Post-processing effects test (SSAO/HDR/Bloom)

[wraitii]

Thing is, I'm pretty sure I can't even do this sort of small checks without the extension (IF doesn't exist in basic ARB)... In which case, I might just as well use it. But then I need a way to check that the computer has the extension, or deactivate parallax mapping (else it'd crash).

By correct, you mean that the other parameters are still important or not?

[myconid]

Thing is, I'm pretty sure I can't even do this sort of small checks without the extension (IF doesn't exist in basic ARB)... In which case, I might just as well use it. But then I need a way to check that the computer has the extension, or deactivate parallax mapping (else it'd crash).

You can. CMP, SLT, SGE are comparison instructions. Look at the reference.

By correct, you mean that the other parameters are still important or not?

Not sure which other parameters you are referring to. You transformed the normal in the vertex shader and then drew that exact result in the fragment shader and it looks wrong, yes?

[wraitii]

Well, no. I multiply the unmodified instancingMatrix with a_normal, and the normals that come out are right, and parallax works properly and everything.

Didn't notice those operators, thanks.

[myconid]

Well, no. I multiply the unmodified instancingMatrix with a_normal, and the normals that come out are right, and parallax works properly and everything.

Oh, you mean the rest of the 4x4 instancingMatrix. No, those aren't needed.

[wraitii]

Well, then, I don't understand why it would be wrong (mathematically, I mean) to multiply the normals using:

vec3 normal = ((instancingTransform) * vec4(a_normal,0.0)).rgb;

instead of

vec3 normal = mat3(instancingTransform) * a_normal; (which is what you, with success, used)

And while the second one doesn't work on my computer, the first does, for some reason, so I'd rather both be correct.

(Same with the tangents using that: vec4 tangent = vec4(((instancingTransform) * vec4(a_tangent.xyz,0.0)).rgb, a_tangent.w); ).

I've pushed the ARB parallax shader (here), in a version that doesn't require the extension (commented that code). Could anybody report if it works?

Edited July 6, 2012 by wraitii

[zoot]

I've pushed the ARB parallax shader (here), in a version that doesn't require the extension (commented that code). Could anybody report if it works?

Which model should I use? I tried with structures/romans/temple_mars2 and got an error message but now I can't reproduce it.

Does anyone know how I can reset the cache thingy used by the game?

[feneur]

Which model should I use? I tried with structures/romans/temple_mars2 and got an error message but now I can't reproduce it.

Does anyone know how I can reset the cache thingy used by the game?

Just delete the files in the appropriate folder http://trac.wildfiregames.com/wiki/GameDataPaths should tell you where to find the folder for your OS

[zoot]

Thanks.

This is the error I get:

ERROR: Failed to compile vertex program 'shaders/arb/model_common.vp' (line 45): line 35, char 39: error: invalid texture coordinate unit selector
ERROR: Failed to compile vertex program 'shaders/arb/solid.vp' (line 1):
ERROR: CRenderer::EndFrame: GL errors occurred

Edited July 6, 2012 by zoot

[wraitii]

What's your graphic card? There might be a limitation on the number of texture coordinate that it can pass... Otherwise I dunno what could cause this.

Edited July 6, 2012 by wraitii

[zoot]

What's your graphic card? There might be a limitation on the number of texture coordinate that it can pass... Otherwise I dunno what could cause this.

It's an integrated Intel graphics controller.

[historic_bruno]

It's an integrated Intel graphics controller.

Can you post your system_info.txt? That should have all the juicy details.

Edit: anyone know the name of the limit / if it exists? We can use http://feedback.wildfiregames.com/report/opengl/ to see how widely supported it's likely to be.

[zoot]

Sure. Here it is.

system_info.txt

[wraitii]

That won't help much actually... I need to know the value of "GL_MAX_TEXTURE_COORDS" for you graphic card (on mine, its height, which is fairly standard, but if you have an Intel Card, it may be way lower.) If you can compile 0 A.D., you can add an std::cout << glGet(GL_MAX_TEXTURE_COORDS) << std::endl;

I can also access it from some headers in my openGL framework, so you might be able to do a search for 'GL_MAX_TEXTURE_COORDS' and get the info.

[zoot]

That won't help much actually... I need to know the value of "GL_MAX_TEXTURE_COORDS" for you graphic card (on mine, its height, which is fairly standard, but if you have an Intel Card, it may be way lower.) If you can compile 0 A.D., you can add an std::cout << glGet(GL_MAX_TEXTURE_COORDS) << std::endl;

I can also access it from some headers in my openGL framework, so you might be able to do a search for 'GL_MAX_TEXTURE_COORDS' and get the info.

Where exactly would I add that? Tried in main():

../../../source/main.cpp: In function ‘int main(int, char**)’:
../../../source/main.cpp:554:41: error: ‘glGet’ was not declared in this scope

[zoot]

This is the exact name reported by my OS:

'Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)'

[wraitii]

Yeah, hadn't thought of that. Let's roll differently. Go to HWDetect.cpp, and change this

#define INTEGER(id) do { \
GLint i = -1; \
glGetIntegerv(GL_##id, &i); \
if (ogl_SquelchError(GL_INVALID_ENUM)) \
scriptInterface.SetProperty(settings.get(), "GL_" #id, errstr); \
else \
scriptInterface.SetProperty(settings.get(), "GL_" #id, i); \
} while (false)

to this

#define INTEGER(id) do { \
GLint i = -1; \
glGetIntegerv(GL_##id, &i); \
if (ogl_SquelchError(GL_INVALID_ENUM)) \
scriptInterface.SetProperty(settings.get(), "GL_" #id, errstr); \
else \
scriptInterface.SetProperty(settings.get(), "GL_" #id, i); \
std::cout << #id << ":" << i << std::endl; \
} while (false)

Then when you start the game you'll have (in the console) MAX_TEXTURE_COORDS_ARB: ... (along with a lot of other info).

I might be able to cram info to reduce the number of texture coords I use, but I'm afraid I won't be able to change that much.

Edited July 7, 2012 by wraitii

[zoot]

I'll try that, but for now I got this from the command line:

$ glxinfo -l | grep GL_MAX_TEXTURE_COORDS
        GL_MAX_TEXTURE_COORDS_ARB = 8

[wraitii]

Mh. Then that shader shouldn't crash. What do you have for "MAX_TEXTURE_IMAGE_UNITS_ARB"?

To debug, you might try changing that line to a lower number, such as "5". It will perhaps crash, but it should crash somewhere else.

[zoot]

Mh. Then that shader shouldn't crash.

As far as I can tell from the code, you are using an coordinate index of 8 on the line where it crashes. Wouldn't that exceed the limit by one because the arrays begins at index 0?

What do you have for "MAX_TEXTURE_IMAGE_UNITS_ARB"?

I get this from the command line:

$ glxinfo -l | grep MAX_TEXTURE_IMAGE_UNITS_ARB
		GL_MAX_TEXTURE_IMAGE_UNITS_ARB = 16

Edited July 7, 2012 by zoot

[wraitii]

Isn't it 7? Looking here, it appears to be, so it shouldn't crash.

Edited July 7, 2012 by wraitii

[zoot]

I swear I haven't touched that parameter. I'll have to try pulling again.

Edit: Ah, I was one commit behind.

Edited July 7, 2012 by zoot

[zoot]

Okay, in your latest version I just get a segfault when trying to add the temple in Atlas:

Segmentation fault (core dumped)

Not sure what info would be useful for you.

[wraitii]

Could you try first loading test.pmp and then adding the temple? I get the same crash, and I'm not sure what's causing it (I also get it in GLSL, so it should come from something else).

[zoot]

Will do. Currently recompiling, for good measure.

[zoot]

Could you try first loading test.pmp

I don't think that file has been added to your commit.