trnr/tlib
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add oversampler

Browse Source
This commit is contained in:
Chris
2024-05-24 13:28:31 +02:00
parent e4a4a661a0
commit 989dba5a6b
484 changed files with 313937 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

446
oversampling/WDL/plush2/pl_pf_tex.h Normal file
View File

@@ -0,0 +1,446 @@
static void
#ifdef PL_PF_MULTITEX
PLMTexTri
#else
PLTexTri
#endif
(LICE_pixel *gmem, int swidth, pl_Face *TriFace, pl_ZBuffer *zbuf, int zfb_width, bool zb_update,
int solidalpha, int solidcomb, LICE_IBitmap *tex, pl_Float *texscales, int texalpha, int texcomb, int texmap
#ifdef PL_PF_MULTITEX
, LICE_IBitmap *tex2, int tex2alpha, int tex2comb, int texmap2
#endif
)
{
pl_sInt32 C1[3], C3[3], C2[3], dC2[3]={0}, dCL[3]={0},dC1[3]={0}, dX2=0, dX1=0;
pl_Float dZ2=0, dZL=0,dZ1=0;
PUTFACE_SORT();
solidcomb&=LICE_BLIT_MODE_MASK;
if (solidcomb == 0 && solidalpha == 256) solidcomb=-1;
else if (solidalpha==0) solidcomb=-2; // ignore
int solidalpha2=(256-solidalpha)*256;
pl_Float dU2=0,dV2=0,dUL=0,dVL=0;
pl_Float dU1=0, dV1=0;
bool bilinear = (texcomb&LICE_BLIT_FILTER_MASK)==LICE_BLIT_FILTER_BILINEAR;
texcomb&=LICE_BLIT_MODE_MASK|LICE_BLIT_USE_ALPHA;
if (texcomb==LICE_BLIT_MODE_COPY && texalpha==256) texcomb=-1;
else if (texalpha==0) texcomb=-2;
int texalpha2=(256-texalpha);
int tex_rowspan=0;
LICE_pixel *texture=NULL;
int tex_w=16,tex_h=16;
#if defined(PLUSH_NO_SOLIDGOURAUD) || defined(PLUSH_NO_TEXTURE)
if (tex)
#endif
{
texture=tex->getBits();
tex_rowspan = tex->getRowSpan();
if (tex->isFlipped())
{
texture += tex_rowspan*(tex->getHeight()-1);
tex_rowspan=-tex_rowspan;
}
tex_w=tex->getWidth();
tex_h=tex->getHeight();
tex_rowspan *= 4;
}
pl_sInt32 MappingU_Max=tex_w<<16, MappingV_Max=tex_h<<16;
texscales[0]*=MappingU_Max;
texscales[1]*=MappingV_Max;
tex_w *= 4;
#ifdef PL_PF_MULTITEX
pl_Float dU2_2=0,dV2_2=0,dUL_2=0,dVL_2=0;
pl_Float dU1_2=0, dV1_2=0;
int tex_w_2=16,tex_h_2=16;
bool bilinear2 = (tex2comb&LICE_BLIT_FILTER_MASK)==LICE_BLIT_FILTER_BILINEAR;
tex2comb&=LICE_BLIT_MODE_MASK|LICE_BLIT_USE_ALPHA;
if (tex2comb==LICE_BLIT_MODE_COPY && tex2alpha==256) tex2comb=-1;
else if (tex2alpha==0) tex2comb=-2;
int tex2alpha2=(256-tex2alpha);
LICE_pixel *texture_2=NULL;
int tex_rowspan_2 = 0;
pl_sInt32 MappingU_Max_2, MappingV_Max_2;
#ifdef PLUSH_NO_TEXTURE
if (tex2)
#endif
{
tex_w_2=tex2->getWidth();
tex_h_2=tex2->getHeight();
texture_2=tex2->getBits();
tex_rowspan_2 = tex2->getRowSpan();
if (tex2->isFlipped())
{
texture_2 += tex_rowspan_2*(tex2->getHeight()-1);
tex_rowspan_2=-tex_rowspan_2;
}
tex_rowspan_2 *= 4;
}
MappingU_Max_2=tex_w_2<<16;
MappingV_Max_2=tex_h_2<<16;
texscales[2]*=MappingU_Max_2;
texscales[3]*=MappingV_Max_2;
tex_w_2 *= 4;
#endif
pl_uChar nm = TriFace->Material->PerspectiveCorrect;
pl_uChar nmb = 0; while (nm) { nmb++; nm >>= 1; }
nmb = plMin(6,nmb);
nm = 1<<nmb;
pl_Float Z1 = TriFace->Scrz[i0];
pl_Float Z2 = TriFace->Scrz[i1];
pl_Float Z3 = TriFace->Scrz[i2];
pl_Float MappingU1=TriFace->MappingU[texmap][i0]*texscales[0]*Z1;
pl_Float MappingV1=TriFace->MappingV[texmap][i0]*texscales[1]*Z1;
pl_Float MappingU2=TriFace->MappingU[texmap][i1]*texscales[0]*Z2;
pl_Float MappingV2=TriFace->MappingV[texmap][i1]*texscales[1]*Z2;
pl_Float MappingU3=TriFace->MappingU[texmap][i2]*texscales[0]*Z3;
pl_Float MappingV3=TriFace->MappingV[texmap][i2]*texscales[1]*Z3;
#ifdef PL_PF_MULTITEX
pl_Float MappingU1_2=TriFace->MappingU[texmap2][i0]*texscales[2]*Z1;
pl_Float MappingV1_2=TriFace->MappingV[texmap2][i0]*texscales[3]*Z1;
pl_Float MappingU2_2=TriFace->MappingU[texmap2][i1]*texscales[2]*Z2;
pl_Float MappingV2_2=TriFace->MappingV[texmap2][i1]*texscales[3]*Z2;
pl_Float MappingU3_2=TriFace->MappingU[texmap2][i2]*texscales[2]*Z3;
pl_Float MappingV3_2=TriFace->MappingV[texmap2][i2]*texscales[3]*Z3;
#endif
int a;
for(a=0;a<3;a++)
{
C1[a] = (pl_sInt32) (TriFace->Shades[i0][a]*(1<<24));
C2[a] = (pl_sInt32) (TriFace->Shades[i1][a]*(1<<24));
C3[a] = (pl_sInt32) (TriFace->Shades[i2][a]*(1<<24));
}
pl_Float U1, U2;
U1 = U2 = MappingU1;
pl_Float V1,V2;
V1 = V2 = MappingV1;
#ifdef PL_PF_MULTITEX
pl_Float U1_2, U2_2;
U1_2 = U2_2 = MappingU1_2;
pl_Float V1_2, V2_2;
V1_2 = V2_2 = MappingV1_2;
#endif
pl_sInt32 X2,X1;
X2 = X1 = Scrx[i0];
pl_sInt32 Y0 = Scry[i0];
pl_sInt32 Y1 = Scry[i1];
pl_sInt32 Y2 = Scry[i2];
{
pl_sInt32 dY = Y2-Y0;
if (dY) {
dX2 = (Scrx[i2] - X1) / dY;
pl_Float v = 1.0/dY;
for(a=0;a<3;a++) dC2[a] = (pl_sInt32) ((C3[a] - C1[a]) * v);
dZ2 = (Z3 - Z1) * v;
dU2 = (MappingU3 - U1) * v;
dV2 = (MappingV3 - V1) * v;
#ifdef PL_PF_MULTITEX
dU2_2 = (MappingU3_2 - U1_2) * v;
dV2_2 = (MappingV3_2 - V1_2) * v;
#endif
}
dY = Y1-Y0;
if (dY) {
dX1 = (Scrx[i1] - X1) / dY;
pl_Float v=1.0/dY;
dZ1 = (Z2 - Z1) * v;
for(a=0;a<3;a++) dC1[a] = (pl_sInt32) ((C2[a] - C1[a]) * v);
dU1 = (MappingU2 - U1) * v;
dV1 = (MappingV2 - V1) * v;
#ifdef PL_PF_MULTITEX
dU1_2 = (MappingU2_2 - U1_2) * v;
dV1_2 = (MappingV2_2 - V1_2) * v;
#endif
if (dX2 < dX1) {
SWAP(dX1,dX2,pl_sInt32);
SWAP(dU1,dU2,pl_Float);
SWAP(dV1,dV2,pl_Float);
#ifdef PL_PF_MULTITEX
SWAP(dU1_2,dU2_2,pl_Float);
SWAP(dV1_2,dV2_2,pl_Float);
#endif
SWAP(dZ1,dZ2,pl_Float);
for(a=0;a<3;a++)
SWAP(dC1[a],dC2[a],pl_sInt32);
stat = 2;
} else stat = 1;
Z2 = Z1;
C2[0] = C1[0];
C2[1] = C1[1];
C2[2] = C1[2];
} else {
if (Scrx[i1] > X1) {
X2 = Scrx[i1];
U2 = MappingU2;
V2 = MappingV2;
#ifdef PL_PF_MULTITEX
U2_2 = MappingU2_2;
V2_2 = MappingV2_2;
#endif
stat = 2|4;
} else {
X1 = Scrx[i1];
SWAP(Z1,Z2,pl_Float)
for(a=0;a<3;a++) SWAP(C1[a],C2[a],pl_sInt32);
U1 = MappingU2;
V1 = MappingV2;
#ifdef PL_PF_MULTITEX
U1_2 = MappingU2_2;
V1_2 = MappingV2_2;
#endif
stat = 1|8;
}
}
pl_sInt32 tmp = (dX1-dX2)*dY;
if (tmp) {
pl_Float v=(1<<XPOS_BITS)/(double)tmp;
dUL = ((dU1-dU2)*dY)*v;
dVL = ((dV1-dV2)*dY)*v;
#ifdef PL_PF_MULTITEX
dUL_2 = ((dU1_2-dU2_2)*dY)*v;
dVL_2 = ((dV1_2-dV2_2)*dY)*v;
#endif
dZL = ((dZ1-dZ2)*dY)*v;
for(a=0;a<3;a++) dCL[a] = (pl_sInt32) ( ((dC1[a]-dC2[a])*dY)*v);
} else {
tmp = X2-X1;
if (tmp) {
pl_Float v=(1<<XPOS_BITS)/(double)tmp;
dUL = (U2-U1)*v;
dVL = (V2-V1)*v;
#ifdef PL_PF_MULTITEX
dUL_2 = (U2_2-U1_2)*v;
dVL_2 = (V2_2-V1_2)*v;
#endif
dZL = (Z2-Z1)*v;
for(a=0;a<3;a++) dCL[a] = (pl_sInt32) ((C2[a]-C1[a])*v);
}
}
}
gmem += (Y0 * swidth);
zbuf += (Y0 * zfb_width);
pl_Float pdZL = dZL * nm;
dUL *= nm;
dVL *= nm;
#ifdef PL_PF_MULTITEX
dUL_2 *= nm;
dVL_2 *= nm;
#endif
Y1 -= Y0;
Y0 = Y2-Y0;
while (Y0--) {
if (!Y1--) {
pl_sInt32 dY = Y2-Scry[i1];
if (dY) {
DO_STAT_XDELTAS
pl_Float tmp=1.0/dY;
dZ1 = (Z3-Z1)*tmp;
for(a=0;a<3;a++) dC1[a] = (pl_sInt32)((C3[a]-C1[a])*tmp);
dV1 = (MappingV3 - V1)*tmp;
dU1 = (MappingU3 - U1)*tmp;
#ifdef PL_PF_MULTITEX
dV1_2 = (MappingV3_2 - V1_2)*tmp;
dU1_2 = (MappingU3_2 - U1_2)*tmp;
#endif
}
}
pl_sInt32 XL1 = (X1+(1<<(XPOS_BITS-1)))>>XPOS_BITS;
pl_sInt32 Xlen = ((X2+(1<<(XPOS_BITS-1)))>>XPOS_BITS) - XL1;
if (Xlen > 0) {
pl_sInt32 iUL, iVL, idUL, idVL, iULnext, iVLnext;
pl_Float UL = U1;
pl_Float VL = V1;
#ifdef PL_PF_MULTITEX
pl_sInt32 iUL_2, iVL_2, idUL_2, idVL_2, iULnext_2, iVLnext_2;
pl_Float UL_2 = U1_2;
pl_Float VL_2 = V1_2;
#endif
pl_sInt32 CL[3] = {C1[0],C1[1], C1[2]};
pl_Float pZL,ZL;
pl_Float t = 1.0f / (pZL = ZL = Z1);
gmem += XL1;
zbuf += XL1;
XL1 += Xlen; // update to new line end pos so we can adjust gmem/zbuf later
iULnext = ((pl_sInt32) (UL*t));
iVLnext = ((pl_sInt32) (VL*t));
#ifdef PL_PF_MULTITEX
iULnext_2 = ((pl_sInt32) (UL_2*t));
iVLnext_2 = ((pl_sInt32) (VL_2*t));
#endif
do {
UL += dUL;
VL += dVL;
iUL = iULnext;
iVL = iVLnext;
pZL += pdZL;
t = 1.0f/pZL;
iULnext = ((pl_sInt32) (UL*t));
iVLnext = ((pl_sInt32) (VL*t));
idUL = (iULnext - iUL)>>nmb;
idVL = (iVLnext - iVL)>>nmb;
if (idUL>MappingU_Max) idUL=MappingU_Max;
else if (idUL<-MappingU_Max) idUL=-MappingU_Max;
if (idVL>MappingV_Max) idVL=MappingV_Max;
else if (idVL<-MappingV_Max) idVL=-MappingV_Max;
// todo: this is slow as shit, should we force textures to be powers of two? hehe
if (iUL<0) do iUL+=MappingU_Max; while (iUL<0);
else while (iUL >= MappingU_Max) iUL-=MappingU_Max;
if (iVL<0) do iVL+=MappingV_Max; while (iVL<0);
else while (iVL >= MappingV_Max) iVL-=MappingV_Max;
#ifdef PL_PF_MULTITEX
UL_2 += dUL_2;
VL_2 += dVL_2;
iUL_2 = iULnext_2;
iVL_2 = iVLnext_2;
iULnext_2 = ((pl_sInt32) (UL_2*t));
iVLnext_2 = ((pl_sInt32) (VL_2*t));
idUL_2 = (iULnext_2 - iUL_2)>>nmb;
idVL_2 = (iVLnext_2 - iVL_2)>>nmb;
if (idUL_2>MappingU_Max_2) idUL_2=MappingU_Max_2;
else if (idUL_2<-MappingU_Max_2) idUL_2=-MappingU_Max_2;
if (idVL_2>MappingV_Max_2) idVL_2=MappingV_Max_2;
else if (idVL_2<-MappingV_Max_2) idVL_2=-MappingV_Max_2;
// todo: this is slow as shit, should we force textures to be powers of two? hehe
if (iUL_2<0) do iUL_2+=MappingU_Max_2; while (iUL_2<0);
else while (iUL_2 >= MappingU_Max_2) iUL_2-=MappingU_Max_2;
if (iVL_2<0) do iVL_2+=MappingV_Max_2; while (iVL_2<0);
else while (iVL_2 >= MappingV_Max_2) iVL_2-=MappingV_Max_2;
#endif
pl_uInt n = nm;
Xlen -= n;
if (Xlen < 0) n += Xlen;
if (zfb_width) do {
if (*zbuf < ZL) {
if (zb_update) *zbuf = (pl_ZBuffer) ZL;
#ifdef PL_PF_MULTITEX
TextureMakePixel2((LICE_pixel_chan *)gmem,solidcomb,solidalpha,solidalpha2,CL, bilinear,
iUL,iVL,tex_w,tex_h,texture,tex_rowspan,texcomb,texalpha,texalpha2,
bilinear2, iUL_2,iVL_2,
tex_w_2,tex_h_2,
texture_2,tex_rowspan_2,tex2comb,tex2alpha,tex2alpha2);
#else
TextureMakePixel((LICE_pixel_chan *)gmem,solidcomb,solidalpha,solidalpha2,CL, bilinear, iUL,iVL,
tex_w,tex_h,
texture,tex_rowspan,texcomb,texalpha,texalpha2);
#endif
}
zbuf++;
gmem++;
ZL += dZL;
CL[0] += dCL[0];
CL[1] += dCL[1];
CL[2] += dCL[2];
iUL += idUL;
iVL += idVL;
if (iUL<0) iUL+=MappingU_Max;
else if (iUL >= MappingU_Max) iUL -= MappingU_Max;
if (iVL<0) iVL+=MappingV_Max;
else if (iVL >= MappingV_Max) iVL -= MappingV_Max;
#ifdef PL_PF_MULTITEX
iUL_2 += idUL_2;
iVL_2 += idVL_2;
if (iUL_2<0) iUL_2+=MappingU_Max_2;
else if (iUL_2 >= MappingU_Max_2) iUL_2 -= MappingU_Max_2;
if (iVL_2<0) iVL_2+=MappingV_Max_2;
else if (iVL_2 >= MappingV_Max_2) iVL_2 -= MappingV_Max_2;
#endif
} while (--n);
else do {
#ifdef PL_PF_MULTITEX
TextureMakePixel2((LICE_pixel_chan *)gmem,solidcomb,solidalpha,solidalpha2,CL, bilinear,
iUL,iVL,tex_w,tex_h,texture,tex_rowspan,texcomb,texalpha,texalpha2,
bilinear2, iUL_2,iVL_2,
tex_w_2,tex_h_2,
texture_2,tex_rowspan_2,tex2comb,tex2alpha,tex2alpha2);
#else
TextureMakePixel((LICE_pixel_chan *)gmem,solidcomb,solidalpha,solidalpha2,CL, bilinear, iUL,iVL,
tex_w,tex_h,
texture,tex_rowspan,texcomb,texalpha,texalpha2);
#endif
gmem++;
CL[0] += dCL[0];
CL[1] += dCL[1];
CL[2] += dCL[2];
iUL += idUL;
iVL += idVL;
if (iUL<0) iUL+=MappingU_Max;
else if (iUL >= MappingU_Max) iUL -= MappingU_Max;
if (iVL<0) iVL+=MappingV_Max;
else if (iVL >= MappingV_Max) iVL -= MappingV_Max;
#ifdef PL_PF_MULTITEX
iUL_2 += idUL_2;
iVL_2 += idVL_2;
if (iUL_2<0) iUL_2+=MappingU_Max_2;
else if (iUL_2 >= MappingU_Max_2) iUL_2 -= MappingU_Max_2;
if (iVL_2<0) iVL_2+=MappingV_Max_2;
else if (iVL_2 >= MappingV_Max_2) iVL_2 -= MappingV_Max_2;
#endif
} while (--n);
} while (Xlen > 0);
gmem += swidth-XL1;
zbuf += zfb_width-XL1;
} else { // xlen <=0 ,no drawing
gmem += swidth;
zbuf += zfb_width;
}
Z1 += dZ1;
U1 += dU1;
V1 += dV1;
#ifdef PL_PF_MULTITEX
U1_2 += dU1_2;
V1_2 += dV1_2;
#endif
X1 += dX1;
X2 += dX2;
C1[0] += dC1[0];
C1[1] += dC1[1];
C1[2] += dC1[2];
}
}