Updating shaders to use no sampling.

[reepblue]

Hey all,

 

I decided for fun to sync up the Vectronic Demo with the latest version. I also wanted to see how the game looked with multi sampling off. when I booted up the game however, it lacked fog. No big deal I thought, I'll just use the #if SAMPLES==0 statements like the light shaders do, and make ms textures 2d textures.

 

I did just that, and it works.... Only if SAMPLES==0 and MSAA is set to 0 as well. I'm curious why there isn't an automatic system in the shader system to detect if MSAA is 0. (I thought #if SAMPLES==0 was it.)

 

I've looked at the decals and they use an int that's most likely toggled under the hood. How would I do this with any Post Process effects that are loaded? In the PBR code, there was a way to toggle an int value, but could I just load the shader and toggle it, or do I need to fetch the active camera, get the shaders of the post processing effects, and do it that way?

 

Perhaps a global sampling value would be nice here so other users don't collide with this issue while making/updating shaders. Maybe actually Set SAMPLES to 0 when Multisampling is 0? This might be more of a request more than anything.

 

Thanks.

 

Oh, for those who want the semi-fix, here:

 

_passthrough.shader (Frag)

#version 400

#ifndef SAMPLES

#define SAMPLES 1

#endif

 

uniform bool isbackbuffer;

uniform vec2 buffersize;

uniform vec2 camerarange;

uniform float currenttime;

 

out vec4 fragData0;

 

uniform sampler2D texture1;

 

#if SAMPLES==0

uniform sampler2DMS texture2;

uniform sampler2DMS texture3;

#else

uniform sampler2D texture2;

uniform sampler2D texture3;

#endif

 

void main()

{

vec2 icoord = vec2(gl_FragCoord.xy/buffersize);

if (isbackbuffer) icoord.y = 1.0 - icoord.y;

vec4 color = texture(texture1,icoord);

fragData0=color;

}

 

_klepto_fog.shader (Frag)

#version 400

#ifndef SAMPLES

#define SAMPLES 1

#endif

 

uniform sampler2D texture1;

uniform samplerCube texture2;

 

#if SAMPLES==0

uniform sampler2D texture3;

#else

uniform sampler2DMS texture3;

#endif

 

uniform bool isbackbuffer;

uniform vec2 buffersize;

out vec4 fragData0;

uniform samplerCube uTexture;

smooth in vec3 eyeDirection;

uniform vec2 camerarange;

uniform float camerazoom;

uniform vec3 cameraposition;

uniform mat4 camerainversematrix;

uniform mat4 projectionmatrix;

uniform mat4 cameramatrix;

 

uniform vec2 fogrange = vec2(0.0,15.0);

uniform vec4 fogcolor = vec4(0.72,0.73,0.67,1.0);

uniform vec2 fogangle = vec2(5.0,15.0);

uniform bool fogislocal = false;

uniform float clipheight= 0.0;

 

 

float DepthToZPosition(in float depth) {

return camerarange.x / (camerarange.y - depth * (camerarange.y - camerarange.x)) * camerarange.y;

}

 

void main() {

//integer screen coordinates

//needed for depth lookup

ivec2 icoord = ivec2(gl_FragCoord.xy);

if (isbackbuffer) icoord.y = int(buffersize.y) - icoord.y;

 

//floating screencoords normalised to range 0-1

vec2 coord = vec2(gl_FragCoord.xy/buffersize);

if (isbackbuffer) coord.y = 1.0 - coord.y;

 

//fetch depth value

float depth;

 

#if SAMPLES==0

depth = texelFetch(texture3,icoord,0).x;

#else

depth = texelFetch(texture3,icoord,0).x;

#endif

 

//calculating worldposition from cameramatrix,screenposition and depth

//normalize it to get the cubecoords

vec3 screencoord;

screencoord = vec3(((coord.x)-0.5) * 2.0,((coord.y)-0.5) * 2.0 / (buffersize.x/buffersize.y),DepthToZPosition( depth ));

screencoord.x *= screencoord.z;

screencoord.y *= -screencoord.z;

vec4 worldpostemp= vec4(screencoord,1.0) * camerainversematrix;

vec3 worldpos = worldpostemp.xyz;// / 1.0-worldpostemp.w;

worldpos+=cameraposition;

vec3 cubecoord = normalize(worldpos.xyz);

 

fragData0 = texture(texture1,coord);

 

if(depth == 1.0) //no geometry rendered --> background

{

vec3 normal=normalize(cubecoord);

normal.y=max(normal.y,0.0);

float angle=asin(normal.y)*57.2957795-fogangle.x;

float fogeffect=1.0-clamp(angle/(fogangle.y-fogangle.x),0.0,1.0);

fogeffect *= fogcolor.w;

fragData0=fragData0*(1.0-fogeffect)+fogeffect*fogcolor;

}

else // no background - render input + fog to output

{

float lineardepth = DepthToZPosition(depth);

float fogeffect = clamp( 1.0 - (fogrange.y - lineardepth) / (fogrange.y - fogrange.x) , 0.0, 1.0 );

if (fogislocal && lineardepth > fogrange.y) fogeffect=0.0;

fragData0 = fragData0 * (1.0 - fogeffect) + fogcolor * fogeffect;

}

//if (cameramatrix[3][1] <= -0.99) fragData0=texture(texture1, icoord);

}

[Josh]

****, that is no good. I don't like this.

 

On the plus side, AMD has responded to me.

[reepblue]

Bout time! I've went to the post to see what they've said, but I figure it was a PM/E-mail as nothing else was posted.

 

I still like the idea of turning multi-sampling completely off personally, but I understand that this is getting real messy, real fast. However, you're already knee deep in this process...