mirror of https://github.com/aap/librw.git
better ps2 swizzling code
This commit is contained in:
parent
e9d53fd834
commit
c60cecbd02
|
@ -654,6 +654,13 @@ skinUninstanceCB(Geometry *geo)
|
|||
case 4: // weights
|
||||
for(int32 i = 0; i < geo->numVertices; i++){
|
||||
unpackattrib(weights, p, a);
|
||||
float sum = weights[0] + weights[1] + weights[2] + weights[3];
|
||||
if(sum){
|
||||
weights[0] /= sum;
|
||||
weights[1] /= sum;
|
||||
weights[2] /= sum;
|
||||
weights[3] /= sum;
|
||||
}
|
||||
weights += 4;
|
||||
p += a->stride;
|
||||
}
|
||||
|
|
|
@ -1396,47 +1396,82 @@ convertCSM1_32(uint32 *dst, uint32 *src)
|
|||
}
|
||||
}
|
||||
|
||||
// TMP
|
||||
// Unswizzle PSMT8 from PSMCT32 format
|
||||
// Original code from http://ps2linux.no-ip.info/playstation2-linux.com/docs/howto/display_docef7c.html?docid=75
|
||||
static void
|
||||
unswizzle8(uint8 *dst, uint8 *src, uint32 w, uint32 h)
|
||||
unswizzle32to8(uint8 *dst, uint8 *src, int32 w, int32 h)
|
||||
{
|
||||
for (uint32 y = 0; y < h; y++)
|
||||
for (uint32 x = 0; x < w; x++) {
|
||||
int32 x, y;
|
||||
for(y = 0; y < h; y++)
|
||||
for(x = 0; x < w; x++) {
|
||||
// byte address of beginning of PSMCT32 block
|
||||
// yblock = y/16 * w/16 * 256
|
||||
int32 block_loc = (y&(~0xF))*w + (x&(~0xF))*2;
|
||||
uint32 swap_sel = (((y+2)>>2)&0x1)*4;
|
||||
int32 ypos = (((y&(~3))>>1) + (y&1))&0x7;
|
||||
int32 column_loc = ypos*w*2 + ((x+swap_sel)&0x7)*4;
|
||||
int32 byte_sum = ((y>>1)&1) + ((x>>2)&2);
|
||||
uint32 swizzled = block_loc + column_loc + byte_sum;
|
||||
// y coordinate in PSMCT32 block
|
||||
int32 ypos = (((y&~3)>>1) // column number
|
||||
+ (y&1))&0x7; // row inside column
|
||||
// indices are swapped in groups of 4 every 4
|
||||
// pixel rows in a block, first at y = 2
|
||||
int32 swap_sel = (y+2)&4;
|
||||
// byte address of word in PSMCT32 block
|
||||
int32 word_loc = ypos*w*2 + ((x+swap_sel)&0x7)*4;
|
||||
// byte offset in 32 bit word
|
||||
int32 byte_loc = ((y>>1)&1) + ((x>>2)&2);
|
||||
int32 swizzled = block_loc + word_loc + byte_loc;
|
||||
dst[y*w+x] = src[swizzled];
|
||||
}
|
||||
}
|
||||
|
||||
// Unswizzle PSMT4 from PSMCT16 format
|
||||
static void
|
||||
unswizzle(uint8 *dst, uint8 *src, int32 w, int32 h, int32 d)
|
||||
unswizzle16to4(uint8 *dst, uint8 *src, int32 w, int32 h)
|
||||
{
|
||||
int32 minw, minh;
|
||||
int32 tw, th;
|
||||
int32 x, y;
|
||||
for(y = 0; y < h; y++)
|
||||
for(x = 0; x < w; x++){
|
||||
// byte address of beginning of PSMCT16 block
|
||||
// yblock = y/16 * w/32 * 256
|
||||
int32 block_loc = (y&(~0xF))*w/2 + (x&(~0x1F));
|
||||
// y coordinate in PSMCT16 block
|
||||
int32 ypos = (((y&~3)>>1) // column number
|
||||
+ (y&1))&0x7; // row inside column
|
||||
// indices are swapped in groups of 4 every 4
|
||||
// pixel rows in a block, first at y = 2
|
||||
int32 swap_sel = (y+2)&4;
|
||||
// byte address of halfword in PSMCT16 block
|
||||
int32 hword_loc = ypos*w
|
||||
+ (((x+swap_sel) & 0x7) + ((x & 0x10)>>1))*2;
|
||||
// byte offset in 16 bit word
|
||||
int32 byte_loc = (x>>3)&1;
|
||||
uint32 swizzled = block_loc + hword_loc + byte_loc;
|
||||
// y&2 selects nibbles
|
||||
int8 c = y & 2 ? src[swizzled] >> 4 : src[swizzled] & 0xF;
|
||||
int32 addr = y*w + x;
|
||||
if(addr & 1)
|
||||
dst[addr>>1] = dst[addr>>1]&0xF | c<<4;
|
||||
else
|
||||
dst[addr>>1] = dst[addr>>1]&0xF0 | c;
|
||||
}
|
||||
}
|
||||
|
||||
if(d == 4)
|
||||
transferMinSize(PSMT4, 4, &minw, &minh);
|
||||
else
|
||||
transferMinSize(PSMT8, 2, &minw, &minh);
|
||||
tw = max(w, minw);
|
||||
th = max(h, minh);
|
||||
if(tw == w && th == h){
|
||||
unswizzle8(dst, src, w, h);
|
||||
return;
|
||||
}
|
||||
static void
|
||||
expandPSMT4(uint8 *dst, uint8 *src, int32 w, int32 h, int32 srcw)
|
||||
{
|
||||
int32 x, y;
|
||||
for(y = 0; y < h; y++)
|
||||
for(x = 0; x < w/2; x++){
|
||||
dst[y*w + x*2 + 0] = src[y*srcw/2 + x] & 0xF;
|
||||
dst[y*w + x*2 + 1] = src[y*srcw/2 + x] >> 4;
|
||||
}
|
||||
}
|
||||
|
||||
// this is more complicated...
|
||||
uint8 *tmp = rwNewT(uint8, tw*th, MEMDUR_FUNCTION | ID_RASTERPS2);
|
||||
unswizzle8(tmp, src, tw, th);
|
||||
static void
|
||||
copyPSMT8(uint8 *dst, uint8 *src, int32 w, int32 h, int32 srcw)
|
||||
{
|
||||
int32 x, y;
|
||||
for(y = 0; y < h; y++)
|
||||
for(x = 0; x < w; x++)
|
||||
dst[y*w + x] = tmp[y*tw + x];
|
||||
rwFree(tmp);
|
||||
dst[y*w + x] = src[y*srcw + x];
|
||||
}
|
||||
|
||||
Image*
|
||||
|
@ -1505,33 +1540,29 @@ rasterToImage(Raster *raster)
|
|||
w = raster->width;
|
||||
h = raster->height;
|
||||
|
||||
int minw, minh;
|
||||
int th, tw;
|
||||
transferMinSize(depth == 4 ? PSMT4 : PSMT8, natras->flags, &minw, &minh);
|
||||
tw = max(w, minw);
|
||||
th = max(h, minh);
|
||||
out = image->pixels;
|
||||
in = raster->lock(0);
|
||||
if(depth == 4){
|
||||
// Actually PSMT4 is transferred as PSMCT16 but
|
||||
// apparently we can "uncompress" the indices into PSMT8
|
||||
// and unswizzle as if transferred as PSMCT32
|
||||
// Would be nice to understand this correctly some day
|
||||
int minw, minh;
|
||||
int th, tw;
|
||||
transferMinSize(PSMT4, 4, &minw, &minh);
|
||||
tw = max(w, minw);
|
||||
th = max(h, minh);
|
||||
uint8 *in8 = rwNewT(uint8, tw*th, MEMDUR_FUNCTION | ID_RASTERPS2);
|
||||
for(uint32 i = 0; i < tw*th/2; i++){
|
||||
in8[i*2+0] = in[i] & 0xF;
|
||||
in8[i*2+1] = in[i] >> 4;
|
||||
}
|
||||
if(natras->flags & Ps2Raster::SWIZZLED4)
|
||||
unswizzle(out, in8, w, h, 4);
|
||||
else
|
||||
memcpy(out, in8, w*h);
|
||||
rwFree(in8);
|
||||
if(natras->flags & Ps2Raster::SWIZZLED4){
|
||||
uint8 *tmp = rwNewT(uint8, tw*th/2, MEMDUR_FUNCTION | ID_RASTERPS2);
|
||||
unswizzle16to4(tmp, in, tw, th);
|
||||
expandPSMT4(out, tmp, w, h, tw);
|
||||
rwFree(tmp);
|
||||
}else
|
||||
expandPSMT4(out, in, w, h, tw);
|
||||
}else if(depth == 8){
|
||||
if(natras->flags & Ps2Raster::SWIZZLED8)
|
||||
unswizzle(out, in, w, h, 8);
|
||||
else
|
||||
memcpy(out, in, w*h);
|
||||
if(natras->flags & Ps2Raster::SWIZZLED8){
|
||||
uint8 *tmp = rwNewT(uint8, tw*th, MEMDUR_FUNCTION | ID_RASTERPS2);
|
||||
unswizzle32to8(tmp, in, tw, th);
|
||||
copyPSMT8(out, tmp, w, h, tw);
|
||||
rwFree(tmp);
|
||||
}else
|
||||
copyPSMT8(out, in, w, h, tw);
|
||||
}else
|
||||
// TODO: stride
|
||||
for(int32 y = 0; y < image->height; y++)
|
||||
|
|
Loading…
Reference in New Issue