Frederic BOUVIER wrote:
> Erik Hofman wrote :
> > > Frederic Bouvier wrote:
> > > Please ignore my previous patch. It only produce white splash. Instead,
> > > please apply the one below. The memory is deleted in the destructor.
> >
> > Well, that way the texture data is kept in memory at three different places:
> >
> > 1. Main memory (because of your patch)
> > 2. Main memory (OpenGL driver)
> > 3. Texture memory.
> >
> > We should be able to delete the texture data after sending it to OpenGL.
>
> Are you sure 1. and 2. are separate areas ? Is the copy specified in
> OpenGL ?
>
> After I submitted the first patch, I noticed 2 white splash in a row, then I
> removed the delete and the clear of the pointer and the white splash
> disappeared.
>
> I can make more test this evening to be sure.
It seems that SGTexture makes the assumption that when texture_data is null,
the texture is not yet created, without looking at texture_id.
For example,
void
SGTexture::finish(unsigned int width, unsigned int height) {
// If a texture hasn't been created then it gets created, and the contents
// of the frame buffer gets copied into it. If the texture has already been
// created then its contents just get updated.
bind();
if (!texture_data)
{
// Copies the contents of the frame buffer into the texture
glCopyTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, 0, 0,
texture_width, texture_height, 0);
} else {
// Copies the contents of the frame buffer into the texture
glCopyTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, 0, 0,
texture_width, texture_height);
}
// Set the OpenGL window back to its previous size
resize(width, height);
// Clear the window back to black
glClearColor(0.0, 0.0, 0.0, 1.0);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
}
The test is on texture_data and not texture_id
-Fred
_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
