GLSL

[Gig]

Okay, this seems to avoid the error. Now I can see something actually different.
But the result isn't exactly "cool"...
Take a look to the screenshot.
I think this is the version that was in the folder "3" of your zip, but I'm not sure. Is it supposed to look like that?

I must go now, I will try again the various versions of the sparks, and I will try the water shader (although some more usage infos would be welcome for that...) tomorrow.

[Hitchhiker]

thanks for trying Gig, actually that is the effect I made 
for some reason I could not get the entire box but only vertical stripes so the space to do more was limited.
all the other versions are more or less the same thing so no need to try them.
I'll see if I can get something else drawn there.

[Gig]

For "the entire box", do you mean the six sides of the brush?
For my "sparks" brushes I used my gig_stripes1 shader on one face, and "nodraw" on the other five (if you check with r_showtris 1, you can see the other faces are not rendered at all): the "box" effect looked bad with my shader, so I made it "2D".

For the GLSL effect of the sparks I thought about something like adding some sort of "bump mapping with light from changing angle" to my texture.
Of course my texture is quite small since the beginning, hence the "space" available to draw cool effects there is very little... if it is a problem, you can focus your attention on other shaders instead.

[Gig]

Well, I'm trying to test the water effect, but I have some problems.
- I placed your waternormal.jpg in my "textures/gig" folder.
- I placed stream_fp.glsl and stream_vp.glsl in a "glsl/water" folder.
- I placed "stream.shader" in my "scripts" folder.
- I edited it, adjusting two paths, and removing all your comments starting with "-" (I thought comments begin with "//" ... I see you used "-" instead, is that ok?) and removing the second (short) version of the shader, trying to avoid conflitcs (more infos about the two versions in your shader file would be welcome). Now the shader file looks like this.

textures/liquids/clear_ripple2
{
   qer_editorimage textures/liquids/pool3d_3.jpg
   surfaceparm trans
   surfaceparm nonsolid
   surfaceparm water
   cull disable
   deformVertexes wave 64 sin .25 .25 0 .5   

   {
      map textures/liquids/pool3d_5.jpg
      blendFunc gl_dst_color gl_one
      rgbgen identity
      tcmod scroll -.025 .001
      //GLSL blendFunc gl_dst_color gl_src_color
      //GLSL map2 textures/gig/pebbles1.jpg
      //GLSL map3 textures/gig/waternormal.jpg
      //GLSL program stream
      //GLSL vertexProgram glsl\water\stream_vp.glsl
      //GLSL fragmentProgram glsl\water\stream_fp.glsl   
   }

   {
      map textures/liquids/pool3d_6.jpg
      blendFunc gl_dst_color gl_one
      rgbgen identity
      tcmod scroll .03 -.001
   }
   {
      map textures/liquids/pool3d_5e.jpg
      blendFunc gl_dst_color gl_one
      rgbgen identity
      tcmod scroll .001 .025
   }
   {
      map $lightmap
      blendFunc gl_dst_color gl_zero
      rgbgen identity      
   }
   

}

I loaded my map, and the water looks like as usual. I've done a vid_restart and checked the console. I can see this error:

Error:
Fragment shader failed to compile with the following errors:
ERROR: 0:63: error(#202) No matching overloaded function found dot
ERROR: 0:64: error(#202) No matching overloaded function found reflect
ERROR: 0:64: error(#202) No matching overloaded function found dot
ERROR: error(#273) 3 compilation errors.  No code generated

Couldn't compile fragment shader for program streamWARNING:
RE_GLSL_RegisterProgram failed for 'program stream' in shader
'textures/liquids/clear_ripple2'

What's the matter?

I also tried keeping the "short" version instead:

textures/liquids/clear_ripple2
{
   qer_editorimage textures/liquids/pool3d_3.jpg
   surfaceparm trans
   surfaceparm nonsolid
   surfaceparm water
   cull disable
   deformVertexes wave 64 sin .25 .25 0 .5   

   {
      map textures/liquids/pool3d_5.jpg
      blendFunc blend
      rgbgen identity
      tcmod scroll -.025 .001
      //GLSL map2 textures/gig/pebbles1.jpg
      //GLSL map3 textures/gig/waternormal.jpg
      //GLSL program stream
      //GLSL vertexProgram glsl\water\stream_vp.glsl
      //GLSL fragmentProgram glsl\water\stream_fp.glsl   
   }
   {
      map $lightmap
      blendFunc gl_dst_color gl_zero
      rgbgen identity      
   }
}

Glsl program compiling still fails with the same errors as above (predictable, calling the same exact glsl program)... but this time the "classic" shader looks very bad, showing a non-transparent greyscale texture.

[Hitchhiker]

Hi Gig and again THANKS for testing!
I see I will have to be more careful about mixing vec3 and vec4 variables as ATI cards seem to be more strict than nvidia..

Could you try replacing the lines 63 and 64 with this:
//calculate diffuse and specular values   
   float diffuse= clamp( dot (bump.xyz,normalize(lpos)) ,0 , 0.8 );
   float specular = pow(clamp(dot(normalize (reflect(normalize(-vpos), bump.xyz) ), normalize(lpos)), 0.0, 1.0), 64.0 );

I think this should do it.

The gray texture shows as I don't know how to switch off the blendfunc once it was set in a shader. So you could remove the  //GLSL blendFunc gl_dst_color gl_src_color from the shader but then the water and refraction effect is kind of lost.

Possibly (as I have few additions to the renderer) I could create a separate renderer dll and then the patch that loads shaders particular to the renderer could resolve these foldername, blendfunc, shader-name and other issues. To be considered.. and maybe Sago007 or fromhell can share their thoughts on this. But I think for this to work for everybody a new release of OA would be needed as the patch for different renderers was added after 0.8.8?

As for sparks, I've remembered a line from tutorial.. the one that says that glsl can modify vertex positions. so I did that and scaled the box a little - the new shader is a bit nicer. I'll post it tonight.

[Hitchhiker]

hi again,
the project fail.. not so fancy sparks effect here attached. I know it sounds fake but I had a much better effect towards the end but being so well organized I managed to overwrite the latest version with the one from the beginning of the evening..
hopefully will manage to recreate it.. not even worth posting this one

[fromhell]

Gig, what's your video chipset again? Are you certain it properly supports GLSL?

[Gig]

Gig, what's your video chipset again? Are you certain it properly supports GLSL?

ATI Radeon HD 5670.
GLSL works, it correctly shows things such as "r_postprocess edges" program.

UPDATE: With that fix in stream_vp.glsl, the water works. AND IT'S AWESOME.
It's so nice I had to immediately post a video of it.
I will comment more in a later post.
Video is .avi in xvid format. I had to place it into a .zip because the forum does not allow .avi extension. Sorry for the poor quality, but the forums allow to upload very little files only.

[Gig]

Hitch, first of all, congratulations for that water: it's really cool!!!! Thank you for that, you just got a cake! Q3 water should have looked like that since the beginning!

Then... I've done more tests, and I noticed some problems.

Using the "short" version of the .shader (see this post), my "pebbles1" texture is the only one players see "under the surface". The shader surface is not transparent: as you can see in the first screenshot, the part of the proximity mine that is under it is not visible at all. "Marks on walls" are not shown.

Using the "long" version of the .shader, my "pebbles1" texture is semi-transparent, and I can see the aqueduct texture below: I like that effect very much (although I suppose transparent pebbles do not exist in real world, and the aqueduct texture does not have refraction -as you can see, blocks lines are not distorted-... but the glance is cool anyway)! As told, the shader is transparent: as you can see in the second screenshot, I can see the part of the proximity mine that is under the water surface, good. Small problem: "marks on walls" appear "doubled", especially if you are not vertically above them (like if a mark is left on the bottom of the pool, and a second mark is left on the water surface?).

Major problem: with both shaders, if I disable r_ext_vertex_shader, then the water looks in ugly black and white (see third screenshot). I really hope there is some way to workaround this problem without the need of a new OA engine version, otherwise that effect could not be in OACMP1, and that would be a shame.
PS: also the non-glsl version of the shader should show "flowing" water (e.g. like in that shader from Akom).

I have not tested yet what happens if applying this water to a proper "pool" deep enough to swim. I imagine pebbles could suddently disappear when you go underwater, but I have to test yet... however for this particular map I don't need to go underwater.

[Gig]

About that second "sparks" package:

Sorry, the glsl program does not compile:

Error:
Fragment shader failed to compile with the following errors:
ERROR: 0:38: error(#202) No matching overloaded function found pow
ERROR: 0:38: error(#160) Cannot convert from 'const float' to 'highp
3-component vector of float'
ERROR: error(#273) 2 compilation errors.  No code generated

Couldn't compile fragment shader for program sparksWARNING:
RE_GLSL_RegisterProgram failed for 'program sparks' in shader
'textures/gig/gig_stripes1'

For completion, I post here also the .shader file (that is the same of the previous sparks .zip... maybe there is something that should be updated to fit the new version of the glsl program?):

textures/gig/gig_stripes1
{
        surfaceparm trans   
        surfaceparm nomarks   
        surfaceparm nonsolid
      surfaceparm nolightmap
        cull none
               
   {
      map textures/gig/gig_stripes1.tga
      //GLSL program sparks
      //GLSL vertexProgram glsl\sparks\sparks_vp.glsl
      //GLSL fragmentProgram glsl\sparks\sparks_fp.glsl
        tcMod Scroll .5 .6
        blendFunc GL_ONE GL_ONE
        }
           
}

[CGB01]

Very cool water effect Hitchhiker and nice video Gig.  Wanted to see if I could do the same and sadly, it's not working for me.

From the console, I'm getting the following message;

glsl programs: enabled  (Ok, so I did the /r_ext_vertex_shader 1)
Warning:  Cannot locate postprocessing glsl program stream (I did the /r_postprocess stream)

The stream_vp.glsl and stream_fp.glsl are in the glsl/water/ directory and I've copied Gig's shader code.

What am I doing wrong? 

[Gig]

glsl programs: enabled  (Ok, so I did the /r_ext_vertex_shader 1)
Warning:  Cannot locate postprocessing glsl program stream (I did the /r_postprocess stream)
What am I doing wrong?  

GLSL programs can be user-side (to be used with /r_postprocess <programname> from console), or mapper-side (invoked from a shader used on a certain brush in a map). In the first case (postprocess mode), they affect the whole screen; in the second case, they affect only that shader/texture. They are two completely different things... Hitch said there is no easy method to guess if a certain glsl program is to be used for postprocess mode or to be invoked by a shader: I suppose the glsl author should mention this in some comment line inside the shader itself (another thing that should be said in the tutorial!).

However, we are talking about glsl programs invoked by shaders ("stream" glsl program requires some texture names as "input variables", and this info comes from the shader).
I had to put together a few files from different posts, place them in the correct folders, apply some fixes Hitch mentioned in other posts, and fix some paths mentioned in the shader (very simple work compared with what Hitch did, but it's not so automatic for a shader beginner like me).

For your convenience, I've just packed them up into pk3, so you can use them more easily. It's only the "flowing water" shader, not the "sparks" one (that does not yet work for me).

So:
1) get udestruction v7 (BETA map from mine) .pk3 from here, and save it under your baseoa folder (it contains textures/gig/pebbles1.jpg, and the map uses textures/liquids/clear_ripple2 on the aqueduct).
2) download the two .pk3 files you find attached to this post, and place GLSL_TEST_SHORT.PK3 in your baseoa folder (this gives you the shader -a glsl powered version of clear_ripple2 that of course will require a different name in the "final" map, because the original clear_ripple2 is not supposed to be "flowing"-, the glsl program, and Hitch's waternormal.jpg file).
3) Launch openarena, be sure you have r_ext_vertex_shader 1, then do /DEVMAP UDESTRUCTION, and go on the aqueduct. You should see the cool water flowing, with the water surface non-transparent.

Then close the game, remove GLSL_TEST_SHORT.PK3 from baseoa and place GLSL_TEST_LONG.PK3 there. Launch the game and load udestruction again. This time, the water surface will be semi-transparent.

If you wanna do experiments on your own, you could extract the content of the .pk3 file (at that point, you can even delete the .pk3), and modify the .shader (e.g. to load a different texture in the place of gig/textures/pebbles1.jpg).
Please remember that, if you need to use stuff that is not packaged into a pk3 (e.g. your personal edited version of the shader), you need to set /sv_pure 0 before loading the map.

[CGB01]

1) get udestruction v7 (BETA map from mine) .pk3 from here, and save it under your baseoa folder (it contains textures/gig/pebbles1.jpg, and the map uses textures/liquids/clear_ripple2 on the aqueduct).
2) download the two .pk3 files you find attached to this post, and place GLSL_TEST_SHORT.PK3 in your baseoa folder (this gives you the shader -a glsl powered version of clear_ripple2 that of course will require a different name in the "final" map, because the original clear_ripple2 is not supposed to be "flowing"-, the glsl program, and Hitch's waternormal.jpg file).
3) Launch openarena, be sure you have r_ext_vertex_shader 1, then do /DEVMAP UDESTRUCTION, and go on the aqueduct. You should see the cool water flowing, with the water surface non-transparent.

This is the console log after downloading the files ...

ERROR: 0:28: '[' :  array must be redeclared with a size before being indexed with a variable
WARNING: 0:52: 'assign' : implict conversion between types allowed from GLSL 1.20
WARNING: 0:56: 'texture' : function not available in current GLSL version
WARNING: 0:58: 'assign' : implict conversion between types allowed from GLSL 1.20
WARNING: 0:58: 'texture' : function not available in current GLSL version
WARNING: 0:63: 'clamp' : implicit conversion is allowed from GLSL 1.20
WARNING: 0:63: 'assign' : implict conversion between types allowed from GLSL 1.20
WARNING: 0:67: 'texture' : function not available in current GLSL version
WARNING: 0:67: 'texture' : function not available in current GLSL version
WARNING: 0:67: 'assign' : implict conversion between types allowed from GLSL 1.20
WARNING: 0:69: 'assing' : implict conversion between types allowed from GLSL 1.20


Couldn't compile fragment shader for program stream
WARNING: RE_GLSL_RegisterProgram failed for 'program stream' in shader 'textures/liquids/clear_ripple2'


Guess there's still something wrong somewhere. 

[Gig]

Did you try also the "long" shader? I can see it noone downloaded it yet.

Which video card are you using? Maybe its glsl support is not good, or for that some other tweak is needed... or I did something wrong in the pk3 (I tried it on a separate oa install, but on the same machine).

Do things like "r_postprocess edge" work with you?

[CGB01]

Sorry Gig, had to run out this morning.  Yes, I just tried downloading the file and the same error occurs.

"\r_postprocess edge" and a "\vid_restart" gives me an error as well 'Cannot locate postprocessing glsl program edge'.

Hmm, wonder if it's because I'm trying to run this on my laptop.  It has a built in graphics card (Intel based), although I'm not getting any errors during the R_init phase of the game start.

[Gig]

AFAIK, Intel integrated graphic cards are not good for gaming. I would suggest you to try on another pc.

By the way, I suppose that adding glsl option to the menu would be good, but that should continue being disabled by default.. cards compatibility is uncertain...

[fromhell]

Generally anything made after 2002 with shader model 2.0 (Radeon 9500/GeforceFX 5200 and above) supports GLSL IIRC

[Gig]

Sorry, yesterday I wrote that post from the cell phone and I didn't notice I left a character out. That was supposed to be 'r_postprocess edges', with the final "s". My fault.

Anyway, I have a PC (dated around 2008-2009) with integrated Intel graphics, where GLSL doesn't work.

@Hitch. Another thing for when you will update the GLSL tutorial: I suppose it would be good to link a GLSL language specification document (http://en.wikipedia.org/wiki/GLSL#References), but which one?
I read that GLSL language evolved with various OpenGL versions (http://en.wikipedia.org/wiki/GLSL#Versions), and each version has got its own document (examples, version 1.1 and version 4.3: http://www.opengl.org/registry/doc/GLSLangSpec.Full.1.10.59.pdf http://www.opengl.org/registry/doc/GLSLangSpec.4.30.8.pdf). Which one should we tell users to follow? The one that OA GLSL implementation follows more stictly? The very first one, in hope to assure compatibility with older video cards? Or what?

[CGB01]

Sorry, yesterday I wrote that post from the cell phone and I didn't notice I left a character out. That was supposed to be 'r_postprocess edges', with the final "s". My fault.

No problem   Yes, I tried this morning with the "edges" and it does work.  So the question still remain, why are your pk3 files not working?

[Gig]

I don't know. Maybe your graphic chipset requires even more strict "formal language" following than AMD cards (Hitch already had to write down small fixes to have the same effects that worked on his NVIDIA card to also work on my AMD card).
Or, looking at the error messages you posted, the "stream" glsl program may use some GLSL features that are not supported by your video card (or by its drivers). From the very little I can guess from those messages, it looks like that GLSL program uses GLSL 1.2 version features, and maybe your graphic card (or its driver!) supports GLSL 1.1 features only. Maybe the "edges" program, instead, just requires 1.1 features. Just guessing. I suppose Hitch could be able to give a better analysys of the problem.

However, Intel drivers GLSL support sounds like a problem also for other people:
http://www.ozone3d.net/blogs/lab/20080611/glsl-support-in-intel-graphics-drivers/
I have not yet found a page listing all the various GLSL versions supported by the various Intel chipsets.
However, I found a page about OpenGL support for Intel 4 Series chipsets (and it mentions Vista and XP driver versions only...):
http://software.intel.com/en-us/articles/opengl-extensions-supported-in-intel-4-series-express-chipsets-and-beyond
It says that GLSL 1.2 is supported since 14.39 drivers for Vista and since 15.14 drivers for XP.

I would suggest you to try updating your Intel video drivers, if possible. Otherwise, try on another machine.

[CGB01]

There's a program on ozone3d.net to test your GPU's capabilities called "Geek3D GPU Cap Viewer".  I've downloaded it and it reports that I have OpenGL version 3.1 (with 116 extensions) and that I support GLSL version 1.40.  My driver version is 15.10 and HP does not have a newer version for my laptop

Guess I'm SOL for getting Hitch's stream version going 

Once I get settled into the new place (I'm currently moving) I'll get the desktop system setup and try again.

Thanks for your help Gig!

Cheers.

[Hitchhiker]

the message about glsl version 1.20 is just a warning. I've modified the 'stream' program so it might work on Intel as well. version 1.1 here attached.

[Hitchhiker]

worked also on pool kind of water (basically a modified version of the water in ZEQ thread).
program and screenshot here attached. I did try to verify all the variables are of the correct type and hope it will work out of the box on ATI and Intel.

[CGB01]

the message about glsl version 1.20 is just a warning. I've modified the 'stream' program so it might work on Intel as well. version 1.1 here attached.

Hey Hitch, thanks for trying.  It didn't work either, I get the same compile errors.

[Hitchhiker]

Hi CGB01 and thank you for trying as well
If you want to give it another go, I've cleaned the programs of multitexturing stuff and the updated version is here attached. If you try and it gives an error again, could you post the error message again please?
Thanks!