Commit graph

27 commits

Author SHA1 Message Date
Christoph Oelckers
5193f6cfef - remove use of builtin deprecated varyings in shaders. 2014-07-15 00:19:41 +02:00
Christoph Oelckers
ed8a21fd86 - replaced deprecated alpha testing with shader code. 2014-07-14 21:14:43 +02:00
Christoph Oelckers
84a49e37ee - handle normals for spheremapped mirror surfaces using non-deprecated features.
- move all WGL references out of global header files so that global wgl header include is no longer necessary
2014-07-14 19:54:07 +02:00
Christoph Oelckers
ce3653f6e1 - remove all uses of builtin matrices. 2014-07-13 23:13:40 +02:00
Christoph Oelckers
dbb05c5f33 - remove use of builtin texture matrices.
- make matrix class single precision.
2014-07-13 20:41:20 +02:00
Christoph Oelckers
00fcf4bc06 - for some reason using world coordinates for clipping in the shader is somewhat imprecise so the clip plane heights have to be adjusted a bit for it. 2014-07-13 13:25:42 +02:00
Christoph Oelckers
60dc2e1122 - some shader fixes. 2014-07-13 13:01:35 +02:00
Christoph Oelckers
d868f60f6c - since the clip planes for plane mirrors did not work anymore I reimplemented them using shader based logic. It still needs to be seen if this affects performance on older hardware. 2014-07-13 12:14:12 +02:00
Christoph Oelckers
a936629cec - use default fragment shader for burn and stencil shader, with the time consuming parts disabled by a #define, to avoid code duplication. 2014-07-10 10:33:07 +02:00
Christoph Oelckers
f710518903 - use a uniform array to store vertex data to render dynamic stuff on GL 3.x hardware without the ARB_buffer_storage extension.
Due to the way the engine works it needs to render a lot of small primitives with frequent state changes.
But due to the performance of buffer uploads it is impossible to upload each primitive's vertices to a buffer separately because buffer uploads nearly always stall the GPU.
On the other hand, in order to reduce the amount of buffer uploads all the necessary state changes would have to be saved in an array until they can finally be used. This method also imposed an unacceptable overhead.
Fortunately, uploading uniform arrays is very fast and doesn't cause GPU stalls, so now the engine puts the vertex data per primitive into a uniform array and uses a static vertex buffer to index the array in the vertex shader.
This method offers the same performance as immediate mode but only uses core profile features.
2014-06-30 18:10:55 +02:00
Christoph Oelckers
1efc2938b7 - implement model vertex buffer and draw MD2 models using it instead of using the GLCommands from the model. 2014-06-29 23:24:16 +02:00
Christoph Oelckers
9d1dbf4eab - fixed: FBufferedUniform1f didn'T work because it used an int as its buffered value. 2014-06-29 14:08:44 +02:00
Christoph Oelckers
ffcb6cb70a - added second vertex coordinate attribute for model interpolation. 2014-06-29 11:00:21 +02:00
Christoph Oelckers
39775cc904 - mystery of desaturation shader solved: The old code passed '1-desaturation' to the shader, the new code 'desaturation', so for the 'mix' function to work its arguments must be swapped. 2014-05-21 15:25:25 +02:00
Christoph Oelckers
54425ee2ef - fixed: Desaturation factor was applied incorrectly.
- Also fixed some very strange thing in the shader's desaturate function. For unknown reasons using the 'mix' function there did not work.
- fixed: The fog boundary special shader could not be used.
2014-05-21 13:40:46 +02:00
Christoph Oelckers
cf45f2d718 - added missing shader files. 2014-05-12 22:24:26 +02:00
Christoph Oelckers
4d005bdfa0 shader rework
All those special shaders have been merged together.
Mostly working but the non-shader lighting seems a bit broken.
2014-05-12 14:45:41 +02:00
Christoph Oelckers
b9a6fe80a4 Do not use the shader to handle STYLEF_RedIsAlpha.
Turns out that the name doesn't accurately describe what it does.
It is correct for images that come with their own palette or are true color.
But for images using the game palette it doesn't use the red channel to determine translucency but the palette index! Ugh...

This means it cannot be done with a simple operation in the shader because it won't get a proper source image. The only solution is to create a separate texture.
2014-05-11 23:56:53 +02:00
Christoph Oelckers
7793bbbcc9 Further cleanup of lighting code.
- remove thing color from lighting calculations.
- implement alpha textures and inverse sprites for infrared as texture modes. This still requires some handling for the alpha texture mode for non-shader rendering because there is no way in the fixed pipeline to do it. The inverted texture effect can be done with a texture combiner.
- fixed: ThingColor for sprites was set in the wrong place. It must be in the Process function, not in the lighting calculation.
- added functions for isolated calculation of sprites' dynlight color.
2014-05-11 17:56:38 +02:00
Christoph Oelckers
52056a05bd - changed handling of DynLight in shader to serve as a global dynamic light color for all lighting modes. 2014-05-11 16:49:17 +02:00
Christoph Oelckers
53f4cd0108 - added objectcolor uniform. This will be used to hold the thingcolor for shader based rendering. 2014-05-11 16:06:25 +02:00
Christoph Oelckers
09f4071436 Ok, it had to be done: Removed shader support for pre GLSL 1.3/GL 3.0 hardware. The compromises needed to accomodate these are just too bad and would block any attempt at streamlining the code. 2014-05-11 13:27:51 +02:00
Christoph Oelckers
b09405a8bd - changed rendering of glowing walls so that it doesn't require an additional vertex attribute, just pass the floor and ceiling planes as uniforms. 2014-05-10 17:09:43 +02:00
Christoph Oelckers
fd767233aa - fixed typo: glColor must be gl_Color in shader. 2013-12-08 10:01:03 +01:00
Christoph Oelckers
d46315c44b - some finetuning of subtractive effect. 2013-12-05 15:39:46 +01:00
Christoph Oelckers
95163e378e - added a fog layer when drawing sprites with render style reverse subtract. 2013-12-05 15:06:10 +01:00
Christoph Oelckers
e00847f64b - GZDoom solution file
- missing GL resources for gzdoom.pk3.
2013-06-23 11:13:01 +02:00