this?

Release without LTCG

Fri, 19 Aug 2011 08:32:05 -0700

Ideally, we would like to work with release physx libraries without using ltcg (no /GL). To that end, would it be possible to get versions of all libraries compiled as per the release configuration without /GL (rather than switching to the checked configuration)?

NTUNE FOR ALL USERS?

Fri, 19 Aug 2011 08:31:13 -0700

We have determined a profile (underclocking, actually) which helps stabalize some machines we have at a remote location that have undervoltage problems during heavy graphics usage (long story.)

Our question is this: Is there any way that the System Tools ("Performance", to be specific) can be installed by a script, or copying some files and/or registry settings to these remote computers, avoiding the need for user interaction at the actual machine, and make it apply the same rules for ALL USERS of the machine(s)?

All we want to do is:

1.) Install whatever is necessary to allow the clock settings to be modified.
2.) Install the profile we have come up with.
3.) Configure it to use that profile when a specific .exe is started, and return to default clocks when that app exits or the machine is rebooted.

Any details would be appreciated. If necessary, we will remote into these machines and do it that way, but if it could be done with simple file copies / registry flashes it would sure save time.

OS developer question.

Thu, 18 Aug 2011 14:24:02 -0700

Since this is an innovation question I am uneasy about giving details in a public forum. Sadly, there doesn't appear to be a location for this kind of question which I think could perhaps impact OEMs who want to implement your hardware directly.

The question is as above though: I am considering OS development and possibly board level design and integration.

As part of my feasibility study I would like to know what support is in place, or could be actioned for a developer who will be writing an OS who will not be going through an existing OS layer?

Ideally I would like low level access from both assembler and C or C++.

I know assembler is a dirty word in today's climate and I expect this is connected to silicon manufacturers wanting to protect their IP, however there are several assembler-only OSs in research which have been hampered by lack of graphics card drivers.

I have come to Nvidia first at this preliminary stage since my experience of your developer and OS support is an example to the industry. I have an assembler 'head' so having access to the silicon is very attractive to me. This could also unleash previously unseen performance in proof of concept products.

I am more than happy to discuss the proposal with official Nvidia representatives.

The immediate interest from me though concerns the question of support for self-start development for OS development for which OpenGL and DirectX drivers do not already exist.

RSVP.

Jonathan M Kendall. CEO TRinc.

How to view several scenes one by one?

Thu, 18 Aug 2011 07:31:03 -0700

Hi!
In my app i create 10 scenes. Then i connect my app to PVD and wanted to view them one by one, but the viewer shows all objects like they are in one scene, not separate. However the objects browser shows them in separate scenes.
Is there any way to view them independently in real time?

P.S. And is there any way to see my actors in objects browser in REALTIME and not after disconnecting my app and replaying? The list is empty while i view the scene in realtime. Objects appear only after disconnection and replaying. This makes the debug process very long and very hard.

GLSL profiles and "gl_" variables not accessibles

Thu, 18 Aug 2011 02:41:24 -0700

Hi,

I'm getting the error C5059 "gl_" variables are not accessible while using the glslv profile.
The profile documentation says that all the uniform constants and variables listed in the glsl v1.20 doc are available but the use of any gl_ variable triggers the above error. I tried to set the version=120 in the profile options but it doesn't change anything.
Any idea on what I should do to be able to use built-in gl_ variables in my vertex shader ?

Thanks!

ps: my very simple failing code:

CODE

void main(
    float4 position : POSITION,
    out float4 oPosition : POSITION
)
{
    oPosition = mul(gl_ModelViewProjectionMatrix, position);
}

How to rotate monitor on WinXP with nvapi?

Thu, 18 Aug 2011 01:31:07 -0700

As the title states I need rotate the output of one or many external monitors 180deg programatically on WinXP.

As far as i know there are three options:
Nvidia control panel - works but not an option since it needs to be done automatically.
Using user32 ChangeDisplaySettings(Ex) - Does not work, disabled in favor of nvidia control panel?
Using nvapi - NvAPI_SetViewEx is only available on Vista and on.

The nvidia control panel shows that the functionality is there, what am i missing?

Mikael

DDs export plugin introducing pink and greenisch artifacts

Wed, 17 Aug 2011 13:54:46 -0700

We're saving our AO maps as a separate image that are rendered on top of the diffuse pass.

We're saving them as "DXT1 RGB 4 bpp | no alpha" too keep the file size low. After the export the images have pinkish and greenish blobs in them that are clearly visible in the engine.

The attached image is an excerpt of an AO map. It's only 37 pixels wide and high. Open it in an image viewer and zoom in. Sorry about that but re-sampling the image or spaming the forum with a huge document wasn't a good choice either.

On the left side is the Photoshop file and on the right side the same excerpt of the saved dds file.

This happens when the image is converted to grayscale or not. It's clearly the plugin that introduces those color artifacts.

Thanks!

Alex

http://forum-archive.developer.nvidia.com/index.php?act=attach&type=post&id=584

Acceleration for rigid bodies and multithreading

Tue, 16 Aug 2011 18:49:09 -0700

Hi, I am trying to simulate thousand of boxes stacked together and measure the time for simulate method. The problem is, it seems having the NX_SDKF_NO_HARDWARE on or off doesn't result in any significant difference. I have already specified the simulation type of my scene to be simulated by the hardware. Can someone explain this to me?

I also tried to use the multithreading. I have 12 cores PC and I followed the step in the PhysX manual to turn on the multithreading. However, I found out that having the thread to be 2 or 3 seemed to improve the performance than having 1 thread, but as I increment the number of threads to the maximum of 12 threads, the average time it takes to do the simulate method seems to increase. Again, any explanation about this?

How to make a monitor EDID 3D compatible?

Tue, 16 Aug 2011 08:02:05 -0700

Hi,

i have here a autostereoskopic 3D LCD Display with a fitting LVDS converter and it runs perfectly with a software that simulated fidellity like 5 view outputs.
So the display is compatible with the monitors from fidelity but without any trademark.

However, the nvidia driver does not detect this as a 3D display because of the missing 3D compatibility in the EDID EEPROM.
So my questions: what I have to do to get the driver detecting this display as a 3D monitor? Which EDID flag?
Do I need some certifications for that?

best regards
sirnoname

download link for PerfKit R195 Package for 64-bit Linux seems broken?

Mon, 15 Aug 2011 05:47:47 -0700

as title
the 64-bit version file size is only 1.91MB, compare with 32-bit version, the file size is 3xMB,
can somebody fix this please?
thanks

which kind of sorting algorithm is used for CUDA Thrust::sort_by_key?

Mon, 15 Aug 2011 02:00:22 -0700

Hi all,

which kind of sorting algorithm is used for CUDA Thrust::sort_by_key?

Could anyone tell me?

Thanks,

Yik

Bitonic Sorting Network vs Thrust::sort_by_key

Sun, 14 Aug 2011 22:45:29 -0700

Hi all,

I implemented an algorithm which used sorting. I tried Thrust::sort_by_key that took around 0.4s to sort an array with 10^7 elements.

I thought bitonic sorting network should be faster than Thrust::sort_by_key. However, bitonic sorting took about 2.5s to sort the same array mentioned above. I used the bitonic sorting network provided by SDK. I just modified the original bitonic sort a little bit.

Could you tell me why? or give me some advice?

Thanks,

Yik

Aug, 15, 2011

Cloth/Cloth collsions

Sun, 14 Aug 2011 22:21:54 -0700

Can someone tell me if PhysX 2.8 supports collision detection between 2 separate cloth meshes? I've tried and can't get it to work either for 2D meshes or closed 3D meshes like spheres.
Will PhysX 3.1 support it when it's released?

-Bird

PhysX cannot support huge num of joint??

Sun, 14 Aug 2011 01:49:43 -0700

Hi,

We are building a plugin product for maya and using PhysX as physical engine.
We created a scene with 2000 agents (26*2000 actors(boxes) and 25*2000 joints) in scene.

But no matter 32 or 64 bit of PhysX, the engine cannot hold to many joints there. Take a look at the attachment pic, you may see the near agents are no connected by joints, and the far agents are OK. That's because far agents are created eariler.

Is there an attribute or parameter we are miss?? or PhysX itself has this limit.

(Please check out the attachment)

Thank you very much!!!

Problem rotation rigid actor

Fri, 12 Aug 2011 11:00:16 -0700

hello,

I'm trying to control rotation of a floating level.
I have a square and i need to control the rotation in 2 directions.
like this game: http://www.topminigames.com/3d/2134.asp
but i can't make it to work.
I'm looking for the best way to control it.
When i try

getGlobalOrientation()
nxmat33 rot;
rot.rotX(m_RotX);
rot.rotZ(m_RotZ);
setGlobarOrientation(rot);

it doesn't work at all. I looked all over the internet.
But couldn't found an answer to this probably easy problem.
I'm trying to make it work for a week, and getting nervous.
please someone help.

tyvm

Multiple Cameras

Fri, 12 Aug 2011 06:59:03 -0700

Hello, i am new here

i would like to ask, how can i use more cameras ?
i'd like to get ViewProjection matrixes of both cameras into single shader, can i get them ?

thank you very much

Actors Dissappear

Fri, 12 Aug 2011 04:23:26 -0700

Hey, I am developing using a .net wrapper of Physx known as Physx.net.
im developing a vehicle which uses WheelShape in XNA4.0. I have one actor which has BoxShape as its body and having 4 WheelShapes for its wheels. (The whole vehicle is one actor). My terrain is a trianglemesh created using Maya. Each mesh of the terrain is a different actor. It basicaly consist of ground(which is one actor) and few buildings (actor per each building), When my vehicle hits walls in the terrain or when vehicles get packed between eachother some of the vehicles disappear. (The actor of vehicles GlobalPosition is NAN and its GlobalPose Matrix doesnt render the vehicle. )

Doesnt anybody knows a solution for this problem?

Friction doesn't work

Fri, 12 Aug 2011 02:03:15 -0700

Shifted to PhysX SDK versions. Posting can be deleted.

ClosestShape for Point (not Ray)

Fri, 12 Aug 2011 01:58:32 -0700

Hello,

Is there any function similar to raycastClosestShape function that I can utilize to find the closest point among all actor's surfaces given "a point".

I have looked at Collision Detection topic as well but could not find something useful yet.

Thank you,

Creating Force Field from Point Cloud

Thu, 11 Aug 2011 04:25:25 -0700

Dear PhysX Developers,

I have made an articulated model composed of rigid bodies and (spherical & revolute) joints. I need to apply forces to that model according to a point cloud (& normals for each point) that are updated in each frame.

I have looked at creating custom force field and applied force to all actors using it from a single point. However, I need to apply forces from each point to the closest actor "preferably" along the direction of its normal.

I have couple of questions: Should I do this by using force fields? Do I need to create multiple force fields? Would this be inefficient? or should I do this by Raycasting and applying forces to local positions of actors? What would be the easiest, fastest and stable way to accomplish this?

I guess that raycastClosestShape function would be suitable for this purpose. Then, I may add forces to rigid bodies. Anybody knows what is the difference between addForceAtPos and addLocalForceAtPos functions?

Regards,

heightField causing low frame rate

Wed, 10 Aug 2011 09:48:15 -0700

Hi,

This is kind of a weird problem but i cant seem to find out whats wrong.
I have this heightmap terrain. Now everything works fine until i generate a collision mesh.
Suddenly my PhysX is causing me framerate to drop to 7 fps.

I used the same code as i used in my other project and there it runes smoothly.
I ran a profiler on my engine. And PhysXCore.dll is taking 70% of my cpu time.

I hope someone can help me out on this one.

Cloth deformed by user

Wed, 10 Aug 2011 02:53:19 -0700

Hi all,

I am looking for a way to have an object with a deformable mesh (deformation based on user input), which interact with other objects in the scene as a solid object would be.
I want the user to be the only one responsible for this object deformation.

Is a cloth the right way to this or there is another way?

I tried to implement the object based on a cloth, I control the mesh but other objects also modify the mesh.

What is the right approach to this scenario?

Thanks
Emanuel

Cloth colision detection

Wed, 10 Aug 2011 02:37:09 -0700

Hi all,

I'm trying to create a cloth that can move objects like spheres and boxes
but the collision with objects should not bend it or cause any deformation.

one more thing is that I need to change the cloth vertexes positions asynchronously.

I tried to create a triangle mesh and use NxCloth::attachToCore (cloth metal method in Doc) however when I moved the cloth using NxCloth::setPositions(points) it moves like it unattached to
the triangle mesh and therefore the objects keep moving on the triangle mesh even tough the cloth is already found in another location.

Does anyone has a clue how to do that?

Thanks in advance,
Eran.

Stereoscopic imaging on OSes other than Windows

Tue, 09 Aug 2011 10:26:35 -0700

Firstly, many apologies if this post is not in the appropriate forum but I could not see one relating to 3D Vision.

I am working on a project for a computing degree which extends an OpenGL application for creating 3D meshes and which is intended to run on Windows, Linux, and OSX. The UI is created entirely using OpenGL. What I would like to do is use the application in conjunction with a 3D monitor to give the impression that the mesh being edited is projected in front of the screen. Unfortunately, from what I can tell, the relevant nVidia 3D Vision driver and developer tools are only available for Windows.

What I am wondering is if there are any current development tools that would allow me to achieve stereoscopic projection of an OpenGL mesh on Linux and OSX with an nVidia GPU. So far, the only indication I've found that this might be possible is the GLUT_STEREO flag in calls to glutInitDisplayMode() - see http://www.tav.net/3d/3d_stereo.htm. However this seems to be dependent on the driver supporting this feature and probably means it won't be supported on the nVidia Linux/OSX drivers.

Incidentally I was planning to buy a GeForce 560 Ti with a 3D Vision-compatible monitor.

How to set a performance state with NVAPI?

Sun, 07 Aug 2011 15:30:36 -0700

Can anyone give me a proper example how to set a desired performance state (PSTATE) regardless of the current GPU load?

I can set PREFERRED_PSTATE_ID, but it is just "a preferred" state. The drivers can abandon "the preferred" state, although not as rapidly as in the case when PREFERRED_PSTATE_ID is set to 0.

Is it possible at all to keep desired PSTATE regardless of the GPU load?

Density Buffer

Sat, 06 Aug 2011 15:41:22 -0700

Hi,

I'm using the density buffer from a fluid.
The SDK says that these densities are normalized between 0 and 1.
But I am getting back values in the range 0->100.
Is this correct behavior or am I missing something.

I'm reading the densities as PXF32.

Thanks.

A need for PhysX SDK 2.6.0

Fri, 05 Aug 2011 13:19:19 -0700

Hello everyone,

I have been setting up a server for an open sourced game of which I must compile the client with the given source myself. Because of this, I have to round up all the 3rd party libraries myself as well. Now this is all fine and dandy for most things, but I have a problem.

The game was designed for PhysX 2.6.0 and the oldest version of the SDK nVidia's site offers is 2.6.4. This becomes a problem for me because I need that specific version of the SDK, but I have no where to get it. Does anyone have a method for me in which I can get the 2.6.0 version of the SDK? I've already tried building it with 2.6.4; however, many components of the game work horrendously because of the version differences. (Such as the player falling through the floor or certain objects becoming unusable, which breaks the game.)

As I said, I need some way to obtain nVidia's 2.6.0 version of the SDK. I really, really need it, and this project (and all the work I've put into it) will go to waste if I can't find it. I cannot 'fix' the code to work with a newer version of the SDK, unfortunately. I have already tried emailing nVidia's PhysX Developer support but they never replied. (I've waited for about 3 days now).

Thanks,
Valdez42

per pixel alpha blending hardware overlay in Quadro 600?

Fri, 05 Aug 2011 07:50:52 -0700

We are developing a GUI application in RHEL 6 (Linux). In our application, we need to blend video (hardware overlay) and drawing (GUI) layers according to the per-pixel alpha values of the GUI layer so that we can draw semi transparent images and transparent gradient bars over the movies. This technique is named as per-pixel alpha blending mode in ATI M9 (RV250) video cards. Anybody knows if this technique is supported by the NVIDIA Quadro 600 (Fermi) cards.
If so, is there any documentation (i.e Fermi register reference) for developing such a feature?

tex2d() returning 8-bit precision depth - stumped.

Wed, 03 Aug 2011 08:49:01 -0700

As part of a more complex shader I'm sampling a depth buffer using tex2D(). Trouble is, the values it's giving me only have 8 bits of precision, leading to noticeable banding.

To try and isolate the problem I've stripped the shader down to this:

void main ( in v2p IN,
uniform sampler2D ScreenDepth : TEX0,
uniform float4 screenres,
float4 scpos : WPOS,
out float4 oColor : COLOR0)
{
scpos.xy/=screenres.xy;
float dpth = tex2D( ScreenDepth, scpos.xy ).r;
oColor.w = 1;
oColor.xyz = fmod(dpth*10.0f,1.0f);
}

The multiply/fmod line is just to make it easier to see what's going on.

I've tried using a straight depthbuffer and a depth/stencil buffer to no avail.

If I use gDebugger to look at the depth texture, I see a smooth depth gradient with no banding, and everything z-sorts properly so it's not a problem with the texture itself.
If I make dpth a float3 and visualise the r,g and b result of the tex2D call, it's still grayscale ( ie, tex2D().r === tex2D().g === tex2D().B )

The exact same shader was working perfectly on the PS3.

I've spent all day banging my head on this and I'm losing the will to live. Nobody else in the history of the internet seems to have had this problem, so far as I can tell from Google.

Please help. Thanks

Am I missing something?

Wed, 03 Aug 2011 00:57:53 -0700

Having finally gotten around to instrumenting my game engine, was looking forward to integrating some of the data that the PerfSDK exposes, but something isn't right.

- Before installing the PerfSDK I did a clean install of the graphics driver following my usual routine for this: Uninstall via Add/Remove programs, boot into safe mode, use driver sweeper to ensure everything was removed, reboot and install most recent driver available.
- Then I proceeded to download and install the Nvidia PerfKit, which is where things went wrong.
- I downloaded the 32-bit version since the page said to pick the version that matches the application.
- It appeared to have installed fine, except I can't find any trace of it except for gDEBugger.
- The user guide on the PerfKit page will not open, Adobe Reader on this computer says it's corrupted, Evince on my laptop just spits out mangled text.
- No new VS2008, VS2010, or PIX for Windows extensions seem to be registered either.

The computer I am trying to install it on:
OS: Windows 7 64-bit
Mainboard: P7N Diamond
CPU: Intel Core 2 Quad Q9450
Graphics: XFX Geforce 9600GT Alpha Dog Edition

Really getting fed up with WheelShapes

Tue, 02 Aug 2011 10:48:34 -0700

hi

okay so I'm doing a sort of kart racing game thing and I've got some wheelshapes for my karts. Problem is that the results are very frustrating to drive around since I'm kinda just throwing numbers at them to see what works best, and this is really getting frustratiing since I'm not really getting anywhere. Try to go up a tube that loops around? Nope, I start driving up and around the sides instead. Go over a slight bump in the road, landing on some other angled surface? My kart decided to spin all over the place. This isn't what I wanted at all D:

I've tried wheels like this one

CODE

Radius = 0.5
Suspension = 0.5
SpringRestitution = 2000
SpringDamping = 800
SpringBias = 0

BrakeForce = 4000
MotorForce = 5000
TurnAngle = 0
MaxSpeed = 100

LatExtremumSlip = 1.0
LatExtremumValue = 1.0
LatAsymptoteSlip = 2.0
LatAsymptoteValue = 0.7
LatStiffnessFactor = 8000

LongExtremumSlip = 1.0
LongExtremumValue = 1.0
LongAsymptoteSlip = 2.0
LongAsymptoteValue = 0.7
LongStiffnessFactor = 8000

or this one

CODE

Radius = 0.5
Suspension = 0.5
SpringRestitution = 7000
SpringDamping = 300
SpringBias = 0

BrakeForce = 5000
MotorForce = 4000
TurnAngle = 0
MaxSpeed = 100

LatExtremumSlip = 0.01
LatExtremumValue = 1
LatAsymptoteSlip = 0.04
LatAsymptoteValue = 0.8
LatStiffnessFactor = 10000

LongExtremumSlip = 0.01
LongExtremumValue = 1
LongAsymptoteSlip = 0.04
LongAsymptoteValue = 0.8
LongStiffnessFactor = 6000

and I really feel that I'm just not getting anywhere.

My kart is 1.5x0.5x1.4 with a box shape, mass is about 400ish

here's an older build: http://db.tt/Gc43ow5 (~10 MB) - I'm using a C# wrapper so you'll also need .net 4.0+ installed

I'm really tempted to just throw this whole thing out and start over with spheres and d6 joints, but I don't know if that'll be any easier either.

Any advice?

NVIDIA Fermi Parallel Data Cache

Tue, 02 Aug 2011 10:17:21 -0700

Hello,

I am wondering if any of the codes provided in the CUDA SDK, test the power of the NVIDIA Parallel DataCache capability on the Fermi architecture. If there are any, what are they ?

Thank you.

Sharing triangle mesh data between scenes

Tue, 02 Aug 2011 06:27:14 -0700

Hi!
I need to create many separate scenes for processing. Each will be created and simulated in a separate thread to spread the load across processor cores. But they all will use the same set of triangle meshes (static level geometry). How can i optimize my memory usage because of this? Will they all copy into each scene during creation or share same space?
Also is this a good idea anyway?
What i'm doing is a game server. Each new scene is a new game instance with players. They (scenes) must not interact with each other, but share same triangle mesh data.

Octree Texture

Mon, 01 Aug 2011 17:13:41 -0700

Hey, I'm trying to get the Octree texture from chapter 37 in GPU Gems 2 to work.

I have the octree loaded into a 3d texture, and everything seems fine in that end (the data is volume data, so instead of storing colour information, I store scalar values which I can then use on the gpu to access another texture, the transfer function). Now I'm trying to get the it working with raycasting.

This is the raycasting code:

CODE

fragment_out fragment_main(VertexShaderOutput input, uniform sampler2D backTex,
                uniform sampler2D frontTex, uniform sampler3D volumeTex,
                uniform sampler1D transFunction,
                uniform float3 stepsize, uniform float iterations)
{
    fragment_out OUT;

    float2 texC = ((input.Pos.xy /= input.Pos.w) + 1 ) / 2;

    float3 front = tex2D(frontTex, texC).rgb;
    float3 back = tex2D(backTex, texC).rgb;

    float3 dir = normalize(back - front);
    float3 pos = front;

    float4 dst = float4(0, 0, 0, 0);
    float4 src = 0;

    float4 value = 0;
    float3 Step = dir * stepsize;

    for(int i = 0; i < iterations; i++)
    {
        value = tex3D(volumeTex, pos);

                src = tex1D(transFunction, value.a);
        src.a = 1 - pow((1 - src.a), 0.5f / 0.5f);

        src.rgb *= src.a;
        dst = (1.0f - dst.a) * src + dst;

        //break from the loop when alpha gets high enough
        if(dst.a >= .95f)
            break;

        //advance the current position
        pos.xyz += Step;

        //break if the position is greater than <1, 1, 1>
        if(pos.x > 1.0f || pos.y > 1.0f || pos.z > 1.0f)
            break;
    }

And the Octree Lookup code from GPU Gems. I switched the discard statement for a break statement as it was not printing anything at all during testing.

CODE

float4 treeLookup( uniform sampler3D IndirPool, uniform float3 invS, uniform float N, float3 M )
{

    float4 I = float4(0.0, 0.0, 0.0, 0.0);

    float3 MND = M;

    // fixed # of iterations
    for (float i = 0; i < 100; i++)
    {
        float3 P;

        // compute lookup coords. within current node
        P = (MND + floor(0.5 + I.xyz * 255.0)) * invS;

        // access indirection pool
        if (I.w < 0.9)                   // already in a leaf?
            I = (float4)tex3D(IndirPool, P); // no, continue to next depth

        if (I.w > 0.9)
            break;

        if (I.w < 0.1) // empty cell
            break;

        // compute pos within next depth grid
        MND = MND * N;
    }

    return (I);
}

To implement the two together, I figured rather than sampling the volumeTexture in the fragment shader, I would call treeLookup with pos, and using the returned value to access the transfer function. However, doing this results in nothing being displayed.

I think maybe I'm storing the pointers to the indirection grids incorrectly. For example, when storing a pointer to the second indirection grid, I set the RGB values to be (1, 0, 0). Is this correct or should I divide each dimension by the size of the texture like (1 / texWidth, 0 / texHeight, 0 / texDepth)?

Any help would be really appreciated, thanks!

How can I limit the speed of bodies?

Mon, 01 Aug 2011 11:55:26 -0700

How can I limit the linear speed of bodies on my simulation? I found just this method setMaxAngularVelocity (NxActor)

OpenCL pi Calculation

Mon, 01 Aug 2011 09:25:08 -0700

Hi Everone;
Is there anybody who can give me a direction about the OpenCL pi number calculation..Or who has the sample pi calculation code on OpenCL.

OpenCL exceptions on particular functions.

Mon, 01 Aug 2011 07:18:05 -0700

Hi all,

I've been recently working on some layer library for OpenCL and found two functions to constantly throw exceptions when called:

clCreateUserEvent
clCreateSubBuffer

I'd tested these calls with other vendors and it worked OK.

I do not think that there is any need to provide samples - these calls are very basic - just pass context and for second one also - r-w buffer object, CL_MEM_READ_WRITE flag and region definition of half of the original buffer...

I have theses exceptions always - even when expecting to receive regular errors (some negative testing...) - like there is something wrong with function calls mapping. It looks that there is no error check done (except for simple CONTEXT != NULL on ICD level - so this is only situation when there is no exception)...

Is this a known Issue and will be addressed in the future? Or I'm doing something wrong?

Regards
SW

Edit:

I'd tested on M880 and GTX 480.

Display 3D image in WPF application

Mon, 01 Aug 2011 00:12:24 -0700

Hi guys,

I have an existing interactive WPF .NET 3.5 application, which displays jpg images on a GUI with overlayed controls. The idea is to make it 3D with NVidia 3D Vision on a 3D TV. As far as I read in this post, there is a Silverlight plugin, which allows 3D video, but no still images. In addition, it's Silverlight and not compatible with WPF, right? Google gave me the following ideas:

Create an XNA (DirectX) application and embed WPF: http://www.c-sharpcorner.com/UploadFile/iersoy/256/
Embedding a D3DImage element in WPF/XAML: http://www.codeproject.com/KB/WPF/D3DImage.aspx

Both solutions require a DirectX 3D model. The problem: I have stereoscopic images only (1 image each for left and right eye) and no 3D (DirectX) model. I can create a .jps file and view it in the Nvidia 3D photo viewer or in the JPSViewer.

It would be nice to "just" embed a stereoscopic image into XAML/WPF. Is this possible? Any hint greatly appreciated! Thank you!

[v.3.x] Motorized

Sun, 31 Jul 2011 10:49:13 -0700

In order to simulate a steerable cylindrical joint like suspension + steer , i m using a d6 joint

CODE

CREATE:
j = PxD6JointCreate(*gPhysicsSDK, a0, P1, a1, ps);
// ALONG Y AXIS
j->setMotion(PxD6Axis::eY,PxD6Motion::eLIMITED);
j->setMotion(PxD6Axis::eSWING1,PxD6Motion::eLIMITED);

PxD6JointDrive drive(1000, 10.0f, PX_MAX_F32, true);

j->setDrive(PxD6Drive::eSWING, drive);
j->setDrivePosition(PxTransform::createIdentity());
j->setDriveVelocity(PxVec3(0),PxVec3(0));

In order to steer joint i using setDrivePosition / setDriveVelocity but joint does not move it.
How can i use drive model?
Thanks.

PerfSDK does not work on Win7 x64

Sun, 31 Jul 2011 06:35:00 -0700

Hi there,
I'm trying PerfSDK to get some system low level counters to help profile my application. But, seems the api does not work...

Win7 x64,
PerfKit 6.7 for win7 x64,
Geforce GTX 460, driver is 6.70.1201.2315.

I try it in a D3D SDK sample , ParallaxOcclusionMapping, running on dx9. Paste the code block below:

********************************************************************************
*********************
DXUTInit( true, true ); // Parse the command line and show msgboxes
DXUTSetHotkeyHandling( true, true, true ); // handle the defaul hotkeys
DXUTCreateWindow( L"Parallax Occlusion Mapping" );
DXUTCreateDevice( true, 800, 600 );

NVPMRESULT hr = NVPMInit(); // it returns NVPM_OK, expected, right?
hr = NVPMEnumCounters(MyEnumFunc); // it returns NVPM_ERROR_NOT_INITIALIZED, I don't know why...
DXUTMainLoop();

hr = NVPMShutdown(); // it returns NVPM_ERROR_NOT_INITIALIZED, I don't konw why..
********************************************************************************
*********************

I found there were any other topics may have the same issues, and I raise this again to pop it to top .

Or your team has any plan updating the tool ?

Welcome any feedback .

Thanks.
-Hui

PhysX Helicopter

Sun, 31 Jul 2011 01:13:11 -0700

Hello, I am semi-new to physX and was wondering if making a helicopter or other aircraft would be even considerably possible, I cant figure out any possible way and I have gone through all the different forces and things in the learning documentation that came with 2.8.1, And cant seem to think of a way, if anyone has any ideas or methods please tell me it would be really awesome and thanks in advance!

DDS and CS3

Sat, 30 Jul 2011 22:17:33 -0700

What is the proper proceedure to effectively install the DDS files in CS3 in a windows 7 64 bit environment?

I tried the 32 bit and 64 bit files and when I use the 32 bit file i get an error opening up photoshop that says missing D3x9_43.dll, and I don't have the option to save in a .dds format. If I uninstall the dds 32 bit program the missing D3x9_43.dll error goes away.

When I load 64 bit version no errors but I still don't have the option to save in .dds format. Kind of frustrating..................

I've just uninstalled and cleaned CS3. I've done a new fresh load of CS3. What are the correct steps from here?

Which file should I download? etc.

Thank you in advance.

Bill

Change gamma programatically on WinXP

Fri, 29 Jul 2011 00:51:52 -0700

Hi,

I am looking for a way to alter the gamma curve set in NVIDIA Control Panel programmatically.

On Windows 7 this can be done by using Gdi32.SetDeviceGamma but on Windows XP Gdi32.SetDeviceGamma does not seem to change the same gamma ramp as the NVIDIA Control Panel.

I have looked in NVAPI but all the gamma related functions seems to apply to video color settings. How can i change the gamma curve for the desktop?

Thanks in advance.

NxArray of NxArray

Thu, 28 Jul 2011 18:53:54 -0700

I am using the SDK 2.8.4 and I would like to use a NxArray of a NxArray, I try this but does not work.

CODE

    NxArray<NxArray<int>> test;
    NxArray<int> t;
    t.push_back(1);
    t.push_back(2);
    test.push_back(t);

I receive the error mensage "Access violation reading location" on line "test.push_back(t);"

How can I do this?

Subclass of NxScene

Thu, 28 Jul 2011 18:48:38 -0700

I would like to create a subclass NxScene1 from NxScene (SDK 2.8.4). I would like that the class NxScene1 has atribbutes like:

CODE

#include <NxScene.h>
#include "NxPhysics.h"

using namespace std;

class NxScene1 : public NxScene{
public:
    Class1* class1;
        Class2* class2;
public:

    NxScene1(void);
    ~NxScene1(void);
    void initNxScene1();
};

To instantiate in memory I create the method initNxScene1:

CODE

void NxScene1::initNxScene1(){
class1 = new Class1();
class2 = new Class2();
}

Because the scenes was created by the method createScene from NxPhysicsSDK (SDK 2.8.4.) but I cast to NxScene1*, like:

CODE

Simulation::gScenes[Simulation::gBaseScene] = (NxScene1*)scene;
Simulation::gScenes[Simulation::gBaseScene]->initNxScene1();
Simulation::gPhysicsSDK->setParameter(NX_VISUALIZATION_SCALE, 1.0f);

But On the line 3 of that code I receive the exception "Access violation writing location".

What I can do to use my class NxScene1?

Set a 3DLUT in GC RAMDAC ?

Thu, 28 Jul 2011 02:38:11 -0700

Hello all !

I'm working on a color calibration and proofing software.
I'm looking for a way to set a 3DLUT directly into a graphic card RAMDAC (and I'm not sure it's possible).

I'm already using some OS API functions (XF86VidModeSetGammaRamp on Gnu/Linux, CGSetDisplayTransferByTable on MacOS and SetDeviceGammaRamp on windows) to set a gamma ramp in the graphic card RAMDAC.
Exactly as I set these gamma ramps (which are in fact three 1D LUTs), I'd like to set a 3DLUT.
Unfortunetly, such function doesn't exist in OS APIs.

However I was told it may be possible to do so via Nvidia's driver (all our graphic cards are Nvidias).
I searched in NVAPI without finding anything.

The point of using such a function is to modify the way colors are displayed everywhere (means colors are corrected even on the desktop or in windows explorer for example).

Have anyone a clue ?

Thanks a lot,

Marie

nvcompress with incredibuild

Wed, 27 Jul 2011 21:33:36 -0700

Hi,

I am trying to farm out some nvcompress processes via incredibuild but it seems that when it runs via incredibuild it does not use CUDA. If I run it manually on the PCs that it fails on it will work correctly with CUDA installed. Does anyone know if this is possible or am I chasing my tail and it just won't work because of crazy incredibuild hooking? I've tried the 64bit versions of texture tools but our version of Incredibuild doesn't seem to handle that very well.

Also, _sometimes_ it will work on some of the PCs that it runs on via incredibuild.

It seems (after putting some debug prints into nvcompress) that it succeeds to find the driver but then it will fail on cudaGetDeviceCount with a status of 10 (which is invalid device I think). It is pretty hard to debug incredibuild to find what dlls it is pulling over and using when it is running on the host pcs so that makes it quite hard to debug :/

arm + cuda

Wed, 27 Jul 2011 13:17:54 -0700

I am thinking to design an arm + GPU system, software: linux + cuda. They connect each other through pci-e.
Does anyone have any idea whether this is possible or not?

about perspective shadow maps implement(DX9)

Wed, 27 Jul 2011 00:01:19 -0700

Hello all!

where is the sample of perspective shadow maps implement(DX9) ?

please tell me, thanks !

Clone Joint

Mon, 25 Jul 2011 15:05:05 -0700

I would like to clone Joint from a source scene to a destination scene. But i use this code and the joint was create on the source scene, not the destination scene (indexDestScene).

CODE

NxJoint* Simulation::cloneJoint(NxJoint* jointSource, int indexDestScene){
    NxJoint* joint;

    NxJointType type = jointSource->getType();
    if(type==NxJointType::NX_JOINT_REVOLUTE){

    }
    else if(type==NxJointType::NX_JOINT_D6){
        NxD6JointDesc d6JointDesc;
        jointSource->isD6Joint()->saveToDesc(d6JointDesc);

        joint = gScenes[indexDestScene]->createJoint(d6JointDesc);
    }

    return joint;
}

Plugin tutorial?

Mon, 25 Jul 2011 13:40:47 -0700

Is anyone aware of a tutorial on building plugins? I'm thinking the best way to get my float textures would be to write a custom file importer. I'm looking at the SDK examples, but a guided overview would be very helpful.

Cloth simulation/examples on 3.0 ? on linux?

Mon, 25 Jul 2011 10:02:18 -0700

Hi,

I'm a new PhysX user and I want to simulate cloth behavior on GPU.
Maybe I have missed something but I don't understand. I have downloaded the 3.0 PhysX SDK but I can't find the cloth classes Nx... Neither on linux nor on windows. Classes which are in the SDK 2.8 for windows.

So are those classes in a plugin or something else for the release 3.0? are they available for linux?

Thanks in advance for your help

Erik

Prismatic Joint: Extension and Spring

Mon, 25 Jul 2011 06:54:37 -0700

Hi,

is there a possibility to read the extension of a prismatic joint in PhysX 2.8.3? I tried it with calculating the distance between the two involved actors but this isn't a generalised solution.

Furthermore I tried to control the prismatic joint like the revolute joint with a spring damper to fix it to a specific position / extension but I couldn't find something similar for the prismatic joint. I don't want to reset the limit planes because this should lead to very fancy behaviour and is also not very continuous.

Thanks for any hints

Cheers
Dennis

How i could create a DDS Header?

Mon, 25 Jul 2011 03:24:31 -0700

Hi,

i would write a program which convert a bmp to DDS File in Visual C++ 2010.
I know that is there are a bunch of softwares to do it, but i want to split the header from the data to save them in seperatly files.

How i could create the header?

My target is to create files with a resolution of 512x512 and type of DXT1, w/o mip-mapping, as DDS-File(where the DDS Headers is separatly stored).

The onliest what i found in this forum was this thread -> http://forums.developer.nvidia.com/index.php?showtopic=944
but it was not the information i searched for.

Could someone tell me please where i'll find some example codes to create a dds file?

I'm installed the nvtt and the Nvidia graphics SDK, but got in stuck, which header (nvtt.h?) i need to use and so on.

thanks in advance,
raanubis

PS: @Mod or Admin, Sorry wrong forum, please delete or move, thank you

NVAPI force a resolution

Sun, 24 Jul 2011 20:54:05 -0700

I am using the NVAPI to set a display resolution, however there are some circumstances where the EDID information for the monitor will not be available (due to the use of a KVM). The monitor in question is known to support the resolution but it can't be set while this EDID information is not available.

I can set a custom resolution through the NVIDIA control panel, and then set the required resolution. This process needs to be done automatically. I am using NvAPI_SetViewEx to set the resolution this takes a NV_DISPLAY_PATH_INFO struct which contains a NV_DISPLAY_PATH struct.

This has a member bForceModeSet with a comment: "(IN) Used only on Win7 and higher during a call to NvAPI_SetViewEx(). Turns off optimization & forces OS to set supplied mode."

I have not been able to successfully use this variable to force a resolution to be set in a Windows 7 environment. Is the purpose of this member to allow a display resolution to be forced? Or is it only used to force clone mode, extended mode etc.?

EDID Issues with KVM

Sun, 24 Jul 2011 04:12:17 -0700

[font="Arial"][/font]hi,
as u can see in my attachment, i use a brandnews DisplayPort KVM and 3 nvidia quadro cards

my problem: no signal at all on my screen

the manufacturer of the kvm gave me now this answer:

it must be an EDID issue that you are experiencing
EDID refers to the protocol that the video card(s) and monitor uses to communicate the "best" resolution supported by both.

KVM's do not pass this information, as there are multiple video cards used in a KVM.

The only way to overcome this obstacle, is to force a specific video resolution on all computers.

Do this you will need a special software utility, generally provided by the video card manufacturer, to force a specific video resolution.

Different manufacturers have different utilities for this, and should be able to assist you with forcing a specific video resolution.

does anybody know what i can do ??

Problem with glutMotionFunc(MotionCallback)

Sat, 23 Jul 2011 13:04:38 -0700

When I rotate qx (move cursor on the x axis of the window) the camera do not rotate on the z axis of the camera, Dir is the direction of view camera.

CODE

    glutMouseFunc(MouseCallback);
    glutMotionFunc(MotionCallback);
    MotionCallback(0, 0);

void Simulation::MouseCallback(int button, int state, int x, int y)
{
    mx = x;
    my = y;
}

void Simulation::MotionCallback(int x, int y)
{
    int dx = mx - x;
    int dy = my - y;

    Dir.normalize();
    N.cross(Dir,NxVec3(0,1,0));

    NxQuat qx(NxPiF32 * dx * 20/ 180.0f, NxVec3(0,1,0));
    qx.rotate(Dir);
    NxQuat qy(NxPiF32 * dy * 20/ 180.0f, N);
    qy.rotate(Dir);

    mx = x;
    my = y;
}

Video of motionCallback problem of GLUT

Where can I get PhysX 2.8.4?

Sat, 23 Jul 2011 11:18:40 -0700

I'm using PhysX SDK 2.8.4, and

CODE

NxGetCookingLib(NX_PHYSICS_SDK_VERSION);

returns 0. Where can I get the appropriate PhysX version?

Render with GLUT

Sat, 23 Jul 2011 09:00:31 -0700

Render with GLUT at SampleSceneExport (PhysX SDK 2.8.4)

I am using the GLUT to render a scene but I have the problem that in certain position cameras some objects of the scene dissapear. Someone knows what is the problem? I add two screenshots of the scene one with the object in scene and other just change the camera position that object dissapear.

The code:

CODE

void DrawWireBox(const NxBox& obb, const NxVec3& color, float lineWidth)
{
    // Compute obb vertices
    NxVec3 pp[8];
    NxComputeBoxPoints(obb, pp);

    // Draw all lines
    const NxU32* Indices = NxGetBoxEdges();
    for(NxU32 i=0;i<12;i++)
    {
        NxU32 VRef0 = *Indices++;
        NxU32 VRef1 = *Indices++;
        DrawLine(pp[VRef0], pp[VRef1], color, lineWidth);
    }
}

void DrawWireBox(NxShape* box, const NxVec3& color, float lineWidth)
{
    NxBox obb;
    box->isBox()->getWorldOBB(obb);

    DrawWireBox(obb, color, lineWidth);
}

void DrawBox(NxShape* box, const NxVec3& color)
{
    NxMat34 pose = box->getGlobalPose();

    glPushMatrix();
    SetupGLMatrix(pose.t, pose.M);
    NxVec3 boxDim = box->isBox()->getDimensions();
    glScalef(boxDim.x, boxDim.y, boxDim.z);
    RenderBox(2.0);
    glPopMatrix();
}

void DrawWireShapeIME(NxShape *shape, const NxVec3& color)
{
    glColor4f(color.x, color.y, color.z, 1.0f);
    switch(shape->getType())
    {
    case NX_SHAPE_PLANE:
        DrawWirePlane(shape, color);
        break;
    case NX_SHAPE_BOX:
        DrawWireBox(shape, color);
        break;
    case NX_SHAPE_SPHERE:
        DrawWireSphere(shape, color);
        break;
    case NX_SHAPE_CAPSULE:
        DrawWireCapsule(shape, color);
        break;
    case NX_SHAPE_CONVEX:
        DrawWireConvex(shape, color);
        break;
    case NX_SHAPE_MESH:
        DrawWireMesh(shape, color);
        break;
    default:
        break;
    }
}

void DrawShapeIME(NxShape* shape, const NxVec3& color)
{
    glColor4f(color.x, color.y, color.z, 1.0f);
    switch(shape->getType())
    {
    case NX_SHAPE_PLANE:
        DrawPlane(shape);
        break;
    case NX_SHAPE_BOX:
        DrawBox(shape, color);
        break;
    case NX_SHAPE_SPHERE:
        DrawSphere(shape, color);
        break;
    case NX_SHAPE_CAPSULE:
        DrawCapsule(shape, color);
        break;
    case NX_SHAPE_CONVEX:
        DrawConvex(shape, color);
        break;
    case NX_SHAPE_MESH:
        DrawMesh(shape, color);
        break;
    case NX_SHAPE_WHEEL:
        DrawWheelShape(shape);
        break;
    default:
        break;
    }
}

void DrawActorIME(NxActor* actor)
{
    if(actor != NULL)
    {
        NxShape*const* shapes = actor->getShapes();
        NxU32 nShapes = actor->getNbShapes();
        if (actor == NULL || actor->getName() == NULL)
        {
            while (nShapes--)
            {
                DrawWireShapeIME(shapes[nShapes], NxVec3(1,1,1));
            }
        }
        else
        {
            if (strcmp(actor->getName(), "Campo") == 0)
            {
                while (nShapes--)
                {
                    DrawShapeIME(shapes[nShapes], NxVec3(0,1,0)); //Verde
                }
            }
            else if(strcmp(actor->getName(), "Bola") == 0)
            {
                while (nShapes--)
                {
                    DrawShapeIME(shapes[nShapes], NxVec3(5,0.5,0)); //Laranja //1,0.5,0
                }
            }
            else if(strcmp(actor->getName(), "Robo1") == 0 || strcmp(actor->getName(), "Robo2") == 0 || strcmp(actor->getName(), "Robo3") == 0 || strcmp(actor->getName(), "Robo4") == 0 || strcmp(actor->getName(), "Robo5") == 0)
            {
                while (nShapes--)
                {
                    DrawShapeIME(shapes[nShapes], NxVec3(0,0,1)); //Azul
                }
            }
            else if(strcmp(actor->getName(), "Robo6") == 0 || strcmp(actor->getName(), "Robo7") == 0 || strcmp(actor->getName(), "Robo8") == 0 || strcmp(actor->getName(), "Robo9") == 0 || strcmp(actor->getName(), "Robo10") == 0)
            {
                while (nShapes--)
                {
                    DrawShapeIME(shapes[nShapes], NxVec3(1,1,0)); //Amarelo
                }
            }
            else
            {
                while (nShapes--)
                {
                    DrawShapeIME(shapes[nShapes], NxVec3(1,1,1)); //Branco
                }
            }
        }
    }
}

void Simulation::IdleCallback()
{
    glutPostRedisplay();
}

void Simulation::ReshapeCallback(int width, int height)
{
    glViewport(0, 0, width, height);
}

void Simulation::setupCamera()
{
    // Setup projection matrix
    glMatrixMode(GL_PROJECTION);
    glLoadIdentity();
    gluPerspective(60.0f, (float)glutGet(GLUT_WINDOW_WIDTH) / (float)glutGet(GLUT_WINDOW_HEIGHT), 1.0f, 10000.0f);
    gluLookAt(gEye.x, gEye.y, gEye.z, gEye.x + gDirection.x, gEye.y + gDirection.y, gEye.z + gDirection.z, 0.0f, 1.0f, 0.0f);

    // Setup modelview matrix
    glMatrixMode(GL_MODELVIEW);
    glLoadIdentity();
}

void Simulation::RenderCallback()
{
    // Clear buffers
    glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);

    //Setup camera
    setupCamera();

    //Draw
    glPushMatrix();
    for (NxU32 i = 0; i < gMaxScenes; ++i)
    {
        if (gScenes[i])
        {
            glColor4f(0.6f,0.6f,0.6f,1.0f);
            for(unsigned int j = 0; j < gScenes[i]->getNbActors(); j++ )
            {
                DrawActorIME(gScenes[i]->getActors()[j]);
            }
        }
    }
    glPopMatrix();
    //glFlush();
    glutSwapBuffers();
}

void Simulation::main(int argc, char **argv)
{
#if defined(_XBOX) || defined(__CELLOS_LV2__)
    glutRemapButtonExt(8, '1', false); // Left shoulder to save
    glutRemapButtonExt(9, 'c', false); // Right shoulder to clear
    glutRemapButtonExt(8, '3', true); // Shift + left shoulder to load
#endif
    // Initialize Glut
    glutInit(&argc, argv);
    glutInitWindowSize(512, 512);
    //glutFullScreen();
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    gMainHandle = glutCreateWindow("Simulacao");
    glutSetWindow(gMainHandle);
    glutDisplayFunc(RenderCallback);
    glutReshapeFunc(ReshapeCallback);
    glutIdleFunc(IdleCallback);
    glutKeyboardFunc(callback_key);
    glutKeyboardUpFunc(callback_keyUp);
    glutSpecialFunc(ArrowKeyCallback);
    glutMouseFunc(MouseCallback);
    glutMotionFunc(MotionCallback);
    MotionCallback(0, 0);
    atexit(ReleaseNx);

    // Setup default render states
    glClearColor(0.3f, 0.4f, 0.5f, 1.0);
    glEnable(GL_DEPTH_TEST);
    glEnable(GL_COLOR_MATERIAL);
#if !defined(__PPCGEKKO__)
    glEnable(GL_CULL_FACE);
#endif

    // Setup lighting
    glEnable(GL_LIGHTING);
    float AmbientColor[] = { 0.0f, 0.1f, 0.2f, 0.0f };
    glLightfv(GL_LIGHT0, GL_AMBIENT, AmbientColor);
    float DiffuseColor[] = { 1.0f, 1.0f, 1.0f, 0.0f };
    glLightfv(GL_LIGHT0, GL_DIFFUSE, DiffuseColor);
    float SpecularColor[] = { 0.0f, 0.0f, 0.0f, 0.0f };
    glLightfv(GL_LIGHT0, GL_SPECULAR, SpecularColor);
    float Position[] = { 100.0f, 100.0f, 400.0f, 1.0f };
    glLightfv(GL_LIGHT0, GL_POSITION, Position);
    glEnable(GL_LIGHT0);

    // Initialize physics scene and start the application main loop if scene was created
    if (true)
        glutMainLoop();
}

Vertex/data texures?

Fri, 22 Jul 2011 14:09:47 -0700

I'm generating a Vector4 texture containing some fairly involved data structures in my XNA app and passing it to a shader. I'd like to use FX Composer to debug that shader, but I'm unsure how to handle this non-image texture?

I can get the byte array containing the texture data and dump it to a file; is there any way to then load that into FX composer? If this isn't natively supported, is it possible to create a texture and populate a texture using the scripting interface?

NxCompartment implementation differences between PC and Xbox 360 ?

Fri, 22 Jul 2011 11:43:09 -0700

hi there,

PhysX 2.8.4 (UE3 integration)

are there any special caveats to using NxCompartments on the Xbox 360 vs. the PC ?

i've put an NxActor into a rigidbody compartment and i want it to collide with a rigidbody in the primary scene. both the RB in the primary scene and the RB in the compartment are moving at the time of expected collision. however, the mesh itself is not changing shape or anything.

if i do not put the RB into a compartment then the two RBs collide just fine. if i put the RB into a compartment, the two RBs collide on the PC but not on the Xbox 360.

any thoughts?

TIA -phil.

Nvapi?

Thu, 21 Jul 2011 14:02:19 -0700

Is anyone alive here? Does anyone have any better example code with Nvapi other than what Nvidia has posted? I'm attempting to change the displays from vertical span to dualview and back. I get an NVAPI_INVALID_ARGUMENT response and don't really know what else to try to move forward.

Getting Cg shaders working on ATI cards

Thu, 21 Jul 2011 02:22:20 -0700

Probably not the first time asked How to get Cg shaders (in an OpenGL program) work on a (modern) ATI card?

Since I never used ATI cards, I programmed all my shaders without worries. But now that another guy wants to test my program, it comes with dozens of bugs. In a nutshell:
- Can't compile geometry programs
- Can't do vertex-texture-fetch
- Can't do MRT on channel 5 and 6
- Can't do tex2Darray
- Can't do loops

Well, this card certainly supports all those techniques. I think the wrong profile is chosen with cgGLgetLatestProfile, as it turns
CG_PROFILE_ARBVP1 and CG_PROFILE_ARBFP1

Ok... but what to chose else? I saw the GLSL profile, but doesn't that require to rewrite all shaders in GLSL as well? When I just change the profiles to GLSL (VP, FP and GP), I get c9999 errors.

* Using Cg / CgGL DLL's version 3.0.0.15.

Rick

Softbody forces within the body and inhomogenous Softbodies

Wed, 20 Jul 2011 22:42:28 -0700

Hi there,

i have two question.
First, is it possible to add a force directly into a softbody. For Deformation within but without a deformation on the surface?

Second is the goal to create an inhomogenous softbody. There should be differences in the stiffnes in the softbody. Is that possible?

Short Update: Is it possible to adjust the mass-spring model. For example a link between to tetras?

Thank you a lot.
Greets
Sascha

Windows XP Display Driver programming

Wed, 20 Jul 2011 11:26:36 -0700

Dear Knowledgeable one.

Please type in the answer to this question.

What files do you have to program if you need to flip and transpose the 800x600 display matrix?

Is there a Windows XP open source tool kit to program an application like Nvidia�s Zoom?

PhysX Visual Indicator always displays CPU

Tue, 19 Jul 2011 20:25:58 -0700

Hi, Im enabling hardware like that:
desc.flags &= ~NX_SDKF_NO_HARDWARE;

pPhysicsSDK = NxCreatePhysicsSDK( NX_PHYSICS_SDK_VERSION, &defAllocator, &myErrorOBJ, desc, &errorCode );

if( pPhysicsSDK->getHWVersion() == NX_HW_VERSION_ATHENA_1_0 )
int x = 7;//breakpoint gets hit here

But the damn visual indicator, enabled trough nVidia ctrl panel, always shows CPU..I got all nervous, and them checked with Batman Arkhan Asylum, but it works just fine( it displays GPU )...so..what the hell is going on?

[Q]: can i adjust the time scale for a specific NxActor within a scene ?

Tue, 19 Jul 2011 19:05:41 -0700

hi there,

i would like to step a particular NxActor (or a few of them) at a slower rate than the rest of the rigidbody actors in my scene. so basically i want to timescale them. but i still want these actors to interact with the other actors in the scene (collisions etc.).

is there any way to do this ? using PhysX 2.x (not 3.0)

TIA

[Q]: how to account for linear damping when predicting the range of a projectile ?

Tue, 19 Jul 2011 19:00:15 -0700

hi there,

i'm launching a rigid body into the air and i want to predict a spot fairly close to where it will land.

if i set the linear velocity damping to 0.0 then i can get a fairly accurate prediction. if i want some "air drag" though, and i set linear damping to 0.01 say, how is that applied in the simulation in each timestep? such that i could account for that into my calculation for the horizontal distance traveled of the projectile ?

TIA

Generated cloth normals blinking

Mon, 18 Jul 2011 11:20:39 -0700

Using PhysX SDK 2.8.4, I am running into an issue with generated normals related to welding verts and the generation of normals.

I have a mesh which is double-sided with backface culling on, created by duplicating the vertices and triangles of the mesh with reverse winding order. The reason I need the mesh to be double-sided with duplicated vertices is so that there can be different UVs on each side. When initially implemented, the cloth looked correct, but it would split in half when simulating, as the two sides of the cloth were not attached to each other.

I looked at the SDK docs, and it suggested using NX_CLOTH_MESH_WELD_VERTICES in the "flags" member of NxClothMeshDesc. Adding this flag did prevent the cloth from splitting in half, but caused the normals output from PhysX to randomly flip back and forth each frame between the front side and back side of the cloth, which makes lighting on the cloth blink like crazy.

How can I avoid both of these issues with double-sided cloth? Specifically, how can I make double-sided cloth not split in half while maintaining correct normals on both sides?

Thanks!

Adding objects in Visual Debugger

Sat, 16 Jul 2011 05:33:50 -0700

Hi!
I tried to add some debug objects to PhysX Visual Debugger with a help of the docs, but no success. Here is how the code looks like:

CODE

    NX_DBG_CREATE_OBJECT(objectID, NX_DBG_OBJECTTYPE_BOX, "DebugBox", DEBUG_OBJECTS_MASK);
    NX_DBG_CREATE_PARAMETER(transform, objectID, "Frame", DEBUG_OBJECTS_MASK);
    NX_DBG_CREATE_PARAMETER(NxVec3(1.2f, 0.2f, 0.1f), objectID, "Corner", DEBUG_OBJECTS_MASK);
    NX_DBG_ADD_CHILD(physicsScene, objectID, DEBUG_OBJECTS_MASK);

As a result i don't see any objects added neither in viewport of the scene browser. However i can add actors of my choise and they appear just fine.
What i forgot?

P.S. The events mask for a debugger and my objects mask are OK.

Blending the last pixels

Fri, 15 Jul 2011 05:50:45 -0700

When I export my textures to DDS files using Nvidias dds tool it blends the sides with each other. For example if I have one red side and one green side that is opposite to each other the last pixel on both side becomes the red and green color mixed. Any idea what I am doing wrong or how I turn this off ?

Why is compiled shader using more instructions?

Thu, 14 Jul 2011 23:31:21 -0700

Hello,

I have a shader that looks like this :

CODE

void main( in   float2              pos         : TEXCOORD0,
           in   uniform sampler2D   data        : TEXUNIT0,
           in   uniform sampler2D   palette     : TEXUNIT1,
           in   uniform float       c,
           in   uniform float       th0,
           in   uniform float       th1,
           in   uniform float       th2,
           in   uniform float4      BackGroundColor,
           out  float4              color       : COLOR
         )
{
    const float4 dataValue = tex2D( data, pos );
    const float vValue = dataValue.x;
    const float tValue = dataValue.y;

    color = BackGroundColor;
    if ( tValue <= th2 )
    {
        if ( tValue < th1 )
        {
            const float vRealValue = abs( vValue - 0.5 );
            if ( vRealValue > th0 )
            {
                // determine value and color
                const float power = ( c > 0.0 ) ? vValue : ( 1.0 - vValue );
                color = tex2D( palette, float2( power, 0.0 ) );
            }
        }
        else
        {
            color = float4( 0.0, tValue, 0.0, 1.0 );
        }
    }
}

If I compile it using cg version 2.2.0006, like this :

CODE

cgc -profile arbfp1 -strict -O3 -q sh.cg -o sh.asm

I get this a compiled shader that uses 18 instructions:

CODE

# 18 instructions, 3 R-regs

Now, if I compile the shader using cg version 3.0.0016 (doesn't matter what optimization option I put), I always get a compiled shader that uses 23 instructions:

CODE

# 23 instructions, 3 R-regs

Can someone explain what is going on? Why is the optimization not working and why is the shader longer when I compiled with cg 3.0?

Mouse picking, object dragging

Thu, 14 Jul 2011 14:35:59 -0700

Hi guys, I found a couple tutorials on the net about how to implement object dragging, but it seems that my code only goes as far to add a spring to an object, but not have the constraint attach to the mouse. I'm using OIS for mouse input, does anyone know how this can be accomplished? I have this code for picking so far -

CODE

bool Nogredex::pick(Ray &r, Ogre::Vector3 &intersectPoint) {
    NxRay nxRay = toNx®;
    NxRaycastHit hit;

    // intersect ray + scene
    NxShape* closestShape = p->mScene->raycastClosestShape(nxRay, NX_DYNAMIC_SHAPES, hit);
	if (!closestShape) return false;
	if (!closestShape->getActor().isDynamic()) return false;
    if (hit.shape) {
        intersectPoint = toOgre(hit.worldImpact);
        p->mPickedActor = &hit.shape->getActor();

		//added
		
		NxActor* gMouseSphere = CreateSphere(hit.worldImpact, 0.1f, 1.0f);
		gMouseSphere->raiseBodyFlag(NX_BF_KINEMATIC);
		gMouseSphere->raiseActorFlag(NX_AF_DISABLE_COLLISION);
		NxDistanceJointDesc desc;
		NxActor* gSelectedActor = &closestShape->getActor();
		gSelectedActor->wakeUp();
		desc.actor[0] = gMouseSphere;

		#ifdef USE_NX_DOUBLE_BUFFERED
			desc.actor[1] = static_cast(gSelectedActor->userData);
		#else
			desc.actor[1] = gSelectedActor;
		#endif

		gMouseSphere->getGlobalPose().multiplyByInverseRT(hit.worldImpact, desc.localAnchor[0]);
		gSelectedActor->getGlobalPose().multiplyByInverseRT(hit.worldImpact, desc.localAnchor[1]);
		desc.maxDistance = 0.0f;
		desc.minDistance = 0.0f;
		desc.spring.damper = 1.0f;
		desc.spring.spring = 200.0f;
		desc.flags |= NX_DJF_MAX_DISTANCE_ENABLED | NX_DJF_SPRING_ENABLED;
		NxJoint* joint = p->mScene->createJoint(desc);
		gMouseJoint = (NxDistanceJoint*)joint->is(NX_JOINT_DISTANCE);
		//end added
		

        // get intersection point in actor's local coordinate frame
        NxMat33 rot;
        NxVec3 v;

        v = hit.worldImpact - p->mPickedActor->getGlobalPosition();
       rot = p->mPickedActor->getGlobalOrientation();
        rot.setTransposed();    // reverses the rotation
        rot.multiply(v, p->mLocalPoint);
		
        return true;
    }
    return false;
}

Raycast filtering Problem

Thu, 14 Jul 2011 11:45:25 -0700

I'm having some problems filtering my raycasts. I am using the "SampleRaycast" example that comes with the PhysX SDK, and I am trying to get the horizontal raycast to ignore the terrain. What I've done is set the shape group to the terrain to 1 inside the "InitTerrain()" function like so:

CODE

NxActorDesc ActorDesc;
ActorDesc.shapes.pushBack(&terrainShapeDesc);
ActorDesc.shapes[0]->group = 1;
gScene->createActor(ActorDesc)->userData = (void*)0;

And I've set my raycast to group 2 like so: (RaycastAllShapes())

CODE

gScene->raycastAllShapes(worldRay, gMyReport, NX_ALL_SHAPES, 2);

Now I've been able to filter the cubes to pass right through the terrain, but I am having problems with the raycast and the terrain. Any ideas?

Thanks.

Ragdolls and referenced geometry

Thu, 14 Jul 2011 02:08:03 -0700

Hi

This question goes hand-in-hand with my other ragdoll query. Can a Kinematic Ragdoll be created on referenced geometry?

My scene looks like this:

I have a RIG file containing the skeleton. I open this up and reference in the MESH file. The MESH file contains the character model and the cape mesh.
Then I bind the referenced geometry to the skeleton.

I cannot create a Kinematic ragdoll for geometry referenced in this way. Am I doing something wrong?

Barry

How to turn on (active) Stereo mode for OpenGL program

Thu, 14 Jul 2011 01:01:03 -0700

My graphic card is Quadro 4000, with acer hn274n monitor, and 3d vision glasses.
When "3D Vision Photo Viewer" switch to stereo mode, the screen flashes and then I can see stereo from the nVidia glasses. After I close the software, the screen flashes again and turn off the switch mode.
For my own OpenGL program, I can not see stereo when the program is running (screen didn't flash) along, but I can see stereo when "3D Vision Photo Viewer" has switched on the stereo mode (window mode). So when I close "3D Vision Photo Viewer", I cannot see stereo from my OpenGL program again.
I think there is an api to turn on/off the stereo mode at run time. Does anybody know about that?

having more then one light affect an object

Wed, 13 Jul 2011 17:20:55 -0700

It's been awhile since i posted here, but i've been re-finding FX Composer lately and have gotten the hang of some of it i still don't know all of it, such as the editor specific SEMANTICS like Ambient, and all the others. Also i'd like to know how to render to a Color Target (render target) and also I noticed only one light can affect an object at once. now is this FX Composer's limitation? or is there something i need to set to make multiple lights affect the object(s) at once? I look forward to hearing from anyone about these issues

Can't create Kinematic Ragdoll

Wed, 13 Jul 2011 07:26:20 -0700

Hi

We're just attempting to push a cape through a Maya - Unreal pipeline. I'm following the Maya tutorial videos for APEX Cloth and have hit an impasse. The tutorial in question is the trenchcoat one. When I select the body and try to create a Kinematic Ragdoll Maya reports:

"// Error: No joint selected/found. Ragdoll not generated"

I'm following the tutorial exactly but something's not right.

I've made the cape and bound it to the character's rig. I've painted maxDistance and pinToNearest values. The cape seems to simulate fine - draping in all the right places. Next I select the character's body mesh and from the menu: PhysX - Ragdolls - Create Kinematic Ragdoll. This is when Maya starts to complain.

Anyone got any tips

Barry

(Maya 2010 - by the way)

please add ppu support

Tue, 12 Jul 2011 15:29:56 -0700

please add old ageia ppu support to new NV physx drivers....i have a lot of friends that own old ageia ppu, please support them... thanks a lot, with big honor Neil Aaron Brown aka 'Tonix The Game'

ESA Specifications

Tue, 12 Jul 2011 10:52:15 -0700

Hello,

I am a programmer with a Dell XPS 730x and I would like to write a custom program for my master control board based on ESA specifications, and I need to know more about ESA specifications in general along with API documentation. It would also be helpful to know exact Memory handles (locations) the MCB has to be ESA compliant. If anyone knows this information, or at the very least can point me to someone or a research that has the information that would be great.

Thanks!

physxcooking.dll was not found

Tue, 12 Jul 2011 09:55:11 -0700

Hi to all,

Just bought this off the site (GamersGate or something like that, listed on the homepage), but when I installed it and tried to run it it says, "This application has failed to start because physxcooking.dll was not found. Re-installing the application may fix this problem." I've reinstalled the game like 6-7 times and it's the same problem. What's wrong with my game?

crash when construct a convex mesh

Tue, 12 Jul 2011 00:21:01 -0700

My program works when I generate a convex hull using a mesh file. But it crashes when I try to create a convex mesh using this same mesh file.
The triangle number is less than 100 in this mesh file.

My code is:

convexDesc->numVertices = objMesh->getNumVertices();
convexDesc->pointStrideBytes = sizeof(NxVec3);
NxVec3 DestVert[256];
for(NxU32 Index=0; Index< objMesh->getNumVertices(); Index++)
{
DestVert[Index]=objMesh->getVertex(Index);
}
convexDesc->points = DestVert;
convexDesc->flags = 0;//NX_CF_COMPUTE_CONVEX;
convexDesc->numTriangles = objMesh->getNumTriangles();
convexDesc->triangles = TempIndices;
convexDesc->triangleStrideBytes = 3 * sizeof(NxU32);

NxConvexShapeDesc convexShapeDesc;
convexShapeDesc.localPose.t = NxVec3(0,0,0);
convexShapeDesc.userData = convexDesc;

NxInitCooking();
MemoryWriteBuffer buf;
printf("start cooking convex!\n");
bool status = NxCookConvexMesh(*convexDesc, buf);

The error happens when I call NxCookConvesMesh. The error message says that "Access violation reading location 0x00000000"
Can any expert help me?

3dVision Pro and Coin 3d (Open Inventor)

Mon, 11 Jul 2011 23:59:24 -0700

Hello,
I'm developing an application based on OpenGL. I'm using Coin3D libraries and Visual Studio.
I set in my application the quad-buffer stereo visualization, but I cannot enable Nvidia glasses.
Depending on the setting in the Nvidia control panel I obtain the left and right images superimposed or a single image.

Should I change something in the 3d settings or should I enable glasses from the c++ code?
I have also downloaded and tried to use Nvapi but no success.

Any help is appreciated. I do not know where the problem is.

Thanks in advance,
Manuela Chessa

physxcooking.dll error

Mon, 11 Jul 2011 10:16:20 -0700

Hi to all,

I have the problem that is when I start to install physxcooking.dll "Error initializing physxcooking.dll please go to the Nvidia website for updated drivers and even when I install physxcooking.dll same message s' displays. Also be aware that I run under vista 64bit with a Radeon HD 4850 1GB which is not normally compatible with physxcooking.dll.At the start the game but I walked format my hard disk and reinstalling impossible to play. I do not know if that has significance, but I buy my game DLgamer. Please help me in this case and let me why the error is been given and what are the ways to solve the same.

NvAPI: How can I detach (disable) a monitor?

Mon, 11 Jul 2011 09:07:48 -0700

I'm using a nVidia GTX 480 and it has three outputs: 2x DVI, 1x Mini-HDMI

However, only two of them can be enabled at a time. I'm looking into the NvAPI and I've found a function to do the opposite (ie. enable a detached/disabled monitor) but I can't find a function to disable a monitor.

Can anybody give me some advice?

NVIDIA driver profile association

Mon, 11 Jul 2011 08:31:09 -0700

Is it possible to create a NVIDIA driver profile and associate this profile with a specific piece of software; I am working for a small medical imaging company and we wish to create a NVIDIA driver profile (using NVAPI) and associate this with our own software as part of the software installation processs; However I havent found any documentation on how I can associate a NVIDIA driver profile with our own software.

Queried raycasting parameters

Mon, 11 Jul 2011 02:22:28 -0700

Hi!
PhysX has an ability to perform batched queries. I need to make 200+ raycasts every tick. So i thought using queries for this is a good idea. Also i need to check for only the closest hit shape. Hence raycastClosestShape().
But this function has some parameters that i don't understand and couldn't find info in docs. These are "NxRaycastHit& hit" structure and "NxShape** cache" pointer. Since queries are executed later on, i don't understand why is the need in "NxRaycastHit& hit". Should i read something from it? Put something in it? What is it used for? The same with "NxShape** cache". What does it do? What is it used for?

P.S. This is used for checking projectile hits.
P.S.S. On a side note, are raycasting good for this? Or better use overlap tests (like sphere, capsule, etc.)? I don't think i will need position and normal info from contact. Just the shape it hit.

cgfx samplers + d3d11

Sun, 10 Jul 2011 02:24:27 -0700

Hello,

In my cgfx shader I have a sampler parameter:

uniform sampler2D texAlbedo;

In my shader load code, I loop over all effect parameters starting at cgGetFirstEffectParameter

For any parameters where cgGetParameterType(param) == CG_SAMPLER2D I cache the CGparameter handle

Now, when I am ready to render, I'd like to call PSSetShaderResources() to bind the ID3D11ShaderResourceView* that I have for a texture.

How do I get the index of the shader resource?

Calling cgGetParameterResourceIndex on the parameter returns 0xFFFFFFFF, but the correct value for this simple shader is just 0... cgGetError() returns CG_NO_ERROR after the call to cgGetParameterResourceIndex

A little hand would be appreciated, thanks!

Issue with getting vertex colors from application

Sat, 09 Jul 2011 14:50:36 -0700

Problem: I would like to work with vertex colors delivered by Maya in my cgfx shader. It should be a trivial problem but I have no luck so far.

Attempt: Below I have written up a super simple shader which should just display the raw vertex colors. Result is pure white though.

Details: Internally the shader works fine, if I set the vert color to red in the vs it comes through just fine. So the issue is getting the value from Maya. Also tried with both COLOR and COLOR0.

Any help or guidance appreciated.

CODE

// string Category = "Effects\\Cg\\BRDF";
// string keywords = "texture";
string description = "Pure vertex color";

//////////////////////////////////////////////////////////
// untweakables //////////////////////////////////////////
//////////////////////////////////////////////////////////

float4x4 WorldViewProjXf : WorldViewProjection < string UIWidget="none";>;

/****************************************************/
/********** SAMPLERS ********************************/
/****************************************************/

texture ColorTex : Diffuse
<
    string ResourceName = "default_color.dds";
    string ResourceType = "2D";
>;

sampler2D ColorSampler = sampler_state
{
    Texture = <ColorTex>;
      MagFilter = Linear;
      MinFilter = LinearMipmapLinear;
};

/****************************************************/
/********** CG SHADER FUNCTIONS *********************/
/****************************************************/

/**************************************/
/***** SHARED STRUCT ******************/
/**** Data from app vertex buffer *****/
/****     for all passes          *****/
/**************************************/

struct appData {
    float3 Position    : POSITION;
    float4 VertColor: COLOR;
};

/****************************************/
/****************************************/

// vertex->fragment registers used for this pass only
struct outVertexData {
    float4 HPosition    : POSITION;
    float4 vertColor     : TEXCOORD1;
};

/****************************************/
/****************************************/

outVertexData textureVS(appData IN)
{
    outVertexData OUT;

    OUT.vertColor = IN.VertColor;
    //OUT.vertColor = float4(1.0f,0.0f,0.0f,0.0f);

    float4 Po = float4(IN.Position.xyz,1.0);
    OUT.HPosition = mul(WorldViewProjXf, Po);

    return OUT;
}

float4 texturePS(outVertexData IN)    : COLOR
{
    return IN.vertColor;
}

/****************************************************/
/********** TECHNIQUES ******************************/
/****************************************************/

technique main {
    pass p0 {
        VertexProgram = compile arbvp1 textureVS();
        DepthTestEnable = true;
        DepthMask = true;
        CullFaceEnable = false;
        FragmentProgram = compile arbfp1 texturePS();
    }
}

/***************************** eof ***/

Simple FPS monitoring program?

Sat, 09 Jul 2011 13:54:50 -0700

In short, I need to build a small application that can monitor the current FPS loadout of the GPU. I know there is software that does this already, but for the purposes I am using it for, they are not an option, therefore I need to build my own. First impressions led me to believe this task was possible with PerfSDK, but now after looking into it more it seems PerfSDK is designed more for monitoring performance of your own application, not monitoring general GPU load to determine FPS of the current application.

Am I looking at the wrong software for what I need? If so can anyone point me in the right direction? If PerfSDK is capable of what I want, is there anywhere I can find better documentation for what I am attempting to do?

Thank you.

Errors compiling CgFX effect using Direct3D9 .FX States

Thu, 07 Jul 2011 13:16:51 -0700

Hello,

I have recently begun using Cg and CgFX and have read through most of the Cg Tutorial book available in the Developer Zone. I am specifically interested in migrating an existing rendering effect system from Direct3D's .FX effect file format to Cg's .CgFX file format because according to the Cg Tutorial, .CgFX is "identical" to the .FX format, and has the added advantage of being compatible with both Direct3D and OpenGL. This is exactly what I need, as it will save me the effort of re-writing all of my effect scripts. Unfortunately, my effect scripts are not compiling under Cg, and I am beginning to doubt that .CgFX and .FX are identical. Consulting the documentation, there seem to be places where the DirectX convention is used, such as with the "CullMode" state, which accepts the DirectX enumerants "None", "CW", "CCW", versus the "DiffuseMaterialSource" state that only appears to accept the enumerants "true" and "false". (Whatever that means.)

I am looking for a definitive answer: Should I expect to be able to compile under Cg any .FX script that compiles successfully under Direct3D 9? If so, what do I have to do to make this possible?

Here is an example to illustrate the problem.

My .CgFX effect, the contents of which compile successfuly as a DirectX .FX effect file under Direct3D 9:

CODE

technique Default
{
	pass P0
	{
		Lighting = False;
		CullMode = None;

		AmbientMaterialSource = Color1;
		DiffuseMaterialSource = Color1;
		SpecularMaterialSource = Material;
		EmissiveMaterialSource = Material;
		
		LightEnable[0] = True;
		LightType[0] = Directional;
		LightDirection[0] = float3(0.0f, 1.0f, 0.0f);
		LightAmbient[0] = float4(0.1f, 0.1f, 0.1f, 1.0f);
		LightDiffuse[0] = float4(1.0f, 0.0f, 0.0f, 1.0f);
		LightSpecular[0] = float4(1.0f, 1.0f, 1.0f, 1.0f);

		MagFilter[0] = Linear;
		MinFilter[0] = Linear;
		MipFilter[0] = Linear;
		
		AddressU[0] = Clamp;
		AddressV[0] = Clamp;

		VertexShader = null;
		PixelShader = null;
	}
}

My set up code, which initializes Cg and loads the effect:

CODE

CGcontext cgContext = cgCreateContext();
cgD3D9RegisterStates(cgContext);
CGeffect cgEffect = cgCreateEffect(cgContext, "Data/Effects/Default.cgfx", nullptr);

The errors that are reported when trying to compile the effect in the code above:

CODE

error C1056: invalid initialization
error C1008: undefined variable "Color1"
error C1008: undefined variable "Color1"
error C1008: undefined variable "Material"
error C1008: undefined variable "Material"
error C1008: undefined variable "Directional"
error C1056: invalid initialization
error C1008: undefined variable "Linear"
error C1008: undefined variable "Linear"

I would greatly appreciate any help with this matter.

game engine with physx

Thu, 07 Jul 2011 07:40:44 -0700

hi all

could anyone recommend me game engine sdk ... works with physx ?

thx

Does NVIDIA have a (windows driver) symbol server?

Wed, 06 Jul 2011 04:50:02 -0700

Well the topic says it all. It would be really useful if I had symbols for nvwgf2um.dll to figure out why exactly it seems to be eating cpu cycles(ie which state changes to try and reduce etc:-)

Thanks,

David

Problem of precision with moveGlobalPose and moveGlobalOrientationQuat

Wed, 06 Jul 2011 01:18:56 -0700

Hi,

I want to rotate a kinematic actor with moveGlobalOrientationQuat or moveGlobalPose and it works but I have a problem of precision.
My rotation needs to be very slow but angle seems to be "grab" to a grid of 0.2�. So my rotation seems to lag despite the frame rate is OK.
The problem doesn't appear with SetGlobalOrientation but I can't use it.
Any ideas ?
Is there a way to define this precision or a work around to do slow rotation with moveGlobalOrientationQuat ?

Thanks,
Harold

How to export some data

Tue, 05 Jul 2011 19:56:09 -0700

such as render states?
like PerfHUD 4. It can export the render target to an image file and render states to an text file.

VBO memory requirements

Tue, 05 Jul 2011 04:46:24 -0700

Hi All,

Firstly I was not sure where to put this question as its very much specific to OpenGL using nVidia cards, hence chose this post.
I am writing a Windows based 3D viewer for CAD file formats, for this I have used Call List feature to render scenes.
Now after good amount of research I found out that I should use VBO if I want to be futuristic about my App.
After doing a sample implementation I got some results about VBO implementation and which are horrifying from memory perspective.
I tested two models one with just 18k vertices arranged in triangle strip and the other model is with 10 million vertices arranged in triangle strip.
I tested the above models on two different nVidia cards namely Quadro FX 1800 and Quadro FX 1600M with latest drivers installed.
Also for 18k model i am creating about 10 VBOs and for the bigger model i am creating ~1000 VBOs.
What I found out was on Quadro FX 1800 the memory (Working Set and peak) for my application process is way higher (40%) than my call list implementation and for 1600M card its more or less same.

When I tried to comment glDrawArrays call in my VBO implementation memory does not shoot, (obviously no scene as well).
Can anyone help me to dig into this issue and check why is this happening.
I am properly creating VBOs.
Also if too much VBO numbers is the problem then the behavior is same for 18k vertices model as well.

-Nikhil

Car Problem

Tue, 05 Jul 2011 01:26:36 -0700

I have a problem with vehicle rotation.
This is my first PhysX try -
I have created a new project and loaded a terrain from file [using some other function and not any of PhysX SDK], made a Triangle mesh and added that to Scene then I have used some fuctions of sampleRayCastCar sample to get a simple cart [without that hamburger].
Now everything simulates with gDeltaTime and each time function keys are checked for RenderCallBack [We do not render anything, as we are not using Glut, instead DirectX, so we are rendering it on Remote Debugger only]

Car moves but stops while turning, so there might be something related with terrain [friction or something] or is there anything I missed while creating cart?

Update global variable

Mon, 04 Jul 2011 14:44:58 -0700

Hello,

I am new to FX composer so maybe I missed something in the documentation.

In CGFX, it seems global variables initial value cannot be changed.

Example:

float4 color1 = float4(1.0, 0.0, 0.0, 1.0);
float4x4 WorldViewProj : WorldViewProjection;

float4 mainVS(float3 pos : POSITION) : POSITION{
return mul(WorldViewProj, float4(pos.xyz, 1.0));
}

float4 mainPS() : COLOR {
return color1;
}

technique technique0 {
pass p0 {
CullFaceEnable = false;
VertexProgram = compile vp40 mainVS();
FragmentProgram = compile fp40 mainPS();
}
}

If I add a sphere in the render view it shows it as a red disk. Correct.
Now if I change color1 to float4(0.0, 0.0, 1.0, 1.0) and re-compile, it should display a blue disk. But it remains red even after restarting the Fx Composer. However, when I define the variable inside the shader function, the right value is used.

???

Thanks for your help

Georges

32-bit NVTT on MacOSX

Mon, 04 Jul 2011 07:35:05 -0700

I'm having serious issues getting nvtt to compile for x86 on macos. Setting the NV_SYSTEM_PROCESSOR cmake var to i686 later gives me a compiler error "nvtt/src/nvcore/Debug.cpp:1: error: CPU you selected does not support x86-64 instruction set". Seeing that the source does not cotain the text for that error message this leads me to believe that message is compiler-generated and due to x86-64 inline assembly beeing used even tough it is configured for x86.

Does not NVTT support MacOSX on 32bit platforms?

NX_NOTIFY_ON_TOUCH not reliable?

Sat, 02 Jul 2011 20:55:32 -0700

Hi
I current have a dynamic body that I need to set to "grounded" when touching anything with a contact normal.y > 0, so Im checking for the ON_TOUCH event in the contact report callback, then I check the normal, and set it to grounded(trough userdata) if normal.y >0, otherwise I do NOTHING..

So, outside this code, before advancing the simulation, I always clear the flag..so it will only be grounded if the ON_TOUCH is reported...

The thing is, sometimes it just doesnt happen, its skipped for some reason, while its clearly sliding on the top of a shape. It happens when the velocity is very slow, BUT not always, I also verified if it was going to sleep, and it isnt...its really the ON_TOUCH that is not happening..

I do something like this:
...
grounded = 0;

advance simulation

if( actor->isGroupSleeping() ) grounded = 1;

if( grounded)
do stuff
else
do other stuff
...

So, even if my dynamic cube is on ground, sliding slowly, I get a bunch(not really a bunch, it flicks 2 to 3 times ) of "not grounded" in a row, before it comes to a more stable state..

Note also that Im verifying on start and on end touch, and none of those is reported when the on touch is also not reported, with doesnt make any sense..
Does the SDK makes any prediction/optimization stuff that skips the callback for on touch events at some frames? Am I doing something wrong?

asset export sample "things go away :("

Sat, 02 Jul 2011 18:25:05 -0700

Hi

I just have a question ...why when i run asset export sample ... if I pressed to create stack or cube it just go away ...in the right direction

how can I handle that issue ...

plz little help here ....it's an urgent

thx

DX11 - Out of memory b/c of MSAA texture Release() issue?

Sat, 02 Jul 2011 14:18:26 -0700

Hi, I am having DX out of memory issues in my application, and tracked it down to an issue with releasing MSAA textures (DX11). As I will describe below this seems to be a driver issue, so I'm not sure in which forum to put it, so I decided to put it here in the NVIDIA Graphics SDK forum.

The code snipped below (which just creates and then releases a texture) causes the issue when enabled with msaa:

CODE

  D3D11_TEXTURE2D_DESC Desc;
  ZeroMemory( &Desc, sizeof( D3D11_TEXTURE2D_DESC ) );
  Desc.ArraySize = 1;
  Desc.BindFlags = D3D11_BIND_RENDER_TARGET | D3D11_BIND_SHADER_RESOURCE;
  Desc.Usage = D3D11_USAGE_DEFAULT;
  Desc.Format = DXGI_FORMAT_R8G8B8A8_UNORM;
  Desc.Width = width;
  Desc.Height = height;
  Desc.MipLevels = 1;
  Desc.CPUAccessFlags = 0;

  Desc.SampleDesc.Count = msaa;
  Desc.SampleDesc.Quality = msaaquality-1;

  window->g_pd3dDevice->CreateTexture2D( &Desc, NULL, &window->g_pTexRender10 );
  if(window->g_pTexRender10)
  {
    window->g_pTexRender10->Release();
  }
  window->g_pTexRender10 = NULL;

When I set 'msaa' to at least 2 and 'msaaquality-1' to any value between 0 and msaa-1, Release() does not seem to release everything, as I get an out of memory error after several hundred frames (executing the code once every frame). Changing the BindFlags makes no difference.
When this happens, Windows also tells me that my memory is low, while my physical memory in task manager is still way below 100% (around 50%, or maybe 75%, not increasing before or during the out of memory errors).

MSAA values have been checked using CheckMultisampleQualityLevels, and the textures work fine when rendered to. HResults from CreateTexture and Release are all checked and all return S_OK (until the create call fails with the out of memory error).

I am using:

- The latest NVIDIA drivers, 275.33
- GeForce GTX 470
- Windows 7 64bit
- DirectX11

The log before the app breaks (I enabled DX debug mode and break on error / corruption in the dx control panel, muting only info messages):

First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cc354..
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cc6e8..
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cc82c..
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cd9a4..
D3D11: ERROR: ID3D11Device::CreateTexture2D: Returning E_OUTOFMEMORY, meaning memory was exhausted. [ STATE_CREATION ERROR #105: CREATETEXTURE2D_OUTOFMEMORY_RETURN ]
D3D11: **BREAK** enabled for the previous D3D11 message, which was: [ STATE_CREATION ERROR #105: CREATETEXTURE2D_OUTOFMEMORY_RETURN ]
First-chance exception at 0x7550b727 in TextureFrank.exe: 0x0000087D: 0x87d.
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cc354..
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cc6e8..
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cc82c..
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cd9a4..
D3D11: ERROR: ID3D11Device::CreateTexture2D: Returning E_OUTOFMEMORY, meaning memory was exhausted. [ STATE_CREATION ERROR #105: CREATETEXTURE2D_OUTOFMEMORY_RETURN ]
D3D11: **BREAK** enabled for the previous D3D11 message, which was: [ STATE_CREATION ERROR #105: CREATETEXTURE2D_OUTOFMEMORY_RETURN ]
First-chance exception at 0x7550b727 in TextureFrank.exe: 0x0000087D: 0x87d.
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cc354..
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cc6e8..
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cc82c..
First-chance exception at 0x7550b727 in TextureFrank.exe: Microsoft C++ exception: _com_error at memory location 0x003cd9a4..
D3D11: ERROR: ID3D11Device::CreateTexture2D: Returning E_OUTOFMEMORY, meaning memory was exhausted. [ STATE_CREATION ERROR #105: CREATETEXTURE2D_OUTOFMEMORY_RETURN ]
D3D11: **BREAK** enabled for the previous D3D11 message, which was: [ STATE_CREATION ERROR #105: CREATETEXTURE2D_OUTOFMEMORY_RETURN ]
First-chance exception at 0x7550b727 in TextureFrank.exe: 0x0000087D: 0x87d.
The program '[2352] TextureFrank.exe: Native' has exited with code 0 (0x0).

Seeing how just using MSAA can cause this crash, and I'm not seeing any errors on my part (please correct me if i'm doing something silly :]), this seems to be a driver issue. If anyone could help me out here that would be great .

Thanks in advance!

Cheers, Frank de Bresser

about asset export sample

Fri, 01 Jul 2011 18:40:49 -0700

hi all

I wonder .. where i can edit the position of the loaded object ?
in which line ?

thx

Nvidia updates

Thu, 30 Jun 2011 13:32:46 -0700

Hello there

Am trying to install the latest version for my card but evertime I try to install it come up as Vista when I have windows 7 OS system installed

275.33-desktop-win7-winvista-64bit-international-whql.exe

can ypu send me an link for windows 7 64 bit

thank you

[ForceField] Linear Kernel formula needed

Thu, 30 Jun 2011 10:46:32 -0700

Hi,
I need the formula for the linear kernel, that is used internally!
I found the normally one in the documentation, but I need the formula with the TYPE = NX_FF_TYPE_GRAVITATIONAL.
I hope a (former) employee can help me.

I need this formula to explain something.

shindai

vs just in time in debugger

Thu, 30 Jun 2011 02:23:12 -0700

when i open the cuda sdk browser4.0.....
iam getting "visual studio just-in-time debugger"
An unhandled win 32 exception occured in browser.exe
possible debugger:
new instance of vs2008

help me to correct this....

file missing

Thu, 30 Jun 2011 02:15:06 -0700

when i compile and build the bandwidth sample program using cuda4.0 in xp win32

i got ended up with fatel error: cannot open input file cutil32D.lib

how can i include that file...

sample file cannot be loaded

Thu, 30 Jun 2011 02:08:38 -0700

The following error has occurred during XML parsing:

File: C:\Documents and Settings\All Users\Application Data\NVIDIA Corporation\NVIDIA GPU Computing SDK 4.0\C\src\deviceQuery\deviceQuery_vs2008.vcproj
Line: 22
Column: 4
Error Message:
Custom build rules file 'c:\Program Files\Microsoft Visual Studio 9.0\VC\VCProjectDefaults\NvCudaRuntimeApi.rules' was not found or failed to load.
The file 'C:\Documents and Settings\All Users\Application Data\NVIDIA Corporation\NVIDIA GPU Computing SDK 4.0\C\src\deviceQuery\deviceQuery_vs2008.vcproj' has failed to load.

ihave installed cuda toolkit and sdk versions of 4.0 and using vs2008 in win32 xp..
when i try to view the device query sample code iam getting the above error.

could any one help in solving me this issue....

missing file

Wed, 29 Jun 2011 19:42:25 -0700

hello

does any one know where i can find file called

PlatformDEBUG-MT.lib

plz it's urgent

thx

clash between webgl, Win7 and Nvidia Geforce

Wed, 29 Jun 2011 12:48:39 -0700

I have done a sphere using J3DI (which is a webGL library) under WIN 7 64bit as OS and Nvidia Geforce GT330 M as graphic card .

firstly I have done it in blue color and it appeared correctly.

then, I tried to make a texture on it but the sphere appeared like in this image: http://s1.postimage.org/1ekqrgolg/earth.jpg

while it has appeared in Mac like this : http://s1.postimage.org/1eksf0138/error.jpg

so, what is the problem? is it from the OS, J3DI or from the graphic card?

for an additional information, the shader script i used this:

VertexShader:

uniform mat4 u_modelViewProjMatrix;
uniform mat4 u_normalMatrix;
uniform vec3 lightDir;

attribute vec3 vNormal;
attribute vec4 vTexCoord;
attribute vec4 vPosition;

varying float v_Dot;
varying vec2 v_texCoord;

void main()
{
gl_Position = u_modelViewProjMatrix * vPosition;
v_texCoord = vTexCoord.st;
vec4 transNormal = u_normalMatrix * vec4(vNormal, 1);
v_Dot = max(dot(normalize(transNormal.xyz), normalize(lightDir)), 0.3);
}
PixelShader:

#ifdef GL_ES
precision highp float;
#endif
uniform sampler2D sampler2d;

varying float v_Dot;
varying vec2 v_texCoord;

void main()
{
vec2 texCoord = vec2(v_texCoord.s, v_texCoord.t);
vec4 color = texture2D(sampler2d, texCoord);
color += vec4(0.1, 0.1, 0.1, 1);
gl_FragColor = vec4(color.xyz * v_Dot, color.a);
}

I'm really concern about this issu

Make fluids/soft acting like rigid

Wed, 29 Jun 2011 05:59:32 -0700

I was wondering if we put for instance a very high viscosity on fluids or very high stiffnesses on soft bodies we could imitate a rigid body behavior ?

Did someone already tested it?

shared object linkage error : Linux 64bit + PhysX3.0.1

Wed, 29 Jun 2011 03:50:10 -0700

We build a dso plug-in for our production system and we encountered such errors

with libPhysX3DEBUG.a

CODE

/usr/bin/ld: /opt/PhysX-3.0.1_LINUX_SDK_Core/PhysXSDK/SDKs/lib/linux64/libPhysX3DEBUG.a(NpPhysics.cpp.o): relocation R_X86_64_32S against `vtable for NpPhysics' can not be used when making a shared object; recompile with -fPIC
/opt/PhysX-3.0.1_LINUX_SDK_Core/PhysXSDK/SDKs/lib/linux64/libPhysX3DEBUG.a: could not read symbols: Bad value

with libPhysX3.so or libPhysX3CHECKED.a

CODE

/usr/bin/ld: /opt/PhysX-3.0.1_LINUX_SDK_Core/PhysXSDK/SDKs/lib/linux64/libPhysX3.a(NpPhysics.cpp.o): relocation R_X86_64_32S against `.rodata' can not be used when making a shared object; recompile with -fPIC
/home/egorch/app/PhysX-3.0.1_LINUX_SDK_Core/PhysXSDK/SDKs/lib/linux64/libPhysX3.a: could not read symbols: Bad value

compiler string contains -fPIC flag, we cannot to wipe it.

what we have to do to solve this issue?

Thank you!

alloca.h - No such file

Wed, 29 Jun 2011 03:40:41 -0700

Dear all,

since 2 weeks i try to understand how i am able to develop physx-programs. But if i want to compile an example project, i get the error, that the file alloca.h wasn�t found. I found some information, which told me, that the use of this file is deprecated, is it right?

I use eclipse with MinGW and the SDK in the version 2.8.4 on a windows (64 bit). Did anyone have the same problems?
Thanks a lot.

Greetings
Sascha

a dynamic actor inside a kinematic actor question

Wed, 29 Jun 2011 03:09:17 -0700

hi

we represent a charactor with 2 different role type of actors

one is NxCapsuleController that is reponsible for moving and the others is Dynamic actor that is responsible for parts and collision detection

and enemy charactor attacks a user character with a gunshort that use raycast to collision detection on character's parts actor

and sometimes raycast detect a collision for these charactor occured simulation crashing

but i don't know what the problem is.

after we delete the parts collision type of actor, just one NxCapsuleController for a character, then it's not never ocurred crashing

can someone explain me what the problem is?

thanks.

Precision : TriangleMesh vs HeightField

Tue, 28 Jun 2011 02:42:57 -0700

Background:
I am working on a project which uses NxTriangleMesh as physics object of terrian.
The raycasting and controller sweeping efficiency is acceptable but cooking time is quite long.

In order to kill cooking time we used NxU code to serialize the cooked meshes of terrian.
But a new problem here is the serialized cooked meshes are too large on hard disk.

For example a 128x128 tile meshes would occupy 3MB for serialized cooked meshes. For a 512x512 scene there must be 48MB.
And then 1440MB for a MMO with 30 so large scenes!

NxHeightField could save some space so I have had a test.
10000 randomly generated raycasts on the same terrain with both NxTriangleMesh and NxHeightField:

1. 4.3% results are totally different, one returned true and the other not.
2. There are 29% results that have different heights with an average error more than 8 unit.
3. 3% height results have an error lower than 0.5 unit.

The result is unacceptable.
Please could someone tell if this data is regular?

Uniform buffer behaving inconsistently across profiles

Mon, 27 Jun 2011 16:01:34 -0700

Hi all!

I'm trying to set up a system using uniform buffers to pass in per frame parameters at once, instead of having to pass them recursively.
I can't seem to use cgSetEffectParameterBuffer to work on profiles such as glslv.

I can reproduce this by forcing the cgfx_buffer_lighting to use the glslv and glslf profiles. Is the buffer API only supported on the latest profiles? If this is the case, should the 'cgfx_buffer_lighting_glsl' technique be removed from the cgfx_buffered_lighting example?

Thanks,
Joe.

PhysXLoader64.dll included with the PhysX Plug-In for 3ds Max 2011 appears to be "bad"

Mon, 27 Jun 2011 12:08:59 -0700

I've installed the PhysX Plug-In for 3ds Max 2011 using this installer:

NVIDIA_PhysX_Plug-in_3dsMax2011_x64_WithAPEX_2.60.0413.1900

obtained here:

http://supportcenteronline.com/ics/support...asp?deptID=1949

And there appears to be an issue with the PhysXLoader64.dll this installer unpacks. First off, when looking at the Properties, under Details tab, many of the fields, such as "Product version" and "Product name" are missing. Also, when I view this DLL with Dependency Walker (http://www.dependencywalker.com/) it reports several errors. I became suspicious of this DLL when I noticed that 3ds Max was loading an older version of the DLL from "Program Files\NVIDIA Corporation\PhysX" while the rest of the PhysX DLL's were being loaded from "3dsmax2011\stdplugs\(PhysX)\PhysX_2.8.4"

I've attached some screenshots of these issues. Is there something legitimately wrong with this DLL, and if so can we get a fixed version along with an updated installer?

Thanks!

Robert Thomas
Blizzard Entertainment

MemoryBuffer for Cooking on Linux

Mon, 27 Jun 2011 04:40:49 -0700

Hello everybody.

I try to cook a TriangleMesh with the example. But I have a problem with the class MemoryBuffer :
In the file MemoryPool.h (PhysXSDK/RepX/src) included by MemoryPoolStreams.h (PhysXSDK/RepX/src), there is this line

CODE

#include "../src/PxProfileFoundationWrapper.h"

But there is no file PxProfileFoundationWrapper.h in the Linux SDK 3.0.1.

I need this file for solve this errors :

CODE

/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:39: error: �profile� is not a namespace-name
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:39: error: expected namespace-name before �;� token
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:61: error: ISO C++ forbids declaration of �ProfileArray� with no type
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:61: error: expected �;� before �<� token
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:63: error: ISO C++ forbids declaration of �FoundationWrapper� with no type
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:63: error: expected �;� before �&� token
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:64: error: �TPxU8PtrList� does not name a type
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:67: error: expected �)� before �&� token
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h: In destructor �physx::repx::CMemoryPool::~CMemoryPool()�:
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:74: error: �TPxU8PtrList� has not been declared
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:74: error: expected �;� before �theEnd�
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:75: error: �TPxU8PtrList� has not been declared
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:75: error: expected �;� before �theIter�
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:76: error: �theIter� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:76: error: �theEnd� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:80: error: �mWrapper� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:82: error: �mAllMemory� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h: In member function �physx::pubfnd3::PxU8* physx::repx::CMemoryPool::allocate()�:
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:97: error: �mWrapper� was not declared in this scope
In file included from /usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPoolStreams.h:31,
                 from WPGeometry.hpp:15,
                 from WPGeometry.cpp:9:
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:102: error: �mAllMemory� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h: At global scope:
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:132: error: ISO C++ forbids declaration of �ProfileArray� with no type
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:132: error: expected �;� before �<� token
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:133: error: ISO C++ forbids declaration of �FoundationWrapper� with no type
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:133: error: expected �;� before �&� token
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:134: error: �TPxU8PtrList� does not name a type
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:139: error: expected �)� before �&� token
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h: In destructor �physx::repx::CVariableMemoryPool::~CVariableMemoryPool()�:
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:148: error: �TPxU8PtrList� has not been declared
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:148: error: expected �;� before �theEnd�
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:149: error: �TPxU8PtrList� has not been declared
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:149: error: expected �;� before �theIter�
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:150: error: �theIter� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:150: error: �theEnd� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:154: error: �mWrapper� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:156: error: �mAllMemory� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h: In member function �physx::pubfnd3::PxU8* physx::repx::CVariableMemoryPool::allocate(physx::pubfnd3::PxU32)�:
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:216: error: �mWrapper� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:217: error: �mAllMemory� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:224: error: �mWrapper� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:226: error: �mAllMemory� was not declared in this scope
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h: At global scope:
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:295: error: �FoundationWrapper� does not name a type
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:322: error: ISO C++ forbids declaration of �FoundationWrapper� with no type
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:322: error: expected �;� before �&� token
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:323: error: expected �;� before �inline�
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h: In constructor �physx::repx::CMemoryPoolManager::CMemoryPoolManager(physx::pubfnd3::PxAllocator
Callback&)�:
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:309: error: class �physx::repx::CMemoryPoolManager� does not have any field named �mWrapper�
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPool.h:310: error: �mWrapper� was not declared in this scope
WPGeometry.cpp: In constructor �Wrapper::WPTriangleMesh::WPTriangleMesh(Wrapper::WActor*, Wrapper::WVector3*, int, int*, int)�:
WPGeometry.cpp:801: error: no matching function for call to �physx::repx::MemoryBuffer::MemoryBuffer()�
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPoolStreams.h:151: note: candidates are: physx::repx::MemoryBuffer::MemoryBuffer(physx::repx::CMemoryPoolManager*)
/usr/include/PhysX/v3.0/SDKs/PhysXSDK/RepX/src/MemoryPoolStreams.h:149: note:                 physx::repx::MemoryBuffer::MemoryBuffer(const physx::repx::MemoryBuffer&)

Thanks

Lpd3dxmesh to nxactor

Mon, 27 Jun 2011 02:21:43 -0700

Hi.
My problem is that I do not know how to convert my lpd3dxmesh into nxactor.

so I tried:

CODE

vector GetVertices(LPD3DXMESH model)
{
    vector vertices;

    DWORD stride =  D3DXGetFVFVertexSize(modellmap["02"]->MyModell->mesh->GetFVF());
	if(stride==NULL)
	{
		::MessageBoxA(NULL,"Not working 1",NULL,NULL);
	}
    BYTE* vbptr = NULL;
   if(FAILED( modellmap["02"]->MyModell->mesh->LockVertexBuffer(0, (LPVOID*)&vbptr)))
   {
		::MessageBoxA(NULL,"Not working 2",NULL,NULL);
   }
   if(vbptr==NULL)
   {
   ::MessageBoxA(NULL,"Not working 3",NULL,NULL);
   }

    int ii = -1;
    for(int i = 0; i < modellmap["02"]->MyModell->mesh->GetNumVertices(); i++)
    {
        ii++;
        D3DXVECTOR3* pos = (D3DXVECTOR3*)vbptr;
        vertices.push_back(pos->x);
        vertices.push_back(pos->y);
        vertices.push_back(pos->z);
        
        vbptr += stride;
    }
   modellmap["02"]->MyModell->mesh->UnlockVertexBuffer();
//    delete vbptr;

    return vertices;
}


vector GetIndices(LPD3DXMESH model)
{
    LPVOID * ppData;
    DWORD stride = sizeof(short);
    BYTE* ibptr = NULL;

    short* indices = new short[modellmap["02"]->MyModell->mesh->GetNumFaces() * 3];
	if(indices==NULL)
	{
		::MessageBoxA(NULL,"Not working 4",NULL,NULL);
	}
    vector copy;

    if(FAILED(modellmap["02"]->MyModell->mesh->LockIndexBuffer(0, (LPVOID*)&indices)))
	{
		::MessageBoxA(hwnd,"Not working 5",NULL,NULL);
	}

    for(int i = 0; i < modellmap["02"]->MyModell->mesh->GetNumFaces() * 3; i++)
    {
        copy.push_back(indices[i]);
    }

   modellmap["02"]->MyModell->mesh->UnlockIndexBuffer();

    delete indices;


    return copy;
}

NxActor* GenTriangleMesh(string type,NxVec3 pos, vector indices, vector vertices, string name)
{
    int NumVerticies = vertices.size() / 3;
    int NumTriangles = indices.size() / 3;

    //Create pointer for vertices
    NxVec3* verts = new NxVec3[NumVerticies];

    int ii = NumVerticies * 3;
    for(int i = NumVerticies - 1; i >= 0; i--)
    {
        --ii;
        verts[i].x = vertices[ii];
        verts[i].y = vertices[--ii];
        verts[i].z = vertices[--ii];
    }

    ofstream out("../box.txt");
    //for(int i = 0; i < vertices.size() / 3; i++)
    //    out<    out.close();

    //Create pointer for indices
    NxU16 *tris = new NxU16[indices.size()];
    for(int i = indices.size() - 1; i >= 0; i--)
        tris[i] = i;
/*
    for(int i = 0; i < indices.size();i ++)
        Message(numtostring((float)tris[i]));
*/
    // Build physical model
    NxTriangleMeshDesc TriMeshDesc;
    TriMeshDesc.numVertices = NumVerticies;
    TriMeshDesc.numTriangles = NumTriangles;
    TriMeshDesc.pointStrideBytes = sizeof(NxVec3);
    TriMeshDesc.triangleStrideBytes = 3*sizeof(NxU16);
    TriMeshDesc.points = verts;
    TriMeshDesc.triangles = tris;
    TriMeshDesc.flags = NX_MF_16_BIT_INDICES | NX_MF_FLIPNORMALS;
    
    NxTriangleMeshShapeDesc ShapeDesc;
    NxInitCooking();

    // Cooking from memory
    MemoryWriteBuffer buf;
    bool status = NxCookTriangleMesh(TriMeshDesc, buf);
    ShapeDesc.meshData = pxsdk->createTriangleMesh(MemoryReadBuffer(buf.data));

    NxActorDesc actorDesc;
    actorDesc.shapes.pushBack(&ShapeDesc);
    actorDesc.globalPose.t = pos;

    if(!actorDesc.isValid())
		::MessageBoxA(NULL,"FAIL",NULL,NULL);


    NxActor* act = pxsc->createActor(actorDesc);
	 NxCloseCooking();
    delete[] verts;
    delete[] tris;

    return act;
}

where the error in the code?
Help me please

new tech

Sun, 26 Jun 2011 22:28:25 -0700

has anyone heard anything new about nvidia using the 28 nanometer process on upcoming graphics cards in late 2011?
or if it will be later

3d max model & collision

Sat, 25 Jun 2011 18:48:02 -0700

Hi all

another silly question

finally I managed to import 3d model from 3dmax to Physx ... and I wonder

does anyone know if that model can collide ?

can it be real rigid body and collide with other object such as "cloth"
?

it's very critical to me .. I'll appreciate help

Regards

Hazem

softbody collision with triangle mesh problem...

Fri, 24 Jun 2011 19:27:29 -0700

i recently switched to 2.8.4.5 of physx from 2.8.4.3 and while soft bodies collide with all actors in the scene when useing software mode, in hardware mode softbodies pass straight through triangle meshes. is there an extra setting or something i might have missed in the change? the soft bodies used to collide with triangle meshes in hardware mode previously. thanks for any advice.

EDIT:
does NX_SBF_HARDWARE only needs to be called if the scene is created in software mode(sceneDesc.simType = NX_SIMULATION_SW;).
as my scene is hardware do the compartment such as cloth/softbodies automatically get set to hardware mode and thus no need to call softBodyDesc.flags |= NX_SBF_HARDWARE;?

PxWheel or PxWheelShape missing ?

Fri, 24 Jun 2011 14:34:49 -0700

Hi,

I am porting my Linux PhysX 2.8.1 application (which is running fine) to PhysX 3.0.1.

I used a NxWheelShape to create wheels and now I am searching for the 3.0.1. counterpart.

When greping through the directory I found an entry in the documentation that PxWheel replaces the old NxWheelShape. Unfortunately I cannot find any class PxWheel in the whole directory?

Is this class not (yet) available in Linux or am I doing something wrong?

I found this sentence in the relase_notes.html

Wheel shapes have been replaced by the more flexible entity PxWheel. A default wheel implementation, encapsulating most of the old wheel functionality, can be found in the PhysX extensions library (see PxDefaultWheel).

But where can I find it ?

Thanks

Compilation of a simple program that uses physx on linux

Fri, 24 Jun 2011 06:05:59 -0700

Hello!

When compiling a simple program that uses basic physx functionality (version 3.0.1 for linux) lots of linker errors occur during compilation. For example:

QUOTE

undefined reference to `Gu::intersectSphereBox...
undefined reference to `Gu::distancePointSegmentSquared...
undefined reference to `Gu::intersectCapsuleMesh...

even though "GeomUtils" and all the other libraries are linked against.

I use qt build system with the following project configuration file:

CODE

TEMPLATE = app

INCLUDEPATH += .

# Input
HEADERS += glwidget.h

SOURCES += glwidget.cpp \
           main.cpp

QT += opengl

INCLUDEPATH += /home/home/libraries/physx/PhysX-3.0.1_LINUX_SDK_Core/PhysXSDK/SDKs/PhysXAPI \
               /home/home/libraries/physx/PhysX-3.0.1_LINUX_SDK_Core/PhysXSDK/PhysXFoundationSDK \
               /home/home/libraries/physx/PhysX-3.0.1_LINUX_SDK_Core/PhysXSDK/SDKs/PhysXAPI/extensions \

LIBS += -L/home/home/libraries/physx/PhysX-3.0.1_LINUX_SDK_Core/PhysXSDK/SDKs/lib/linux32 \
        -losg \
        -losgViewer \
        -lPhysX3 \
        -lFoundation \
        -lPhysX3Common \
        -lPhysX3Extensions \
        -lPhysX3CharacterKinematic \
        -lPhysX3Vehicle \
        -lPhysX3Cooking \
        -lPhysX3MetaData \
        -lLowLevel \
        -lGeomUtils \
        -lSceneQuery \
        -lSimulationController \
        -lPvdRuntime \
        -lRepX3

DEFINES += _DEBUG

QMAKE_CXXFLAGS += -malign-double

Thank you for help.

Resolved: Cg 3, tesselation

Thu, 23 Jun 2011 22:32:11 -0700

Hello. It seems like in Cg3 we have only PATCH inputs for tessellation control program for now.
How could one make a triangle tessellation then?

Thanks for your attention.

physxsdk error

Thu, 23 Jun 2011 16:48:34 -0700

Hi all ...

um working with physx 2.8.4 ... and after installation i tried to build some samples but i get that error

SDK create error (1 - NXCE_PHYSX_NOT_FOUND).
Unable to initialize the PhysX SDK, exiting the sample.

visual studio 2010
and I installed latest version of nvidia system software

and also installed nvidia 2.8.1 " because of training progs"

help me ... it's necessary ..

thx

Need setTiming explanation

Thu, 23 Jun 2011 11:27:16 -0700

Hi,
I want to ask sth. about the setTiming-Method from NxScene and want to know if my thoughts are correct.
The setTiming-Method has the following parameters:

NxReal maxTimestep
NxU32 maxIter
NxTimeStepMethod method

I read that if I use the method: NX_TIMESTEP_FIXED:
the elapsedTime is divided in several maxTimesteps, but the maximum is maxIter.

Example:
maxTimestep = 1/120
maxIter = 8
elapsedTime = 0.1ms
-> 0.1 / (1/120) = 12
that means that the simulationsteps with the size of maxTimestep should be 12 times executed,
but because of the maxIter of 8, it will be executed 8 times and the remaining 4 will be suspended to the next simulate call.
but if the elapsedTime is the same, physx will never catch up.

if something is wrong so plz correct me!!!

I wrote a programm that should simulate a newtons cradle,
but sometimes when the ball is colliding, there are mistakes.
It seems that this mistakes are happening because of the impreciseness of the simulation.
it worked better with a higher skinWidth, but there are always some mistakes.

I tried smaller simulationsteps,
instead of 1/60 (default) -> 1/240 or 1/480 and with smaller timesteps there are less mistakes.
With default (1/60) there are mistakes after the second collision.
I ran the program 5 minutes with a timestep of 1/480 and I didnt see any mistake in the simulation progress.

But WHY?
Why does a smaller simulationstep is leading to more precision, while the maxIter and the elapsedTime are constant?
There have to be a mistake in my thoughts about the simulation procedure I mentioned before.

tex2dproj not supported?

Wed, 22 Jun 2011 13:01:29 -0700

I'm trying to use tex2dproj in the HLSL profile, but the profile doesn't support it. It seems like it would pretty simple to pass tex2dproj on to the generated HLSL code. Is this something that could be added for a future version?

Just in general, I've often been frustrated by seemingly arbitrary limits imposed by the HLSL profile. It only supports 8 TEXCOORD interpolators for example. It would be great if the HLSL/GLSL profiles accepted any valid Cg code, and just let the underlying HLSL/GLSL compiler complain about any platform limitations. These limits could even be exposed as profile options, if the current default behavior is sometimes desirable.

How disable this message

Tue, 21 Jun 2011 23:47:53 -0700

Hello i'm directx developer programmer(C++, MFC)

3d vision that appears when in full screen mode

"This Application is not Rated by NVIDIA corp
Rating : UnKnown"

i'll find disable message looking for code

please Help me