ShuriZma
/
duckstation
mirror of https://github.com/stenzek/duckstation
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

GPU/HW: Truncate colours to 16-bit before applying alpha 

Fixes battle screen transitions blowing out in brightness in Final
Fantasy 7.
This commit is contained in:
Connor McLaughlin 2020-04-24 03:26:52 +10:00
parent 4a4c3d5e3b
commit 715209e178
1 changed files with 19 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
src/core/gpu_hw_shadergen.cpp
View File

@ -637,13 +637,22 @@ float4 SampleFromVRAM(uint4 texpage, uint2 icoord)
#endif #endif
#endif #endif
// Clip to 15-bit range // Premultiply alpha so we don't need to use a colour output for it.
#if !TRUE_COLOR float premultiply_alpha = ialpha;
icolor = TruncateTo15Bit(icolor); #if TRANSPARENCY
premultiply_alpha = ialpha * (semitransparent ? u_src_alpha_factor : 1.0);
#endif #endif
// Normalize float3 color;
float3 color = float3(icolor) / float3(255.0, 255.0, 255.0); #if !TRUE_COLOR
// We want to apply the alpha before the truncation to 16-bit, otherwise we'll be passing a 32-bit precision color
// into the blend unit, which can cause a small amount of error to accumulate.
icolor = int3(((float3(icolor) / float3(255.0, 255.0, 255.0)) * premultiply_alpha) * float3(255.0, 255.0, 255.0));
color = (float3(icolor >> 3) / float3(31.0, 31.0, 31.0));
#else
// True color is actually simpler here since we want to preserve the precision.
color = (float3(icolor) / float3(255.0, 255.0, 255.0)) * premultiply_alpha;
#endif
#if TRANSPARENCY #if TRANSPARENCY
// Apply semitransparency. If not a semitransparent texel, destination alpha is ignored. // Apply semitransparency. If not a semitransparent texel, destination alpha is ignored.
@ -654,10 +663,10 @@ float4 SampleFromVRAM(uint4 texpage, uint2 icoord)
#endif #endif
#if USE_DUAL_SOURCE #if USE_DUAL_SOURCE
o_col0 = float4(color * (u_src_alpha_factor * ialpha), oalpha); o_col0 = float4(color, oalpha);
o_col1 = float4(0.0, 0.0, 0.0, u_dst_alpha_factor / ialpha); o_col1 = float4(0.0, 0.0, 0.0, u_dst_alpha_factor / ialpha);
#else #else
o_col0 = float4(color * (u_src_alpha_factor * ialpha), u_dst_alpha_factor / ialpha); o_col0 = float4(color, u_dst_alpha_factor / ialpha);
#endif #endif
} }
else else
@ -667,15 +676,15 @@ float4 SampleFromVRAM(uint4 texpage, uint2 icoord)
#endif #endif
#if USE_DUAL_SOURCE #if USE_DUAL_SOURCE
o_col0 = float4(color * ialpha, oalpha); o_col0 = float4(color, oalpha);
o_col1 = float4(0.0, 0.0, 0.0, 0.0); o_col1 = float4(0.0, 0.0, 0.0, 0.0);
#else #else
o_col0 = float4(color * ialpha, 1.0 - ialpha); o_col0 = float4(color, 1.0 - ialpha);
#endif #endif
} }
#else #else
// Non-transparency won't enable blending so we can write the mask here regardless. // Non-transparency won't enable blending so we can write the mask here regardless.
o_col0 = float4(color * ialpha, oalpha); o_col0 = float4(color, oalpha);
#if USE_DUAL_SOURCE #if USE_DUAL_SOURCE
o_col1 = float4(0.0, 0.0, 0.0, 1.0 - ialpha); o_col1 = float4(0.0, 0.0, 0.0, 1.0 - ialpha);