2142//this is a post-processing shader, drawn in 2d
2143//there will be a render target containing sky surfaces drawn with crepuscular_sky, and everything else drawn with crepuscular_opaque (to mask out the sky)
2144//this shader then just smudges the sky out a bit as though its coming from the sun or whatever through the clouds.
2145//yoinked from http://fabiensanglard.net/lightScattering/index.php
2906//meant to be used for additive stuff. presumably particles and sprites. though actually its only flashblend effects that use this at the time of writing.
3321//FIXME: Software rendering imitation should possibly push out normals by half a pixel or something to approximate software's over-estimation of distant model sizes (small models are drawn using JUST their verticies using the nearest pixel, which results in larger meshes)
3510//FIXME: with this extra flag, half the permutations are redundant.
3511"lightlev *= 0.5; //counter the fact that the colourmap contains overbright values and logically ranges from 0 to 2 intead of to 1.\n"
3512"float pal = texture2D(s_paletted, tc).r; //the palette index. hopefully not interpolated.\n"
3513// lightlev -= 1.0 / 128.0; //software rendering appears to round down, so make sure we favour the lower values instead of rounding to the nearest
3514"col.r = texture2D(s_colourmap, vec2(pal, 1.0-lightlev.r)).r; //do 3 lookups. this is to cope with lit files, would be a waste to not support those.\n"
3515"col.g = texture2D(s_colourmap, vec2(pal, 1.0-lightlev.g)).g; //its not very softwarey, but re-palettizing is ugly.\n"
3516"col.b = texture2D(s_colourmap, vec2(pal, 1.0-lightlev.b)).b; //without lits, it should be identical.\n"
6311"#ifdef EIGHTBIT //FIXME: with this extra flag, half the permutations are redundant.\n"
6312"lightmaps *= 0.5; //counter the fact that the colourmap contains overbright values and logically ranges from 0 to 2 intead of to 1.\n"
6313"float pal = texture2D(s_paletted, tc).r; //the palette index. hopefully not interpolated.\n"
6314"lightmaps -= 1.0 / 128.0; //software rendering appears to round down, so make sure we favour the lower values instead of rounding to the nearest\n"
6315"col.r = texture2D(s_colourmap, vec2(pal, 1.0-lightmaps.r)).r; //do 3 lookups. this is to cope with lit files, would be a waste to not support those.\n"
6316"col.g = texture2D(s_colourmap, vec2(pal, 1.0-lightmaps.g)).g; //its not very softwarey, but re-palettizing is ugly.\n"
6317"col.b = texture2D(s_colourmap, vec2(pal, 1.0-lightmaps.b)).b; //without lits, it should be identical.\n"
6318"#else\n"
6319//now we have our diffuse+specular terms, modulate by lightmap values.
7236"#ifdef EIGHTBIT //FIXME: with this extra flag, half the permutations are redundant.\n"
7237"lightmaps *= 0.5; //counter the fact that the colourmap contains overbright values and logically ranges from 0 to 2 intead of to 1.\n"
7238"float pal = texture2D(s_paletted, tc).r; //the palette index. hopefully not interpolated.\n"
7239"lightmaps -= 1.0 / 128.0; //software rendering appears to round down, so make sure we favour the lower values instead of rounding to the nearest\n"
7240"gl_FragColor.r = texture2D(s_colourmap, vec2(pal, 1.0-lightmaps.r)).r; //do 3 lookups. this is to cope with lit files, would be a waste to not support those.\n"
7241"gl_FragColor.g = texture2D(s_colourmap, vec2(pal, 1.0-lightmaps.g)).g; //its not very softwarey, but re-palettizing is ugly.\n"
7242"gl_FragColor.b = texture2D(s_colourmap, vec2(pal, 1.0-lightmaps.b)).b; //without lits, it should be identical.\n"
7243"#else\n"
7244//now we have our diffuse+specular terms, modulate by lightmap values.
8809"vec3 halfdir = normalize(normalize(e_eyepos - worldPos) + lightDir); //ASSUMPTION: e_eyepos requires an identity modelmatrix (true for world+sprites, but usually not for models/bsps)\n"
10977//no bumpmapping, so we can just use distance without regard for actual surface direction. we still do scalecos stuff. you might notice it on steep slopes.
11157//no bumpmapping, so we can just use distance without regard for actual surface direction. we still do scalecos stuff. you might notice it on steep slopes.
12812//we also need to pass the result through the light's projection matrix too
12813//the 'matrix' we need only contains 5 actual values. and one of them is a -1. So we might as well just use a vec4.
12814//note: the projection matrix also includes scalers to pinch the image inwards to avoid sampling over borders, as well as to cope with non-square source image
12815//the resulting z is prescaled to result in a value between -0.5 and 0.5.
12816//also make sure we're in the right quadrant type thing
1.9.3 
