tmaq
/
tornavis
1
0
Fork 0
Code Issues 2 Pull Requests 29 Packages Projects 24 Releases Wiki Activity

Correction to the mipmaps generation 

Seem we've always were wrong with multiplying alpha by 255,
other channels seems to be multiplied by 256 with the shift
operations.
This commit is contained in:
Sergey Sharybin 2014-08-18 18:03:56 +06:00
parent 6c7467e0da
commit dc05d817eb
1 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
source/blender/imbuf/intern/scaling.c
View File

@ -320,7 +320,7 @@ MINLINE void straight_uchar_to_premul_ushort(unsigned short result[4], const uns
result[0] = color[0] * alpha;
result[1] = color[1] * alpha;
result[2] = color[2] * alpha;
result[3] = alpha * 255;
result[3] = alpha * 256;
}
MINLINE void premul_ushort_to_straight_uchar(unsigned char *result, const unsigned short color[4])
@ -332,11 +332,11 @@ MINLINE void premul_ushort_to_straight_uchar(unsigned char *result, const unsign
result[3] = USHORTTOUCHAR(color[3]);
}
else {
unsigned short alpha = color[3] / 255;
unsigned short alpha = color[3] / 256;
result[0] = USHORTTOUCHAR(color[0] / alpha * 255);
result[1] = USHORTTOUCHAR(color[1] / alpha * 255);
result[2] = USHORTTOUCHAR(color[2] / alpha * 255);
result[0] = USHORTTOUCHAR(color[0] / alpha * 256);
result[1] = USHORTTOUCHAR(color[1] / alpha * 256);
result[2] = USHORTTOUCHAR(color[2] / alpha * 256);
result[3] = USHORTTOUCHAR(color[3]);
}
}