ShuriZma
/
pcsx2
mirror of https://github.com/PCSX2/pcsx2.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

glsl: expand tab into space 

The mix of the 2 was awful
This commit is contained in:
Gregory Hainaut 2015-08-21 00:33:45 +02:00
parent a46c11b088
commit bfadd884c9
3 changed files with 500 additions and 500 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

474
plugins/GSdx/res/glsl/tfx_fs.glsl
View File

@ -26,9 +26,9 @@
in SHADER
{
vec4 t;
vec4 c;
flat vec4 fc;
vec4 t;
vec4 c;
flat vec4 fc;
} PSin;
#define PSin_t (PSin.t)
@ -62,234 +62,234 @@ layout(r32i, binding = 2) coherent uniform iimage2D img_prim_min;
// Warning duplicated in both GLSL file
layout(std140, binding = 21) uniform cb21
{
vec3 FogColor;
float AREF;
vec3 FogColor;
float AREF;
vec4 WH;
vec4 WH;
vec2 _pad0;
vec2 TA;
vec2 _pad0;
vec2 TA;
uvec4 MskFix;
uvec4 MskFix;
uvec4 FbMask;
uvec4 FbMask;
vec3 _pad1;
float Af;
vec3 _pad1;
float Af;
vec4 HalfTexel;
vec4 HalfTexel;
vec4 MinMax;
vec4 MinMax;
vec2 TextureScale;
vec2 TC_OffsetHack;
vec2 TextureScale;
vec2 TC_OffsetHack;
};
vec4 sample_c(vec2 uv)
{
return texture(TextureSampler, uv);
return texture(TextureSampler, uv);
}
vec4 sample_p(float idx)
{
return texture(PaletteSampler, vec2(idx, 0.0f));
return texture(PaletteSampler, vec2(idx, 0.0f));
}
vec4 clamp_wrap_uv(vec4 uv)
{
vec4 uv_out = uv;
vec4 uv_out = uv;
#if PS_WMS == PS_WMT
#if PS_WMS == 2
uv_out = clamp(uv, MinMax.xyxy, MinMax.zwzw);
uv_out = clamp(uv, MinMax.xyxy, MinMax.zwzw);
#elif PS_WMS == 3
uv_out = vec4((ivec4(uv * WH.xyxy) & ivec4(MskFix.xyxy)) | ivec4(MskFix.zwzw)) / WH.xyxy;
uv_out = vec4((ivec4(uv * WH.xyxy) & ivec4(MskFix.xyxy)) | ivec4(MskFix.zwzw)) / WH.xyxy;
#endif
#else // PS_WMS != PS_WMT
#if PS_WMS == 2
uv_out.xz = clamp(uv.xz, MinMax.xx, MinMax.zz);
uv_out.xz = clamp(uv.xz, MinMax.xx, MinMax.zz);
#elif PS_WMS == 3
uv_out.xz = vec2((ivec2(uv.xz * WH.xx) & ivec2(MskFix.xx)) | ivec2(MskFix.zz)) / WH.xx;
uv_out.xz = vec2((ivec2(uv.xz * WH.xx) & ivec2(MskFix.xx)) | ivec2(MskFix.zz)) / WH.xx;
#endif
#if PS_WMT == 2
uv_out.yw = clamp(uv.yw, MinMax.yy, MinMax.ww);
uv_out.yw = clamp(uv.yw, MinMax.yy, MinMax.ww);
#elif PS_WMT == 3
uv_out.yw = vec2((ivec2(uv.yw * WH.yy) & ivec2(MskFix.yy)) | ivec2(MskFix.ww)) / WH.yy;
uv_out.yw = vec2((ivec2(uv.yw * WH.yy) & ivec2(MskFix.yy)) | ivec2(MskFix.ww)) / WH.yy;
#endif
#endif
return uv_out;
return uv_out;
}
mat4 sample_4c(vec4 uv)
{
mat4 c;
mat4 c;
// Note: texture gather can't be used because of special clamping/wrapping
// Also it doesn't support lod
c[0] = sample_c(uv.xy);
c[1] = sample_c(uv.zy);
c[2] = sample_c(uv.xw);
c[3] = sample_c(uv.zw);
c[0] = sample_c(uv.xy);
c[1] = sample_c(uv.zy);
c[2] = sample_c(uv.xw);
c[3] = sample_c(uv.zw);
return c;
return c;
}
vec4 sample_4_index(vec4 uv)
{
vec4 c;
vec4 c;
// Either GSdx will send a texture that contains a single channel
// in this case the red channel is remapped as alpha channel
//
// Or we have an old RT (ie RGBA8) that contains index (4/8) in the alpha channel
// Either GSdx will send a texture that contains a single channel
// in this case the red channel is remapped as alpha channel
//
// Or we have an old RT (ie RGBA8) that contains index (4/8) in the alpha channel
// Note: texture gather can't be used because of special clamping/wrapping
// Also it doesn't support lod
c.x = sample_c(uv.xy).a;
c.y = sample_c(uv.zy).a;
c.z = sample_c(uv.xw).a;
c.w = sample_c(uv.zw).a;
c.x = sample_c(uv.xy).a;
c.y = sample_c(uv.zy).a;
c.z = sample_c(uv.xw).a;
c.w = sample_c(uv.zw).a;
uvec4 i = uvec4(c * 255.0f + 0.5f); // Denormalize value
uvec4 i = uvec4(c * 255.0f + 0.5f); // Denormalize value
#if PS_IFMT == 1
// 4HH
return vec4(i >> 4u) / 255.0f;
// 4HH
return vec4(i >> 4u) / 255.0f;
#elif PS_IFMT == 2
// 4HL
return vec4(i & 0xFu) / 255.0f;
// 4HL
return vec4(i & 0xFu) / 255.0f;
#else
// Most of texture will hit this code so keep normalized float value
// Most of texture will hit this code so keep normalized float value
// 8 bits
return c;
// 8 bits
return c;
#endif
}
mat4 sample_4p(vec4 u)
{
mat4 c;
mat4 c;
c[0] = sample_p(u.x);
c[1] = sample_p(u.y);
c[2] = sample_p(u.z);
c[3] = sample_p(u.w);
c[0] = sample_p(u.x);
c[1] = sample_p(u.y);
c[2] = sample_p(u.z);
c[3] = sample_p(u.w);
return c;
return c;
}
vec4 sample_color(vec2 st, float q)
{
//FIXME: maybe we can set gl_Position.w = q in VS
//FIXME: maybe we can set gl_Position.w = q in VS
#if (PS_FST == 0)
st /= q;
st /= q;
#endif
#if (PS_TCOFFSETHACK == 1)
st += TC_OffsetHack.xy;
st += TC_OffsetHack.xy;
#endif
vec4 t;
mat4 c;
vec2 dd;
vec4 t;
mat4 c;
vec2 dd;
// FIXME I'm not sure this condition is useful (I think code will be optimized)
#if (PS_LTF == 0 && PS_FMT == FMT_32 && PS_WMS < 2 && PS_WMT < 2)
// No software LTF and pure 32 bits RGBA texure without special texture wrapping
c[0] = sample_c(st);
// No software LTF and pure 32 bits RGBA texure without special texture wrapping
c[0] = sample_c(st);
#ifdef TEX_COORD_DEBUG
c[0].rg = st.xy;
c[0].rg = st.xy;
#endif
#else
vec4 uv;
vec4 uv;
if(PS_LTF != 0)
{
uv = st.xyxy + HalfTexel;
dd = fract(uv.xy * WH.zw);
}
else
{
uv = st.xyxy;
}
if(PS_LTF != 0)
{
uv = st.xyxy + HalfTexel;
dd = fract(uv.xy * WH.zw);
}
else
{
uv = st.xyxy;
}
uv = clamp_wrap_uv(uv);
uv = clamp_wrap_uv(uv);
if((PS_FMT & FMT_PAL) != 0)
{
c = sample_4p(sample_4_index(uv));
}
else
{
c = sample_4c(uv);
}
if((PS_FMT & FMT_PAL) != 0)
{
c = sample_4p(sample_4_index(uv));
}
else
{
c = sample_4c(uv);
}
#ifdef TEX_COORD_DEBUG
c[0].rg = uv.xy;
c[1].rg = uv.xy;
c[2].rg = uv.xy;
c[3].rg = uv.xy;
c[0].rg = uv.xy;
c[1].rg = uv.xy;
c[2].rg = uv.xy;
c[3].rg = uv.xy;
#endif
#endif
// PERF: see the impact of the exansion before/after the interpolation
for (int i = 0; i < 4; i++)
{
// PERF: see the impact of the exansion before/after the interpolation
for (int i = 0; i < 4; i++)
{
// PERF note: using dot product reduces by 1 the number of instruction
// but I'm not sure it is equivalent neither faster.
//float sum = dot(c[i].rgb, vec3(1.0f));
#if ((PS_FMT & ~FMT_PAL) == FMT_24)
c[i].a = ( (PS_AEM == 0) || any(bvec3(c[i].rgb)) ) ? TA.x : 0.0f;
//c[i].a = ( (PS_AEM == 0) || (sum > 0.0f) ) ? TA.x : 0.0f;
c[i].a = ( (PS_AEM == 0) || any(bvec3(c[i].rgb)) ) ? TA.x : 0.0f;
//c[i].a = ( (PS_AEM == 0) || (sum > 0.0f) ) ? TA.x : 0.0f;
#elif ((PS_FMT & ~FMT_PAL) == FMT_16)
c[i].a = c[i].a >= 0.5 ? TA.y : ( (PS_AEM == 0) || any(bvec3(c[i].rgb)) ) ? TA.x : 0.0f;
//c[i].a = c[i].a >= 0.5 ? TA.y : ( (PS_AEM == 0) || (sum > 0.0f) ) ? TA.x : 0.0f;
c[i].a = c[i].a >= 0.5 ? TA.y : ( (PS_AEM == 0) || any(bvec3(c[i].rgb)) ) ? TA.x : 0.0f;
//c[i].a = c[i].a >= 0.5 ? TA.y : ( (PS_AEM == 0) || (sum > 0.0f) ) ? TA.x : 0.0f;
#endif
}
}
#if(PS_LTF != 0)
t = mix(mix(c[0], c[1], dd.x), mix(c[2], c[3], dd.x), dd.y);
t = mix(mix(c[0], c[1], dd.x), mix(c[2], c[3], dd.x), dd.y);
#else
t = c[0];
t = c[0];
#endif
// The 0.05f helps to fix the overbloom of sotc
// I think the issue is related to the rounding of texture coodinate. The linear (from fixed unit)
// interpolation could be slightly below the correct one.
return trunc(t * 255.0f + 0.05f);
// The 0.05f helps to fix the overbloom of sotc
// I think the issue is related to the rounding of texture coodinate. The linear (from fixed unit)
// interpolation could be slightly below the correct one.
return trunc(t * 255.0f + 0.05f);
}
vec4 tfx(vec4 T, vec4 C)
{
vec4 C_out;
vec4 FxT = trunc(trunc(C) * T / 128.0f);
vec4 C_out;
vec4 FxT = trunc(trunc(C) * T / 128.0f);
#if (PS_TFX == 0)
C_out = FxT;
C_out = FxT;
#elif (PS_TFX == 1)
C_out = T;
C_out = T;
#elif (PS_TFX == 2)
C_out.rgb = FxT.rgb + C.a;
C_out.a = T.a + C.a;
C_out.rgb = FxT.rgb + C.a;
C_out.a = T.a + C.a;
#elif (PS_TFX == 3)
C_out.rgb = FxT.rgb + C.a;
C_out.a = T.a;
C_out.rgb = FxT.rgb + C.a;
C_out.a = T.a;
#else
C_out = C;
C_out = C;
#endif
#if (PS_TCC == 0)
@ -297,96 +297,96 @@ vec4 tfx(vec4 T, vec4 C)
#endif
#if (PS_TFX == 0) || (PS_TFX == 2) || (PS_TFX == 3)
// Clamp only when it is useful
C_out = min(C_out, 255.0f);
// Clamp only when it is useful
C_out = min(C_out, 255.0f);
#endif
return C_out;
return C_out;
}
void atst(vec4 C)
{
// FIXME use integer cmp
float a = C.a;
// FIXME use integer cmp
float a = C.a;
#if (PS_ATST == 0) // never
discard;
discard;
#elif (PS_ATST == 1) // always
// nothing to do
// nothing to do
#elif (PS_ATST == 2) // l
if ((AREF - a - 0.5f) < 0.0f)
discard;
if ((AREF - a - 0.5f) < 0.0f)
discard;
#elif (PS_ATST == 3 ) // le
if ((AREF - a + 0.5f) < 0.0f)
discard;
if ((AREF - a + 0.5f) < 0.0f)
discard;
#elif (PS_ATST == 4) // e
if ((0.5f - abs(a - AREF)) < 0.0f)
discard;
if ((0.5f - abs(a - AREF)) < 0.0f)
discard;
#elif (PS_ATST == 5) // ge
if ((a-AREF + 0.5f) < 0.0f)
discard;
if ((a-AREF + 0.5f) < 0.0f)
discard;
#elif (PS_ATST == 6) // g
if ((a-AREF - 0.5f) < 0.0f)
discard;
if ((a-AREF - 0.5f) < 0.0f)
discard;
#elif (PS_ATST == 7) // ne
if ((abs(a - AREF) - 0.5f) < 0.0f)
discard;
if ((abs(a - AREF) - 0.5f) < 0.0f)
discard;
#endif
}
void fog(inout vec4 C, float f)
{
#if PS_FOG != 0
C.rgb = trunc(mix(FogColor, C.rgb, f));
C.rgb = trunc(mix(FogColor, C.rgb, f));
#endif
}
vec4 ps_color()
{
vec4 T = sample_color(PSin_t.xy, PSin_t.w);
vec4 T = sample_color(PSin_t.xy, PSin_t.w);
#if PS_IIP == 1
vec4 C = tfx(T, PSin_c);
vec4 C = tfx(T, PSin_c);
#else
vec4 C = tfx(T, PSin_fc);
vec4 C = tfx(T, PSin_fc);
#endif
atst(C);
atst(C);
fog(C, PSin_t.z);
fog(C, PSin_t.z);
#if (PS_CLR1 != 0) // needed for Cd * (As/Ad/F + 1) blending modes
C.rgb = vec3(255.0f);
C.rgb = vec3(255.0f);
#endif
return C;
return C;
}
void ps_fbmask(inout vec4 C)
{
// FIXME do I need special case for 16 bits
// FIXME do I need special case for 16 bits
#if PS_FBMASK
vec4 RT = trunc(texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0) * 255.0f + 0.1f);
C = vec4((uvec4(C) & ~FbMask) | (uvec4(RT) & FbMask));
vec4 RT = trunc(texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0) * 255.0f + 0.1f);
C = vec4((uvec4(C) & ~FbMask) | (uvec4(RT) & FbMask));
#endif
}
void ps_blend(inout vec4 Color, float As)
{
#if SW_BLEND
vec4 RT = trunc(texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0) * 255.0f + 0.1f);
vec4 RT = trunc(texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0) * 255.0f + 0.1f);
#if PS_DFMT == FMT_24
float Ad = 1.0f;
float Ad = 1.0f;
#else
// FIXME FMT_16 case
// FIXME Ad or Ad * 2?
float Ad = RT.a / 128.0f;
// FIXME FMT_16 case
// FIXME Ad or Ad * 2?
float Ad = RT.a / 128.0f;
#endif
// Let the compiler do its jobs !
vec3 Cd = RT.rgb;
vec3 Cs = Color.rgb;
// Let the compiler do its jobs !
vec3 Cd = RT.rgb;
vec3 Cs = Color.rgb;
#if PS_BLEND_A == 0
vec3 A = Cs;
@ -426,26 +426,26 @@ void ps_blend(inout vec4 Color, float As)
Color.rgb = trunc((A - B) * C + D);
#endif
// FIXME dithering
// FIXME dithering
// Correct the Color value based on the output format
// Correct the Color value based on the output format
#if PS_COLCLIP == 0 && PS_HDR == 0
// Standard Clamp
Color.rgb = clamp(Color.rgb, vec3(0.0f), vec3(255.0f));
// Standard Clamp
Color.rgb = clamp(Color.rgb, vec3(0.0f), vec3(255.0f));
#endif
// FIXME rouding of negative float?
// compiler uses trunc but it might need floor
// FIXME rouding of negative float?
// compiler uses trunc but it might need floor
// Warning: normally blending equation is mult(A, B) = A * B >> 7. GPU have the full accuracy
// GS: Color = 1, Alpha = 255 => output 1
// GPU: Color = 1/255, Alpha = 255/255 * 255/128 => output 1.9921875
#if PS_DFMT == FMT_16
// In 16 bits format, only 5 bits of colors are used. It impacts shadows computation of Castlevania
// In 16 bits format, only 5 bits of colors are used. It impacts shadows computation of Castlevania
Color.rgb = vec3(ivec3(Color.rgb) & ivec3(0xF8));
Color.rgb = vec3(ivec3(Color.rgb) & ivec3(0xF8));
#elif PS_COLCLIP == 1 && PS_HDR == 0
Color.rgb = vec3(ivec3(Color.rgb) & ivec3(0xFF));
Color.rgb = vec3(ivec3(Color.rgb) & ivec3(0xFF));
#endif
#endif
@ -456,141 +456,141 @@ void ps_main()
#if ((PS_DATE & 3) == 1 || (PS_DATE & 3) == 2) && !defined(DISABLE_GL42_image)
#if PS_WRITE_RG == 1
// Pseudo 16 bits access.
float rt_a = texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0).g;
// Pseudo 16 bits access.
float rt_a = texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0).g;
#else
float rt_a = texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0).a;
float rt_a = texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0).a;
#endif
#if (PS_DATE & 3) == 1
// DATM == 0: Pixel with alpha equal to 1 will failed
bool bad = (127.5f / 255.0f) < rt_a;
// DATM == 0: Pixel with alpha equal to 1 will failed
bool bad = (127.5f / 255.0f) < rt_a;
#elif (PS_DATE & 3) == 2
// DATM == 1: Pixel with alpha equal to 0 will failed
bool bad = rt_a < (127.5f / 255.0f);
// DATM == 1: Pixel with alpha equal to 0 will failed
bool bad = rt_a < (127.5f / 255.0f);
#endif
if (bad) {
if (bad) {
#if PS_DATE >= 5
discard;
discard;
#else
imageStore(img_prim_min, ivec2(gl_FragCoord.xy), ivec4(-1));
return;
imageStore(img_prim_min, ivec2(gl_FragCoord.xy), ivec4(-1));
return;
#endif
}
}
#endif
#if PS_DATE == 3 && !defined(DISABLE_GL42_image)
int stencil_ceil = imageLoad(img_prim_min, ivec2(gl_FragCoord.xy)).r;
// Note gl_PrimitiveID == stencil_ceil will be the primitive that will update
// the bad alpha value so we must keep it.
int stencil_ceil = imageLoad(img_prim_min, ivec2(gl_FragCoord.xy)).r;
// Note gl_PrimitiveID == stencil_ceil will be the primitive that will update
// the bad alpha value so we must keep it.
if (gl_PrimitiveID > stencil_ceil) {
discard;
}
if (gl_PrimitiveID > stencil_ceil) {
discard;
}
#endif
vec4 C = ps_color();
vec4 C = ps_color();
#if (APITRACE_DEBUG & 1) == 1
C.r = 255f;
C.r = 255f;
#endif
#if (APITRACE_DEBUG & 2) == 2
C.g = 255f;
C.g = 255f;
#endif
#if (APITRACE_DEBUG & 4) == 4
C.b = 255f;
C.b = 255f;
#endif
#if (APITRACE_DEBUG & 8) == 8
C.a = 128f;
C.a = 128f;
#endif
#if PS_SHUFFLE
uvec4 denorm_c = uvec4(C);
uvec2 denorm_TA = uvec2(vec2(TA.xy) * 255.0f + 0.5f);
uvec4 denorm_c = uvec4(C);
uvec2 denorm_TA = uvec2(vec2(TA.xy) * 255.0f + 0.5f);
// Write RB part. Mask will take care of the correct destination
// Write RB part. Mask will take care of the correct destination
#if PS_READ_BA
C.rb = C.bb;
C.rb = C.bb;
#else
C.rb = C.rr;
C.rb = C.rr;
#endif
// FIXME precompute my_TA & 0x80
// FIXME precompute my_TA & 0x80
// Write GA part. Mask will take care of the correct destination
// Note: GLSL 4.50/GL_EXT_shader_integer_mix support a mix instruction to select a component\n"
// However Nvidia emulate it with an if (at least on kepler arch) ...\n"
// Write GA part. Mask will take care of the correct destination
// Note: GLSL 4.50/GL_EXT_shader_integer_mix support a mix instruction to select a component\n"
// However Nvidia emulate it with an if (at least on kepler arch) ...\n"
#if PS_READ_BA
// bit field operation requires GL4 HW. Could be nice to merge it with step/mix below
// uint my_ta = (bool(bitfieldExtract(denorm_c.a, 7, 1))) ? denorm_TA.y : denorm_TA.x;
// denorm_c.a = bitfieldInsert(denorm_c.a, bitfieldExtract(my_ta, 7, 1), 7, 1);
// c.ga = vec2(float(denorm_c.a));
// bit field operation requires GL4 HW. Could be nice to merge it with step/mix below
// uint my_ta = (bool(bitfieldExtract(denorm_c.a, 7, 1))) ? denorm_TA.y : denorm_TA.x;
// denorm_c.a = bitfieldInsert(denorm_c.a, bitfieldExtract(my_ta, 7, 1), 7, 1);
// c.ga = vec2(float(denorm_c.a));
if (bool(denorm_c.a & 0x80u))
C.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.y & 0x80u)));
else
C.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.x & 0x80u)));
if (bool(denorm_c.a & 0x80u))
C.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.y & 0x80u)));
else
C.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.x & 0x80u)));
#else
if (bool(denorm_c.g & 0x80u))
C.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.y & 0x80u)));
else
C.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.x & 0x80u)));
if (bool(denorm_c.g & 0x80u))
C.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.y & 0x80u)));
else
C.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.x & 0x80u)));
// Nice idea but step/mix requires 4 instructions
// set / trunc / I2F / Mad
//
// float sel = step(128.0f, c.g);
// vec2 c_shuffle = vec2((denorm_c.gg & 0x7Fu) | (denorm_TA & 0x80u));
// c.ga = mix(c_shuffle.xx, c_shuffle.yy, sel);
// Nice idea but step/mix requires 4 instructions
// set / trunc / I2F / Mad
//
// float sel = step(128.0f, c.g);
// vec2 c_shuffle = vec2((denorm_c.gg & 0x7Fu) | (denorm_TA & 0x80u));
// c.ga = mix(c_shuffle.xx, c_shuffle.yy, sel);
#endif
#endif
// Must be done before alpha correction
float alpha_blend = C.a / 128.0f;
// Must be done before alpha correction
float alpha_blend = C.a / 128.0f;
// Correct the ALPHA value based on the output format
// Correct the ALPHA value based on the output format
#if (PS_DFMT == FMT_16)
float A_one = 128.0f; // alpha output will be 0x80
C.a = (PS_FBA != 0) ? A_one : step(128.0f, C.a) * A_one;
float A_one = 128.0f; // alpha output will be 0x80
C.a = (PS_FBA != 0) ? A_one : step(128.0f, C.a) * A_one;
#elif (PS_DFMT == FMT_32) && (PS_FBA != 0)
if(C.a < 128.0f) C.a += 128.0f;
if(C.a < 128.0f) C.a += 128.0f;
#endif
// Get first primitive that will write a failling alpha value
// Get first primitive that will write a failling alpha value
#if PS_DATE == 1 && !defined(DISABLE_GL42_image)
// DATM == 0
// Pixel with alpha equal to 1 will failed (128-255)
if (C.a > 127.5f) {
imageAtomicMin(img_prim_min, ivec2(gl_FragCoord.xy), gl_PrimitiveID);
return;
}
// DATM == 0
// Pixel with alpha equal to 1 will failed (128-255)
if (C.a > 127.5f) {
imageAtomicMin(img_prim_min, ivec2(gl_FragCoord.xy), gl_PrimitiveID);
return;
}
#elif PS_DATE == 2 && !defined(DISABLE_GL42_image)
// DATM == 1
// Pixel with alpha equal to 0 will failed (0-127)
if (C.a < 127.5f) {
imageAtomicMin(img_prim_min, ivec2(gl_FragCoord.xy), gl_PrimitiveID);
return;
}
// DATM == 1
// Pixel with alpha equal to 0 will failed (0-127)
if (C.a < 127.5f) {
imageAtomicMin(img_prim_min, ivec2(gl_FragCoord.xy), gl_PrimitiveID);
return;
}
#endif
ps_blend(C, alpha_blend);
ps_blend(C, alpha_blend);
ps_fbmask(C);
ps_fbmask(C);
#if PS_HDR == 1
// Use negative value to avoid overflow of the texture (in accumulation mode)
// Note: code were initially done for an Half-Float texture. Due to overflow
// the texture was upgraded to a full float. Maybe this code is useless now!
// Good testcase is castlevania
if (any(greaterThan(C.rgb, vec3(128.0f)))) {
C.rgb = (C.rgb - 256.0f);
}
// Use negative value to avoid overflow of the texture (in accumulation mode)
// Note: code were initially done for an Half-Float texture. Due to overflow
// the texture was upgraded to a full float. Maybe this code is useless now!
// Good testcase is castlevania
if (any(greaterThan(C.rgb, vec3(128.0f)))) {
C.rgb = (C.rgb - 256.0f);
}
#endif
SV_Target0 = C / 255.0f;
SV_Target1 = vec4(alpha_blend);
SV_Target0 = C / 255.0f;
SV_Target1 = vec4(alpha_blend);
}
#endif

26
plugins/GSdx/res/glsl/tfx_vgs.glsl
View File

@ -11,27 +11,27 @@ layout(std140, binding = 20) uniform cb20
// Warning duplicated in both GLSL file
layout(std140, binding = 21) uniform cb21
{
vec3 FogColor;
float AREF;
vec3 FogColor;
float AREF;
vec4 WH;
vec4 WH;
vec2 _pad0;
vec2 TA;
vec2 _pad0;
vec2 TA;
uvec4 MskFix;
uvec4 MskFix;
uvec4 FbMask;
uvec4 FbMask;
vec3 _pad1;
float Af;
vec3 _pad1;
float Af;
vec4 HalfTexel;
vec4 HalfTexel;
vec4 MinMax;
vec4 MinMax;
vec2 TextureScale;
vec2 TC_OffsetHack;
vec2 TextureScale;
vec2 TC_OffsetHack;
};
#ifdef VERTEX_SHADER

500
plugins/GSdx/res/glsl_source.h
View File

@ -617,27 +617,27 @@ static const char* tfx_vgs_glsl =
"// Warning duplicated in both GLSL file\n"
"layout(std140, binding = 21) uniform cb21\n"
"{\n"
" vec3 FogColor;\n"
" float AREF;\n"
" vec3 FogColor;\n"
" float AREF;\n"
"\n"
" vec4 WH;\n"
" vec4 WH;\n"
"\n"
" vec2 _pad0;\n"
" vec2 TA;\n"
" vec2 _pad0;\n"
" vec2 TA;\n"
"\n"
" uvec4 MskFix;\n"
" uvec4 MskFix;\n"
"\n"
" uvec4 FbMask;\n"
" uvec4 FbMask;\n"
"\n"
" vec3 _pad1;\n"
" float Af;\n"
" vec3 _pad1;\n"
" float Af;\n"
"\n"
" vec4 HalfTexel;\n"
" vec4 HalfTexel;\n"
"\n"
" vec4 MinMax;\n"
" vec4 MinMax;\n"
"\n"
" vec2 TextureScale;\n"
" vec2 TC_OffsetHack;\n"
" vec2 TextureScale;\n"
" vec2 TC_OffsetHack;\n"
"};\n"
"\n"
"#ifdef VERTEX_SHADER\n"
@ -911,9 +911,9 @@ static const char* tfx_fs_all_glsl =
"\n"
"in SHADER\n"
"{\n"
" vec4 t;\n"
" vec4 c;\n"
" flat vec4 fc;\n"
" vec4 t;\n"
" vec4 c;\n"
" flat vec4 fc;\n"
"} PSin;\n"
"\n"
"#define PSin_t (PSin.t)\n"
@ -947,234 +947,234 @@ static const char* tfx_fs_all_glsl =
"// Warning duplicated in both GLSL file\n"
"layout(std140, binding = 21) uniform cb21\n"
"{\n"
" vec3 FogColor;\n"
" float AREF;\n"
" vec3 FogColor;\n"
" float AREF;\n"
"\n"
" vec4 WH;\n"
" vec4 WH;\n"
"\n"
" vec2 _pad0;\n"
" vec2 TA;\n"
" vec2 _pad0;\n"
" vec2 TA;\n"
"\n"
" uvec4 MskFix;\n"
" uvec4 MskFix;\n"
"\n"
" uvec4 FbMask;\n"
" uvec4 FbMask;\n"
"\n"
" vec3 _pad1;\n"
" float Af;\n"
" vec3 _pad1;\n"
" float Af;\n"
"\n"
" vec4 HalfTexel;\n"
" vec4 HalfTexel;\n"
"\n"
" vec4 MinMax;\n"
" vec4 MinMax;\n"
"\n"
" vec2 TextureScale;\n"
" vec2 TC_OffsetHack;\n"
" vec2 TextureScale;\n"
" vec2 TC_OffsetHack;\n"
"};\n"
"\n"
"vec4 sample_c(vec2 uv)\n"
"{\n"
" return texture(TextureSampler, uv);\n"
" return texture(TextureSampler, uv);\n"
"}\n"
"\n"
"vec4 sample_p(float idx)\n"
"{\n"
" return texture(PaletteSampler, vec2(idx, 0.0f));\n"
" return texture(PaletteSampler, vec2(idx, 0.0f));\n"
"}\n"
"\n"
"vec4 clamp_wrap_uv(vec4 uv)\n"
"{\n"
" vec4 uv_out = uv;\n"
" vec4 uv_out = uv;\n"
"\n"
"#if PS_WMS == PS_WMT\n"
"\n"
"#if PS_WMS == 2\n"
" uv_out = clamp(uv, MinMax.xyxy, MinMax.zwzw);\n"
" uv_out = clamp(uv, MinMax.xyxy, MinMax.zwzw);\n"
"#elif PS_WMS == 3\n"
" uv_out = vec4((ivec4(uv * WH.xyxy) & ivec4(MskFix.xyxy)) | ivec4(MskFix.zwzw)) / WH.xyxy;\n"
" uv_out = vec4((ivec4(uv * WH.xyxy) & ivec4(MskFix.xyxy)) | ivec4(MskFix.zwzw)) / WH.xyxy;\n"
"#endif\n"
"\n"
"#else // PS_WMS != PS_WMT\n"
"\n"
"#if PS_WMS == 2\n"
" uv_out.xz = clamp(uv.xz, MinMax.xx, MinMax.zz);\n"
" uv_out.xz = clamp(uv.xz, MinMax.xx, MinMax.zz);\n"
"\n"
"#elif PS_WMS == 3\n"
" uv_out.xz = vec2((ivec2(uv.xz * WH.xx) & ivec2(MskFix.xx)) | ivec2(MskFix.zz)) / WH.xx;\n"
" uv_out.xz = vec2((ivec2(uv.xz * WH.xx) & ivec2(MskFix.xx)) | ivec2(MskFix.zz)) / WH.xx;\n"
"\n"
"#endif\n"
"\n"
"#if PS_WMT == 2\n"
" uv_out.yw = clamp(uv.yw, MinMax.yy, MinMax.ww);\n"
" uv_out.yw = clamp(uv.yw, MinMax.yy, MinMax.ww);\n"
"\n"
"#elif PS_WMT == 3\n"
"\n"
" uv_out.yw = vec2((ivec2(uv.yw * WH.yy) & ivec2(MskFix.yy)) | ivec2(MskFix.ww)) / WH.yy;\n"
" uv_out.yw = vec2((ivec2(uv.yw * WH.yy) & ivec2(MskFix.yy)) | ivec2(MskFix.ww)) / WH.yy;\n"
"#endif\n"
"\n"
"#endif\n"
"\n"
" return uv_out;\n"
" return uv_out;\n"
"}\n"
"\n"
"mat4 sample_4c(vec4 uv)\n"
"{\n"
" mat4 c;\n"
" mat4 c;\n"
"\n"
" // Note: texture gather can't be used because of special clamping/wrapping\n"
" // Also it doesn't support lod\n"
" c[0] = sample_c(uv.xy);\n"
" c[1] = sample_c(uv.zy);\n"
" c[2] = sample_c(uv.xw);\n"
" c[3] = sample_c(uv.zw);\n"
" c[0] = sample_c(uv.xy);\n"
" c[1] = sample_c(uv.zy);\n"
" c[2] = sample_c(uv.xw);\n"
" c[3] = sample_c(uv.zw);\n"
"\n"
" return c;\n"
" return c;\n"
"}\n"
"\n"
"vec4 sample_4_index(vec4 uv)\n"
"{\n"
" vec4 c;\n"
" vec4 c;\n"
"\n"
" // Either GSdx will send a texture that contains a single channel\n"
" // in this case the red channel is remapped as alpha channel\n"
" //\n"
" // Or we have an old RT (ie RGBA8) that contains index (4/8) in the alpha channel\n"
" // Either GSdx will send a texture that contains a single channel\n"
" // in this case the red channel is remapped as alpha channel\n"
" //\n"
" // Or we have an old RT (ie RGBA8) that contains index (4/8) in the alpha channel\n"
"\n"
" // Note: texture gather can't be used because of special clamping/wrapping\n"
" // Also it doesn't support lod\n"
" c.x = sample_c(uv.xy).a;\n"
" c.y = sample_c(uv.zy).a;\n"
" c.z = sample_c(uv.xw).a;\n"
" c.w = sample_c(uv.zw).a;\n"
" c.x = sample_c(uv.xy).a;\n"
" c.y = sample_c(uv.zy).a;\n"
" c.z = sample_c(uv.xw).a;\n"
" c.w = sample_c(uv.zw).a;\n"
"\n"
" uvec4 i = uvec4(c * 255.0f + 0.5f); // Denormalize value\n"
" uvec4 i = uvec4(c * 255.0f + 0.5f); // Denormalize value\n"
"\n"
"#if PS_IFMT == 1\n"
" // 4HH\n"
" return vec4(i >> 4u) / 255.0f;\n"
" // 4HH\n"
" return vec4(i >> 4u) / 255.0f;\n"
"\n"
"#elif PS_IFMT == 2\n"
" // 4HL\n"
" return vec4(i & 0xFu) / 255.0f;\n"
" // 4HL\n"
" return vec4(i & 0xFu) / 255.0f;\n"
"\n"
"#else\n"
" // Most of texture will hit this code so keep normalized float value\n"
" // Most of texture will hit this code so keep normalized float value\n"
"\n"
" // 8 bits\n"
" return c;\n"
" // 8 bits\n"
" return c;\n"
"#endif\n"
"\n"
"}\n"
"\n"
"mat4 sample_4p(vec4 u)\n"
"{\n"
" mat4 c;\n"
" mat4 c;\n"
"\n"
" c[0] = sample_p(u.x);\n"
" c[1] = sample_p(u.y);\n"
" c[2] = sample_p(u.z);\n"
" c[3] = sample_p(u.w);\n"
" c[0] = sample_p(u.x);\n"
" c[1] = sample_p(u.y);\n"
" c[2] = sample_p(u.z);\n"
" c[3] = sample_p(u.w);\n"
"\n"
" return c;\n"
" return c;\n"
"}\n"
"\n"
"vec4 sample_color(vec2 st, float q)\n"
"{\n"
" //FIXME: maybe we can set gl_Position.w = q in VS\n"
" //FIXME: maybe we can set gl_Position.w = q in VS\n"
"#if (PS_FST == 0)\n"
" st /= q;\n"
" st /= q;\n"
"#endif\n"
"\n"
"#if (PS_TCOFFSETHACK == 1)\n"
" st += TC_OffsetHack.xy;\n"
" st += TC_OffsetHack.xy;\n"
"#endif\n"
"\n"
" vec4 t;\n"
" mat4 c;\n"
" vec2 dd;\n"
" vec4 t;\n"
" mat4 c;\n"
" vec2 dd;\n"
"\n"
" // FIXME I'm not sure this condition is useful (I think code will be optimized)\n"
"#if (PS_LTF == 0 && PS_FMT == FMT_32 && PS_WMS < 2 && PS_WMT < 2)\n"
" // No software LTF and pure 32 bits RGBA texure without special texture wrapping\n"
" c[0] = sample_c(st);\n"
" // No software LTF and pure 32 bits RGBA texure without special texture wrapping\n"
" c[0] = sample_c(st);\n"
"#ifdef TEX_COORD_DEBUG\n"
" c[0].rg = st.xy;\n"
" c[0].rg = st.xy;\n"
"#endif\n"
"\n"
"#else\n"
" vec4 uv;\n"
" vec4 uv;\n"
"\n"
" if(PS_LTF != 0)\n"
" {\n"
" uv = st.xyxy + HalfTexel;\n"
" dd = fract(uv.xy * WH.zw);\n"
" }\n"
" else\n"
" {\n"
" uv = st.xyxy;\n"
" }\n"
" if(PS_LTF != 0)\n"
" {\n"
" uv = st.xyxy + HalfTexel;\n"
" dd = fract(uv.xy * WH.zw);\n"
" }\n"
" else\n"
" {\n"
" uv = st.xyxy;\n"
" }\n"
"\n"
" uv = clamp_wrap_uv(uv);\n"
" uv = clamp_wrap_uv(uv);\n"
"\n"
" if((PS_FMT & FMT_PAL) != 0)\n"
" {\n"
" c = sample_4p(sample_4_index(uv));\n"
" }\n"
" else\n"
" {\n"
" c = sample_4c(uv);\n"
" }\n"
" if((PS_FMT & FMT_PAL) != 0)\n"
" {\n"
" c = sample_4p(sample_4_index(uv));\n"
" }\n"
" else\n"
" {\n"
" c = sample_4c(uv);\n"
" }\n"
"#ifdef TEX_COORD_DEBUG\n"
" c[0].rg = uv.xy;\n"
" c[1].rg = uv.xy;\n"
" c[2].rg = uv.xy;\n"
" c[3].rg = uv.xy;\n"
" c[0].rg = uv.xy;\n"
" c[1].rg = uv.xy;\n"
" c[2].rg = uv.xy;\n"
" c[3].rg = uv.xy;\n"
"#endif\n"
"\n"
"#endif\n"
"\n"
" // PERF: see the impact of the exansion before/after the interpolation\n"
" for (int i = 0; i < 4; i++)\n"
" {\n"
" // PERF: see the impact of the exansion before/after the interpolation\n"
" for (int i = 0; i < 4; i++)\n"
" {\n"
" // PERF note: using dot product reduces by 1 the number of instruction\n"
" // but I'm not sure it is equivalent neither faster.\n"
" //float sum = dot(c[i].rgb, vec3(1.0f));\n"
"#if ((PS_FMT & ~FMT_PAL) == FMT_24)\n"
" c[i].a = ( (PS_AEM == 0) || any(bvec3(c[i].rgb)) ) ? TA.x : 0.0f;\n"
" //c[i].a = ( (PS_AEM == 0) || (sum > 0.0f) ) ? TA.x : 0.0f;\n"
" c[i].a = ( (PS_AEM == 0) || any(bvec3(c[i].rgb)) ) ? TA.x : 0.0f;\n"
" //c[i].a = ( (PS_AEM == 0) || (sum > 0.0f) ) ? TA.x : 0.0f;\n"
"#elif ((PS_FMT & ~FMT_PAL) == FMT_16)\n"
" c[i].a = c[i].a >= 0.5 ? TA.y : ( (PS_AEM == 0) || any(bvec3(c[i].rgb)) ) ? TA.x : 0.0f;\n"
" //c[i].a = c[i].a >= 0.5 ? TA.y : ( (PS_AEM == 0) || (sum > 0.0f) ) ? TA.x : 0.0f;\n"
" c[i].a = c[i].a >= 0.5 ? TA.y : ( (PS_AEM == 0) || any(bvec3(c[i].rgb)) ) ? TA.x : 0.0f;\n"
" //c[i].a = c[i].a >= 0.5 ? TA.y : ( (PS_AEM == 0) || (sum > 0.0f) ) ? TA.x : 0.0f;\n"
"#endif\n"
" }\n"
" }\n"
"\n"
"#if(PS_LTF != 0)\n"
" t = mix(mix(c[0], c[1], dd.x), mix(c[2], c[3], dd.x), dd.y);\n"
" t = mix(mix(c[0], c[1], dd.x), mix(c[2], c[3], dd.x), dd.y);\n"
"#else\n"
" t = c[0];\n"
" t = c[0];\n"
"#endif\n"
"\n"
" // The 0.05f helps to fix the overbloom of sotc\n"
" // I think the issue is related to the rounding of texture coodinate. The linear (from fixed unit)\n"
" // interpolation could be slightly below the correct one.\n"
" return trunc(t * 255.0f + 0.05f);\n"
" // The 0.05f helps to fix the overbloom of sotc\n"
" // I think the issue is related to the rounding of texture coodinate. The linear (from fixed unit)\n"
" // interpolation could be slightly below the correct one.\n"
" return trunc(t * 255.0f + 0.05f);\n"
"}\n"
"\n"
"vec4 tfx(vec4 T, vec4 C)\n"
"{\n"
" vec4 C_out;\n"
" vec4 FxT = trunc(trunc(C) * T / 128.0f);\n"
" vec4 C_out;\n"
" vec4 FxT = trunc(trunc(C) * T / 128.0f);\n"
"\n"
"#if (PS_TFX == 0)\n"
" C_out = FxT;\n"
" C_out = FxT;\n"
"#elif (PS_TFX == 1)\n"
" C_out = T;\n"
" C_out = T;\n"
"#elif (PS_TFX == 2)\n"
" C_out.rgb = FxT.rgb + C.a;\n"
" C_out.a = T.a + C.a;\n"
" C_out.rgb = FxT.rgb + C.a;\n"
" C_out.a = T.a + C.a;\n"
"#elif (PS_TFX == 3)\n"
" C_out.rgb = FxT.rgb + C.a;\n"
" C_out.a = T.a;\n"
" C_out.rgb = FxT.rgb + C.a;\n"
" C_out.a = T.a;\n"
"#else\n"
" C_out = C;\n"
" C_out = C;\n"
"#endif\n"
"\n"
"#if (PS_TCC == 0)\n"
@ -1182,96 +1182,96 @@ static const char* tfx_fs_all_glsl =
"#endif\n"
"\n"
"#if (PS_TFX == 0) || (PS_TFX == 2) || (PS_TFX == 3)\n"
" // Clamp only when it is useful\n"
" C_out = min(C_out, 255.0f);\n"
" // Clamp only when it is useful\n"
" C_out = min(C_out, 255.0f);\n"
"#endif\n"
"\n"
" return C_out;\n"
" return C_out;\n"
"}\n"
"\n"
"void atst(vec4 C)\n"
"{\n"
" // FIXME use integer cmp\n"
" float a = C.a;\n"
" // FIXME use integer cmp\n"
" float a = C.a;\n"
"\n"
"#if (PS_ATST == 0) // never\n"
" discard;\n"
" discard;\n"
"#elif (PS_ATST == 1) // always\n"
" // nothing to do\n"
" // nothing to do\n"
"#elif (PS_ATST == 2) // l\n"
" if ((AREF - a - 0.5f) < 0.0f)\n"
" discard;\n"
" if ((AREF - a - 0.5f) < 0.0f)\n"
" discard;\n"
"#elif (PS_ATST == 3 ) // le\n"
" if ((AREF - a + 0.5f) < 0.0f)\n"
" discard;\n"
" if ((AREF - a + 0.5f) < 0.0f)\n"
" discard;\n"
"#elif (PS_ATST == 4) // e\n"
" if ((0.5f - abs(a - AREF)) < 0.0f)\n"
" discard;\n"
" if ((0.5f - abs(a - AREF)) < 0.0f)\n"
" discard;\n"
"#elif (PS_ATST == 5) // ge\n"
" if ((a-AREF + 0.5f) < 0.0f)\n"
" discard;\n"
" if ((a-AREF + 0.5f) < 0.0f)\n"
" discard;\n"
"#elif (PS_ATST == 6) // g\n"
" if ((a-AREF - 0.5f) < 0.0f)\n"
" discard;\n"
" if ((a-AREF - 0.5f) < 0.0f)\n"
" discard;\n"
"#elif (PS_ATST == 7) // ne\n"
" if ((abs(a - AREF) - 0.5f) < 0.0f)\n"
" discard;\n"
" if ((abs(a - AREF) - 0.5f) < 0.0f)\n"
" discard;\n"
"#endif\n"
"}\n"
"\n"
"void fog(inout vec4 C, float f)\n"
"{\n"
"#if PS_FOG != 0\n"
" C.rgb = trunc(mix(FogColor, C.rgb, f));\n"
" C.rgb = trunc(mix(FogColor, C.rgb, f));\n"
"#endif\n"
"}\n"
"\n"
"vec4 ps_color()\n"
"{\n"
" vec4 T = sample_color(PSin_t.xy, PSin_t.w);\n"
" vec4 T = sample_color(PSin_t.xy, PSin_t.w);\n"
"\n"
"#if PS_IIP == 1\n"
" vec4 C = tfx(T, PSin_c);\n"
" vec4 C = tfx(T, PSin_c);\n"
"#else\n"
" vec4 C = tfx(T, PSin_fc);\n"
" vec4 C = tfx(T, PSin_fc);\n"
"#endif\n"
"\n"
" atst(C);\n"
" atst(C);\n"
"\n"
" fog(C, PSin_t.z);\n"
" fog(C, PSin_t.z);\n"
"\n"
"#if (PS_CLR1 != 0) // needed for Cd * (As/Ad/F + 1) blending modes\n"
" C.rgb = vec3(255.0f);\n"
" C.rgb = vec3(255.0f);\n"
"#endif\n"
"\n"
" return C;\n"
" return C;\n"
"}\n"
"\n"
"void ps_fbmask(inout vec4 C)\n"
"{\n"
" // FIXME do I need special case for 16 bits\n"
" // FIXME do I need special case for 16 bits\n"
"#if PS_FBMASK\n"
" vec4 RT = trunc(texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0) * 255.0f + 0.1f);\n"
" C = vec4((uvec4(C) & ~FbMask) | (uvec4(RT) & FbMask));\n"
" vec4 RT = trunc(texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0) * 255.0f + 0.1f);\n"
" C = vec4((uvec4(C) & ~FbMask) | (uvec4(RT) & FbMask));\n"
"#endif\n"
"}\n"
"\n"
"void ps_blend(inout vec4 Color, float As)\n"
"{\n"
"#if SW_BLEND\n"
" vec4 RT = trunc(texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0) * 255.0f + 0.1f);\n"
" vec4 RT = trunc(texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0) * 255.0f + 0.1f);\n"
"\n"
"#if PS_DFMT == FMT_24\n"
" float Ad = 1.0f;\n"
" float Ad = 1.0f;\n"
"#else\n"
" // FIXME FMT_16 case\n"
" // FIXME Ad or Ad * 2?\n"
" float Ad = RT.a / 128.0f;\n"
" // FIXME FMT_16 case\n"
" // FIXME Ad or Ad * 2?\n"
" float Ad = RT.a / 128.0f;\n"
"#endif\n"
"\n"
" // Let the compiler do its jobs !\n"
" vec3 Cd = RT.rgb;\n"
" vec3 Cs = Color.rgb;\n"
" // Let the compiler do its jobs !\n"
" vec3 Cd = RT.rgb;\n"
" vec3 Cs = Color.rgb;\n"
"\n"
"#if PS_BLEND_A == 0\n"
" vec3 A = Cs;\n"
@ -1311,26 +1311,26 @@ static const char* tfx_fs_all_glsl =
" Color.rgb = trunc((A - B) * C + D);\n"
"#endif\n"
"\n"
" // FIXME dithering\n"
" // FIXME dithering\n"
"\n"
" // Correct the Color value based on the output format\n"
" // Correct the Color value based on the output format\n"
"#if PS_COLCLIP == 0 && PS_HDR == 0\n"
" // Standard Clamp\n"
" Color.rgb = clamp(Color.rgb, vec3(0.0f), vec3(255.0f));\n"
" // Standard Clamp\n"
" Color.rgb = clamp(Color.rgb, vec3(0.0f), vec3(255.0f));\n"
"#endif\n"
"\n"
" // FIXME rouding of negative float?\n"
" // compiler uses trunc but it might need floor\n"
" // FIXME rouding of negative float?\n"
" // compiler uses trunc but it might need floor\n"
"\n"
" // Warning: normally blending equation is mult(A, B) = A * B >> 7. GPU have the full accuracy\n"
" // GS: Color = 1, Alpha = 255 => output 1\n"
" // GPU: Color = 1/255, Alpha = 255/255 * 255/128 => output 1.9921875\n"
"#if PS_DFMT == FMT_16\n"
" // In 16 bits format, only 5 bits of colors are used. It impacts shadows computation of Castlevania\n"
" // In 16 bits format, only 5 bits of colors are used. It impacts shadows computation of Castlevania\n"
"\n"
" Color.rgb = vec3(ivec3(Color.rgb) & ivec3(0xF8));\n"
" Color.rgb = vec3(ivec3(Color.rgb) & ivec3(0xF8));\n"
"#elif PS_COLCLIP == 1 && PS_HDR == 0\n"
" Color.rgb = vec3(ivec3(Color.rgb) & ivec3(0xFF));\n"
" Color.rgb = vec3(ivec3(Color.rgb) & ivec3(0xFF));\n"
"#endif\n"
"\n"
"#endif\n"
@ -1341,141 +1341,141 @@ static const char* tfx_fs_all_glsl =
"#if ((PS_DATE & 3) == 1 || (PS_DATE & 3) == 2) && !defined(DISABLE_GL42_image)\n"
"\n"
"#if PS_WRITE_RG == 1\n"
" // Pseudo 16 bits access.\n"
" float rt_a = texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0).g;\n"
" // Pseudo 16 bits access.\n"
" float rt_a = texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0).g;\n"
"#else\n"
" float rt_a = texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0).a;\n"
" float rt_a = texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0).a;\n"
"#endif\n"
"\n"
"#if (PS_DATE & 3) == 1\n"
" // DATM == 0: Pixel with alpha equal to 1 will failed\n"
" bool bad = (127.5f / 255.0f) < rt_a;\n"
" // DATM == 0: Pixel with alpha equal to 1 will failed\n"
" bool bad = (127.5f / 255.0f) < rt_a;\n"
"#elif (PS_DATE & 3) == 2\n"
" // DATM == 1: Pixel with alpha equal to 0 will failed\n"
" bool bad = rt_a < (127.5f / 255.0f);\n"
" // DATM == 1: Pixel with alpha equal to 0 will failed\n"
" bool bad = rt_a < (127.5f / 255.0f);\n"
"#endif\n"
"\n"
" if (bad) {\n"
" if (bad) {\n"
"#if PS_DATE >= 5\n"
" discard;\n"
" discard;\n"
"#else\n"
" imageStore(img_prim_min, ivec2(gl_FragCoord.xy), ivec4(-1));\n"
" return;\n"
" imageStore(img_prim_min, ivec2(gl_FragCoord.xy), ivec4(-1));\n"
" return;\n"
"#endif\n"
" }\n"
" }\n"
"\n"
"#endif\n"
"\n"
"#if PS_DATE == 3 && !defined(DISABLE_GL42_image)\n"
" int stencil_ceil = imageLoad(img_prim_min, ivec2(gl_FragCoord.xy)).r;\n"
" // Note gl_PrimitiveID == stencil_ceil will be the primitive that will update\n"
" // the bad alpha value so we must keep it.\n"
" int stencil_ceil = imageLoad(img_prim_min, ivec2(gl_FragCoord.xy)).r;\n"
" // Note gl_PrimitiveID == stencil_ceil will be the primitive that will update\n"
" // the bad alpha value so we must keep it.\n"
"\n"
" if (gl_PrimitiveID > stencil_ceil) {\n"
" discard;\n"
" }\n"
" if (gl_PrimitiveID > stencil_ceil) {\n"
" discard;\n"
" }\n"
"#endif\n"
"\n"
" vec4 C = ps_color();\n"
" vec4 C = ps_color();\n"
"#if (APITRACE_DEBUG & 1) == 1\n"
" C.r = 255f;\n"
" C.r = 255f;\n"
"#endif\n"
"#if (APITRACE_DEBUG & 2) == 2\n"
" C.g = 255f;\n"
" C.g = 255f;\n"
"#endif\n"
"#if (APITRACE_DEBUG & 4) == 4\n"
" C.b = 255f;\n"
" C.b = 255f;\n"
"#endif\n"
"#if (APITRACE_DEBUG & 8) == 8\n"
" C.a = 128f;\n"
" C.a = 128f;\n"
"#endif\n"
"\n"
"#if PS_SHUFFLE\n"
" uvec4 denorm_c = uvec4(C);\n"
" uvec2 denorm_TA = uvec2(vec2(TA.xy) * 255.0f + 0.5f);\n"
" uvec4 denorm_c = uvec4(C);\n"
" uvec2 denorm_TA = uvec2(vec2(TA.xy) * 255.0f + 0.5f);\n"
"\n"
" // Write RB part. Mask will take care of the correct destination\n"
" // Write RB part. Mask will take care of the correct destination\n"
"#if PS_READ_BA\n"
" C.rb = C.bb;\n"
" C.rb = C.bb;\n"
"#else\n"
" C.rb = C.rr;\n"
" C.rb = C.rr;\n"
"#endif\n"
"\n"
" // FIXME precompute my_TA & 0x80\n"
" // FIXME precompute my_TA & 0x80\n"
"\n"
" // Write GA part. Mask will take care of the correct destination\n"
" // Note: GLSL 4.50/GL_EXT_shader_integer_mix support a mix instruction to select a component\\n\"\n"
" // However Nvidia emulate it with an if (at least on kepler arch) ...\\n\"\n"
" // Write GA part. Mask will take care of the correct destination\n"
" // Note: GLSL 4.50/GL_EXT_shader_integer_mix support a mix instruction to select a component\\n\"\n"
" // However Nvidia emulate it with an if (at least on kepler arch) ...\\n\"\n"
"#if PS_READ_BA\n"
" // bit field operation requires GL4 HW. Could be nice to merge it with step/mix below\n"
" // uint my_ta = (bool(bitfieldExtract(denorm_c.a, 7, 1))) ? denorm_TA.y : denorm_TA.x;\n"
" // denorm_c.a = bitfieldInsert(denorm_c.a, bitfieldExtract(my_ta, 7, 1), 7, 1);\n"
" // c.ga = vec2(float(denorm_c.a));\n"
" // bit field operation requires GL4 HW. Could be nice to merge it with step/mix below\n"
" // uint my_ta = (bool(bitfieldExtract(denorm_c.a, 7, 1))) ? denorm_TA.y : denorm_TA.x;\n"
" // denorm_c.a = bitfieldInsert(denorm_c.a, bitfieldExtract(my_ta, 7, 1), 7, 1);\n"
" // c.ga = vec2(float(denorm_c.a));\n"
"\n"
" if (bool(denorm_c.a & 0x80u))\n"
" C.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.y & 0x80u)));\n"
" else\n"
" C.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.x & 0x80u)));\n"
" if (bool(denorm_c.a & 0x80u))\n"
" C.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.y & 0x80u)));\n"
" else\n"
" C.ga = vec2(float((denorm_c.a & 0x7Fu) | (denorm_TA.x & 0x80u)));\n"
"\n"
"#else\n"
" if (bool(denorm_c.g & 0x80u))\n"
" C.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.y & 0x80u)));\n"
" else\n"
" C.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.x & 0x80u)));\n"
" if (bool(denorm_c.g & 0x80u))\n"
" C.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.y & 0x80u)));\n"
" else\n"
" C.ga = vec2(float((denorm_c.g & 0x7Fu) | (denorm_TA.x & 0x80u)));\n"
"\n"
" // Nice idea but step/mix requires 4 instructions\n"
" // set / trunc / I2F / Mad\n"
" //\n"
" // float sel = step(128.0f, c.g);\n"
" // vec2 c_shuffle = vec2((denorm_c.gg & 0x7Fu) | (denorm_TA & 0x80u));\n"
" // c.ga = mix(c_shuffle.xx, c_shuffle.yy, sel);\n"
" // Nice idea but step/mix requires 4 instructions\n"
" // set / trunc / I2F / Mad\n"
" //\n"
" // float sel = step(128.0f, c.g);\n"
" // vec2 c_shuffle = vec2((denorm_c.gg & 0x7Fu) | (denorm_TA & 0x80u));\n"
" // c.ga = mix(c_shuffle.xx, c_shuffle.yy, sel);\n"
"#endif\n"
"\n"
"#endif\n"
"\n"
" // Must be done before alpha correction\n"
" float alpha_blend = C.a / 128.0f;\n"
" // Must be done before alpha correction\n"
" float alpha_blend = C.a / 128.0f;\n"
"\n"
" // Correct the ALPHA value based on the output format\n"
" // Correct the ALPHA value based on the output format\n"
"#if (PS_DFMT == FMT_16)\n"
" float A_one = 128.0f; // alpha output will be 0x80\n"
" C.a = (PS_FBA != 0) ? A_one : step(128.0f, C.a) * A_one;\n"
" float A_one = 128.0f; // alpha output will be 0x80\n"
" C.a = (PS_FBA != 0) ? A_one : step(128.0f, C.a) * A_one;\n"
"#elif (PS_DFMT == FMT_32) && (PS_FBA != 0)\n"
" if(C.a < 128.0f) C.a += 128.0f;\n"
" if(C.a < 128.0f) C.a += 128.0f;\n"
"#endif\n"
"\n"
" // Get first primitive that will write a failling alpha value\n"
" // Get first primitive that will write a failling alpha value\n"
"#if PS_DATE == 1 && !defined(DISABLE_GL42_image)\n"
" // DATM == 0\n"
" // Pixel with alpha equal to 1 will failed (128-255)\n"
" if (C.a > 127.5f) {\n"
" imageAtomicMin(img_prim_min, ivec2(gl_FragCoord.xy), gl_PrimitiveID);\n"
" return;\n"
" }\n"
" // DATM == 0\n"
" // Pixel with alpha equal to 1 will failed (128-255)\n"
" if (C.a > 127.5f) {\n"
" imageAtomicMin(img_prim_min, ivec2(gl_FragCoord.xy), gl_PrimitiveID);\n"
" return;\n"
" }\n"
"#elif PS_DATE == 2 && !defined(DISABLE_GL42_image)\n"
" // DATM == 1\n"
" // Pixel with alpha equal to 0 will failed (0-127)\n"
" if (C.a < 127.5f) {\n"
" imageAtomicMin(img_prim_min, ivec2(gl_FragCoord.xy), gl_PrimitiveID);\n"
" return;\n"
" }\n"
" // DATM == 1\n"
" // Pixel with alpha equal to 0 will failed (0-127)\n"
" if (C.a < 127.5f) {\n"
" imageAtomicMin(img_prim_min, ivec2(gl_FragCoord.xy), gl_PrimitiveID);\n"
" return;\n"
" }\n"
"#endif\n"
"\n"
" ps_blend(C, alpha_blend);\n"
" ps_blend(C, alpha_blend);\n"
"\n"
" ps_fbmask(C);\n"
" ps_fbmask(C);\n"
"\n"
"#if PS_HDR == 1\n"
" // Use negative value to avoid overflow of the texture (in accumulation mode)\n"
" // Note: code were initially done for an Half-Float texture. Due to overflow\n"
" // the texture was upgraded to a full float. Maybe this code is useless now!\n"
" // Good testcase is castlevania\n"
" if (any(greaterThan(C.rgb, vec3(128.0f)))) {\n"
" C.rgb = (C.rgb - 256.0f);\n"
" }\n"
" // Use negative value to avoid overflow of the texture (in accumulation mode)\n"
" // Note: code were initially done for an Half-Float texture. Due to overflow\n"
" // the texture was upgraded to a full float. Maybe this code is useless now!\n"
" // Good testcase is castlevania\n"
" if (any(greaterThan(C.rgb, vec3(128.0f)))) {\n"
" C.rgb = (C.rgb - 256.0f);\n"
" }\n"
"#endif\n"
" SV_Target0 = C / 255.0f;\n"
" SV_Target1 = vec4(alpha_blend);\n"
" SV_Target0 = C / 255.0f;\n"
" SV_Target1 = vec4(alpha_blend);\n"
"}\n"
"\n"
"#endif\n"