Hi,every one here,
I am really confused with the usage of multiple pbuffer. When I use them to render to texture with Fragment Program, also with multiple texture units, I can not get the correct result.
How to manage the OpenGL states correctly with pbuffers?
Would some one here like to provide a example or a link?
Thanks in advance.
Kind of vague isn’t it?
Each p-buffer has it’s own context and all states are default as usual until you change them.
Each p-buffer has it’s own context and all states are default as usual until you change them
Yes. But I am really confused with it when I use multiple textures units with Fragment program.
For example, I render something to one texture with one pbuffer which I have set to share texture objects. And when I use this result texture in another pbuffer or in the same pbuffer, there are problems. This texture has been cleared to zero.
Hope to find a clear outway. Thanks a lot.
Lefohn had mentioned in his turtorial Dynamic Volume Computation and Visualization
On the GPU. But he didnot provide the technical details what I really need.
Does he come here?
If this thread is related to the other thread, I didn’t respond because it’s not clear what you are doing.
That said, there are things the spec isn’t clear about … but that’a another matter.
#1 Make sure wglShareList is returning TRUE
for both p-buffers.
The calls should be like this
wglShareLists(GLRC_PBuffer1, GLRC_MainWindow);
wglShareLists(GLRC_PBuffer2, GLRC_MainWindow);
You could try
wglShareLists(GLRC_PBuffer1, GLRC_MainWindow);
wglShareLists(GLRC_PBuffer2, GLRC_PBuffer1);
but I wouldn’t
#2 Which GLRC is current when you are calling the gl functions?
#3 It’s probably not a good idea to create a self-dependency if that is what you are doing.
If the p-buffer is a RTT and this RTT is created in GLRC_MainWindow and at the same time wglShareLists is used on these.
#4 I don’t know what happens if you use 2 tex ID with 1 p-buffer : releasing the p-buffer binding to another tex object, then released, binding to another.
– Try to remain on the safe side –
Thanks for your kindness. But the problem is still there.
#1 Make sure wglShareList is returning TRUE
for both p-buffers.
Yes, there is no problem about this.
wglShareLists(GLRC_PBuffer1, GLRC_MainWindow);
wglShareLists(GLRC_PBuffer2, GLRC_MainWindow);
You could try
wglShareLists(GLRC_PBuffer1, GLRC_MainWindow);
wglShareLists(GLRC_PBuffer2, GLRC_PBuffer1);
I tyied both of them, not work
#2 Which GLRC is current when you are calling the gl functions?
Yes, I have to make it current, otherwise errors occur.
#3 It’s probably not a good idea to create a self-dependency if that is what you are doing.
If the p-buffer is a RTT and this RTT is created in GLRC_MainWindow and at the same time wglShareLists is used on these.
#4 I don’t know what happens if you use 2 tex ID with 1 p-buffer : releasing the p-buffer binding to another tex object, then released, binding to another.
– Try to remain on the safe side --[/b][/QUOTE]
Here I provide an example to illustrate my prblem, and hope to have a discussion here. Thanks for joining.
#include “pbuffer.h”
int Width = 16;
int Height = 16;
unsigned int ShaderID1;
unsigned int ShaderID2;
unsigned int tex0_id;
unsigned int tex1_id;
void testtwopbuffer()
{
int i;
///////////////////////////////////////////////////////
///// Bind the Fragment Program to current drawing
LoadFragmentProgramFromFile(“…\ est1.fp”, ShaderID1);
LoadFragmentProgramFromFile(“…\ est2.fp”, ShaderID2);
///Setup two pbuffers, and the class CFloatPBuffer from RTT of Harris
CFloatPBuffer* pbuffer[2];
for(i=0;i<2;i++)
{
pbuffer[i] = new CFloatPBuffer(Width, Height);
///Here I have set to share resource contexts
pbuffer[i]->Initialize();
pbuffer[i]->BeginCapture();glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); glViewport(0,0,Width,Height); glMatrixMode ( GL_PROJECTION ); glLoadIdentity(); glOrtho( 0, Width, 0, Height, -1.0, 1.0 ); glMatrixMode ( GL_MODELVIEW ); glLoadIdentity(); glEnable(GL_FRAGMENT_PROGRAM_NV); pbuffer[i]->EndCapture();}
int whichone=0;
///Here I have to make one pbuffer current, so the texture data can be correctly used in the following program
pbuffer[whichone]->BeginCapture();
LoadData(tex0_id);
LoadData(tex1_id);
pbuffer[whichone]->EndCapture();
whichone++;
whichone = whichone%2;char filename[20];
///Output the textures, the result is correct.
output(“…\ ex0-1.txt”, tex0_id);
output(“…\ ex1-1.txt”, tex1_id);////Set the tex1_id as the target of RTT
for(i=0;i<3;i++)
{
glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex1_id);
pbuffer[whichone]->BeginCapture();
glBindProgramNV(GL_FRAGMENT_PROGRAM_NV, ShaderID1);glActiveTextureARB(GL_TEXTURE0_ARB); glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex0_id); /// The target can also be used as an input texture glActiveTextureARB(GL_TEXTURE1_ARB); glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex1_id); glBegin(GL_QUADS); glVertex2i(0,0); glVertex2i(Width,0); glVertex2i(Width,Height); glVertex2i(0,Height); glEnd(); glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex1_id); pbuffer[whichone]->EndCapture(); ////In order to get correct result, two pbuffers must be used and swapped in turn. whichone++; whichone = whichone%2;}
///output to have check, no problems here!
output(“…\ ex0-2.txt”, tex0_id);
output(“…\ ex1-2.txt”, tex1_id);///Here is important to show my problem.
///If the target of RTT is same to the above one, NO errors.
///But if I change it to other textures, error occurs.
///That’s to say
///target = tex1_id, NO Wrong.
///target = tex0_id, no error to the first pass(i=0),
///but in the next pass, the texture data of tex1_id has been changed.
unsigned int target = tex0_id;
for(i=0;i<2;i++)
{
glBindTexture(GL_TEXTURE_RECTANGLE_NV, target );
pbuffer[whichone]->BeginCapture();
glBindProgramNV(GL_FRAGMENT_PROGRAM_NV, ShaderID2);glActiveTextureARB(GL_TEXTURE0_ARB); glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex0_id); glActiveTextureARB(GL_TEXTURE1_ARB); glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex1_id); glBegin(GL_QUADS); glVertex2i(0,0); glVertex2i(Width,0); glVertex2i(Width,Height); glVertex2i(0,Height); glEnd(); ///Here will be changed accordingly. glBindTexture(GL_TEXTURE_RECTANGLE_NV, target ); pbuffer[whichone]->EndCapture(); //// TO check the immediate texture data sprintf(filename,"..\ ex0----%d.txt", i); output(filename, tex0_id); ///Here if target = tex0_id, the data of tex1_id is changed arbitrarily(i>0) sprintf(filename,"..\ ex1----%d.txt", i); output(filename, tex1_id); whichone++; whichone = whichone%2;}
}
Thanks for any idea.
You have to be careful which context is current when you are calling gl functions like in this case.
///////////////
for(i=0;i<2;i++)
{
glBindTexture(GL_TEXTURE_RECTANGLE_NV, target );
pbuffer[whichone]->BeginCapture();
///////////////
I don’t know what BeginCapture and EndCapture does.
If BeginCapture is
wglReleaseTexImageARB(HDC_PBUFFER, WGL_FRONT_LEFT_ARB);
wglMakeCurrent(HDC_PBUFFER, GLRC_PBUFFER);
and EndCapture
wglBindTexImageARB(HDC_PBUFFER, GL_FRONT_LEFT_ARB);
then it should be fine. But before you call EndCapture, be sure to have the right context current (probably the main window)
//////////////////////Warning!
glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex1_id);
pbuffer[whichone]->EndCapture();
//////////////////////
maybe you should make something downloadable. I’ll have a look when I have time.
[This message has been edited by V-man (edited 02-17-2004).]
[This message has been edited by V-man (edited 02-17-2004).]