AuroraMiddleware/freetype
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

[base] Further clean up color bitmap conversion.

Browse Source 
* src/base/ftbitmap.c (ft_gray_for_premultiplied_srgb_bgra): Stop
using FT_MulFix and FT_DivFix since all calculations fit into 32 bits.
This commit is contained in:
Alexei Podtelezhnikov 2014-07-14 22:30:55 -04:00
parent 441b3f3898
commit 1183cd4e17
2 changed files with 17 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
ChangeLog
View File

@ -1,3 +1,10 @@
2014-07-11 Alexei Podtelezhnikov <apodtele@gmail.com>
[base] Further clean up color bitmap conversion.
* src/base/ftbitmap.c (ft_gray_for_premultiplied_srgb_bgra): Stop
using FT_MulFix and FT_DivFix since all calculations fit into 32 bits.
2014-07-13 Werner Lemberg <wl@gnu.org>
[truetype] Improve handling of buggy `prep' tables.

33
src/base/ftbitmap.c
View File

@ -378,11 +378,8 @@
static FT_Byte
ft_gray_for_premultiplied_srgb_bgra( const FT_Byte* bgra )
{
FT_Long a = bgra[3];
FT_Long b = bgra[0];
FT_Long g = bgra[1];
FT_Long r = bgra[2];
FT_Long l;
FT_Byte a = bgra[3];
FT_ULong l;
/* Short-circuit transparent color to avoid div-by-zero. */
@ -397,25 +394,14 @@
*
* http://accessibility.kde.org/hsl-adjusted.php
*
* We do the computation with integers only.
* We do the computation with integers only applying gamma of 2.0,
* The following will never overflow 32 bits. This is a scaled up
* luminosity with premultiplication not yet undone.
*/
/* Undo premultification, get the number in a 16.16 form. */
b = FT_DivFix( b, a );
g = FT_DivFix( g, a );
r = FT_DivFix( r, a );
/* Apply gamma of 2.0 instead of 2.2. */
b = FT_MulFix( b, b );
g = FT_MulFix( g, g );
r = FT_MulFix( r, r );
/* Apply coefficients. */
b = FT_MulFix( b, 4731 /* 0.0722 * 65536 */ );
g = FT_MulFix( g, 46871 /* 0.7152 * 65536 */ );
r = FT_MulFix( r, 13933 /* 0.2126 * 65536 */ );
l = r + g + b;
l = 4731UL /* 0.0722 * 65536 */ * bgra[0] * bgra[0] +
46871UL /* 0.7152 * 65536 */ * bgra[1] * bgra[1] +
13933UL /* 0.2126 * 65536 */ * bgra[2] * bgra[2];
/*
* Final transparency can be determined this way:
@ -425,9 +411,10 @@
* - If alpha is zero and luminosity is one, we want 0.
*
* So the formula is a * (1 - l) = a - l * a.
* Undoing premultiplication and scaling back down we get
*/
return a - (FT_Byte)FT_MulFix( l, a );
return a - (FT_Byte)( ( l / a ) >> 16 );
}