[SOLVED] Shader question

C++ coding using Panda3D
1

hello,

how would this code be translated into panda3d c++ ?

SHADER m_terrainVS.Init(Dev, "shaders/terrain.vs", VERTEX_SHADER);
D3DXHANDLE m_vsDirToSun = m_terrainVS.GetConstant("DirToSun");

I’ve loaded the shader so far like this

PT(Shader) mTerrainVS->load("Shaders/Terrain.vs", Shader::SL_GLSL);

but I haven’t found anything in the manual about getting variables (see GetConstant(…);

anyone knows ?

Chrys

2

I use python for panda, but the shader would maybe translate to :

void vshader(uniform float4 vspos_sun) 
{ 
  //view space direction of the light
  float4 vs_lightray = vspos_sun;
}

Send the sun object with :

model.set_shader_input("sun", sun);

EDIT : give more info about your shader; here it is CG code.

3

Thanks for you suggestion Manou but I’m afraid you miss-understood the problem.

I have done some progress I think… would be nice if one of the developers could confirm if this code is correct :slight_smile:

	PT(Shader) mTerrainPS;
	PT(Shader) mTerrainVS;
	CPT(RenderAttrib) mTerrainRA;

	mTerrainPS->load("Shaders/Terrain.ps", Shader::SL_GLSL);
	mTerrainVS->load("Shaders/Terrain.vs", Shader::SL_GLSL);

	mTerrainRA = ShaderAttrib::make();
	const ShaderAttrib* pTerrainSA = DCAST(ShaderAttrib, mTerrainRA);
	mTerrainRA = pTerrainSA->set_shader_input("matrixWorld", NEED_GET_THIS_MATRIX_SOMEHOW);
	pTerrainSA = DCAST(ShaderAttrib, mTerrainRA);
	mTerrainRA = pTerrainSA->set_shader_input("matrixViewProjection", NEED_GET_THIS_MATRIX_SOMEHOW );
	pTerrainSA = DCAST(ShaderAttrib, mTerrainRA);
	mTerrainRA = pTerrainSA->set_shader_input("DirToSun", mDirToSun);
	pTerrainSA = DCAST(ShaderAttrib, mTerrainRA);
	mTerrainRA = pTerrainSA->set_shader(mTerrainVS);

couple of questions here :

  1. is this code correct ?
  2. can I store the PT(RenderAttrib) for each “variable” ?
  3. What’s the best way to get the 2 matrices that are missing ? (see NEED_GET_THIS_MATRIX_SOMEHOW)
4

I don’t know if I understand what you are asking clearly enough to answer your questions.

  1. I don’t see any obvious problems in the syntax. Whether it’s “correct” is, of course, a much more abstract question that I cannot answer.

  2. I don’t know what you mean. You can store any PT(RenderAttrib) objects you like. You don’t need a separate one for each shader variable, but I guess you could store them that way if you wanted to.

  3. I don’t know what matrices you are looking for here, so how can I tell you how to get them?

David

5

my bad I thought I was clear :slight_smile:

the two matrices I am looking for are :

“matrixWorld” which I usually get as follows in directX :

D3DXMATRIX world;
D3DXMatrixIdentity(&world);
m_pDevice->SetTransform(D3DTS_WORLD, &world);

and “matrixViewProjection” which I compute like this :

D3DXMATRIX viewProjection = camera.GetViewMatrix() * camera.GetProjectionMatrix();

I was refering to the syntax indeed.
And I wanted to know if it is correct to set all my shader inputs on the same render attribute or if I should store a PT(RenderAttribute) for each shader input variable.

in directx I would usually set the shader variables on each “render” call - but panda hasn’t got a render call as such (so I will be doing it in my “update” loop).

In direct X I would get the handle to those shader variables and just set them through the handle.

hope that is clearer.

6

You can store them all in the same ShaderAttrib object; that’s the intended way to do it. But you might find it easier to use NodePath::set_shader_input(), which is a higher-level interface, instead constructing a ShaderAttrib laboriously by hand like you are doing.

It looks like you are loading the identity matrix for what you are calling “world”? In Panda, the identity matrix is LMatrix4f::ident_mat().

I guess you are computing the composition of the camera’s net transform and its lens’s projection mat for what you are calling “viewProjection”? Normally, Panda would do this for you; and if you use Cg, you can have Panda automatically compute the transform using the naming convention as described in the manual. Since you are re-using an existing shader, though, you can do something like camera.node()->get_lens()->get_projection_mat() * camera.get_net_transform()->get_mat() for this matrix. You might have to experiment to get precisely the value you are looking for. I can’t tell you precisely what you need to type since what you are doing is very specific to your own application.

David

7

okay think that answers most of my questions… one last one.

once I’ve set the “directionToSun” parameter like this

   mTerrainRA = pTerrainSA->set_shader_input("DirToSun", mDirToSun); 
   pTerrainSA = DCAST(ShaderAttrib, mTerrainRA);

will this now automatically update the “sun parameter” in the shader everytime my mDirToSun is set in my code ?
(mDirToSun is a LVector3f )

or do I need set the shader input parameter each time I want to update its content ?

8

The value of mDirToSun is copied into the shader attrib, so if it changes in the future you will need to reassign it.

David