GS/HW: Improve Round Sprite upscaling fix to cause less problems

This commit is contained in:
refractionpcsx2 2023-11-15 18:44:44 +00:00 committed by lightningterror
parent 671255c684
commit 5f4d2b3905
1 changed files with 122 additions and 82 deletions

View File

@ -1859,42 +1859,136 @@ void GSRendererHW::RoundSpriteOffset()
for (u32 i = 0; i < count; i += 2)
{
// Performance note: if it had any impact on perf, someone would port it to SSE (AKA GSVector)
// Compute the coordinate of first and last texels (in native with a linear filtering)
// if the draw is page aligned, then don't round it.
const int tex_width = std::max(1, (v[i + 1].U - v[i].U) >> 4);
const int ox = m_context->XYOFFSET.OFX;
const int X0 = v[i].XYZ.X - ox;
const int X1 = v[i + 1].XYZ.X - ox;
const int Lx = (v[i + 1].XYZ.X - v[i].XYZ.X);
const float ax0 = alpha0(Lx, X0, X1);
const float ax1 = alpha1(Lx, X0, X1);
const u16 tx0 = Interpolate_UV(ax0, v[i].U, v[i + 1].U);
const u16 tx1 = Interpolate_UV(ax1, v[i].U, v[i + 1].U);
if ((((Lx - ox) >> 4) % tex_width) != 0)
{
// Compute the coordinate of first and last texels (in native with a linear filtering)
const int X0 = v[i].XYZ.X - ox;
const int X1 = v[i + 1].XYZ.X - ox;
const float ax0 = alpha0(Lx, X0, X1);
const float ax1 = alpha1(Lx, X0, X1);
const u16 tx0 = Interpolate_UV(ax0, v[i].U, v[i + 1].U);
const u16 tx1 = Interpolate_UV(ax1, v[i].U, v[i + 1].U);
//DevCon.Warning("Tex width %d draw width %d (X %x -> %x U %x -> %x", tex_width, ((v[i + 1].XYZ.X - v[i].XYZ.X) >> 4), X0, X1, v[i].U, v[i+1].U);
#ifdef DEBUG_U
if (debug)
{
fprintf(stderr, "u0:%d and u1:%d\n", v[i].U, v[i + 1].U);
fprintf(stderr, "a0:%f and a1:%f\n", ax0, ax1);
fprintf(stderr, "t0:%d and t1:%d\n", tx0, tx1);
}
if (debug)
{
fprintf(stderr, "u0:%d and u1:%d\n", v[i].U, v[i + 1].U);
fprintf(stderr, "a0:%f and a1:%f\n", ax0, ax1);
fprintf(stderr, "t0:%d and t1:%d\n", tx0, tx1);
}
#endif
#if 1
// Use rounded value of the newly computed texture coordinate. It ensures
// that sampling will remains inside texture boundary
//
// Note for bilinear: by definition it will never work correctly! A sligh modification
// of interpolation migth trigger a discard (with alpha testing)
// Let's use something simple that correct really bad case (for a couple of 2D games).
// I hope it won't create too much glitches.
if (linear)
{
const int Lu = v[i + 1].U - v[i].U;
// Note 32 is based on taisho-mononoke
if ((Lu > 0) && (Lu <= (Lx + 32)))
{
v[i + 1].U -= 8;
}
}
else
{
if (tx0 <= tx1)
{
v[i].U = tx0;
v[i + 1].U = tx1 + 16;
}
else
{
v[i].U = tx0 + 15;
v[i + 1].U = tx1;
}
}
#endif
}
else
{
if (((v[i + 1].U & 0xf) ^ ((v[i + 1].XYZ.X - ox) & 0xf)) && ((v[i + 1].U - v[i].U) >> 4) == tex_width && (Lx >> 4) / tex_width <= 2)
{
v[i].U &= ~0xf;
v[i + 1].U -= 8;
}
else
{
v[i].U &= ~0xf;
v[i + 1].U &= ~0xf;
v[i].U |= (v[i].XYZ.X - ox) & 0xf;
v[i + 1].U |= (v[i + 1].XYZ.X - ox) & 0xf;
}
}
const int tex_height = std::max(1, (v[i + 1].V - v[i].V) >> 4);
const int oy = m_context->XYOFFSET.OFY;
const int Y0 = v[i].XYZ.Y - oy;
const int Y1 = v[i + 1].XYZ.Y - oy;
const int Ly = (v[i + 1].XYZ.Y - v[i].XYZ.Y);
const float ay0 = alpha0(Ly, Y0, Y1);
const float ay1 = alpha1(Ly, Y0, Y1);
const u16 ty0 = Interpolate_UV(ay0, v[i].V, v[i + 1].V);
const u16 ty1 = Interpolate_UV(ay1, v[i].V, v[i + 1].V);
#ifdef DEBUG_V
if (debug)
{
fprintf(stderr, "v0:%d and v1:%d\n", v[i].V, v[i + 1].V);
fprintf(stderr, "a0:%f and a1:%f\n", ay0, ay1);
fprintf(stderr, "t0:%d and t1:%d\n", ty0, ty1);
}
#endif
if ((((Ly - oy) >> 4) % tex_height) != 0)
{
const int Y0 = v[i].XYZ.Y - oy;
const int Y1 = v[i + 1].XYZ.Y - oy;
const float ay0 = alpha0(Ly, Y0, Y1);
const float ay1 = alpha1(Ly, Y0, Y1);
const u16 ty0 = Interpolate_UV(ay0, v[i].V, v[i + 1].V);
const u16 ty1 = Interpolate_UV(ay1, v[i].V, v[i + 1].V);
#ifdef DEBUG_V
if (debug)
{
fprintf(stderr, "v0:%d and v1:%d\n", v[i].V, v[i + 1].V);
fprintf(stderr, "a0:%f and a1:%f\n", ay0, ay1);
fprintf(stderr, "t0:%d and t1:%d\n", ty0, ty1);
}
#endif
#if 1
if (linear)
{
const int Lv = v[i + 1].V - v[i].V;
if ((Lv > 0) && (Lv <= (Ly + 32)))
{
v[i + 1].V -= 8;
}
}
else
{
if (ty0 <= ty1)
{
v[i].V = ty0;
v[i + 1].V = ty1 + 16;
}
else
{
v[i].V = ty0 + 15;
v[i + 1].V = ty1;
}
}
#endif
}
else
{
if (((v[i + 1].V & 0xf) ^ ((v[i + 1].XYZ.Y - oy) & 0xf)) && ((v[i + 1].V - v[i].V) >> 4) == tex_height && (Ly >> 4) / tex_height <= 2)
{
v[i].V &= ~0xf;
v[i + 1].V -= 8;
}
else
{
v[i].V &= ~0xf;
v[i + 1].V &= ~0xf;
v[i].V |= (v[i].XYZ.Y - oy) & 0xf;
v[i + 1].V |= (v[i + 1].XYZ.Y - oy) & 0xf;
}
}
#ifdef DEBUG_U
if (debug)
fprintf(stderr, "GREP_BEFORE %d => %d\n", v[i].U, v[i + 1].U);
@ -1904,60 +1998,6 @@ void GSRendererHW::RoundSpriteOffset()
fprintf(stderr, "GREP_BEFORE %d => %d\n", v[i].V, v[i + 1].V);
#endif
#if 1
// Use rounded value of the newly computed texture coordinate. It ensures
// that sampling will remains inside texture boundary
//
// Note for bilinear: by definition it will never work correctly! A sligh modification
// of interpolation migth trigger a discard (with alpha testing)
// Let's use something simple that correct really bad case (for a couple of 2D games).
// I hope it won't create too much glitches.
if (linear)
{
const int Lu = v[i + 1].U - v[i].U;
// Note 32 is based on taisho-mononoke
if ((Lu > 0) && (Lu <= (Lx + 32)))
{
v[i + 1].U -= 8;
}
}
else
{
if (tx0 <= tx1)
{
v[i].U = tx0;
v[i + 1].U = tx1 + 16;
}
else
{
v[i].U = tx0 + 15;
v[i + 1].U = tx1;
}
}
#endif
#if 1
if (linear)
{
const int Lv = v[i + 1].V - v[i].V;
if ((Lv > 0) && (Lv <= (Ly + 32)))
{
v[i + 1].V -= 8;
}
}
else
{
if (ty0 <= ty1)
{
v[i].V = ty0;
v[i + 1].V = ty1 + 16;
}
else
{
v[i].V = ty0 + 15;
v[i + 1].V = ty1;
}
}
#endif
#ifdef DEBUG_U
if (debug)