Do not distort images scaled by integer multiples

This commit is contained in:
sfan5 2021-03-31 16:55:48 +02:00
parent 3fb23fc423
commit d7127df9f8
1 changed files with 20 additions and 4 deletions

View File

@ -224,13 +224,29 @@ void CImage::copyToScaling(void* target, u32 width, u32 height, ECOLOR_FORMAT fo
// Similar for y.
// As scaling is done without any antialiasing it doesn't matter too much which outermost pixels we use and keeping
// border pixels intact is probably mostly better (with AA the other solution would be more correct).
const f32 sourceXStep = width > 1 ? (f32)(Size.Width-1) / (f32)(width-1) : 0.f;
const f32 sourceYStep = height > 1 ? (f32)(Size.Height-1) / (f32)(height-1) : 0.f;
// This is however unnecessary (and unexpected) for scaling to integer multiples, so don't do it there.
f32 sourceXStep, sourceYStep;
f32 sourceXStart = 0.f, sourceYStart = 0.f;
if (width % Size.Width == 0)
sourceXStep = (f32)(Size.Width) / (f32)(width);
else
{
sourceXStep = width > 1 ? (f32)(Size.Width-1) / (f32)(width-1) : 0.f;
sourceXStart = 0.5f; // for rounding to nearest pixel
}
if (height % Size.Height == 0)
sourceYStep = (f32)(Size.Height) / (f32)(height);
else
{
sourceYStep = height > 1 ? (f32)(Size.Height-1) / (f32)(height-1) : 0.f;
sourceYStart = 0.5f; // for rounding to nearest pixel
}
s32 yval=0, syval=0;
f32 sy = 0.5f; // for rounding to nearest pixel
f32 sy = sourceYStart;
for (u32 y=0; y<height; ++y)
{
f32 sx = 0.5f; // for rounding to nearest pixel
f32 sx = sourceXStart;
for (u32 x=0; x<width; ++x)
{
CColorConverter::convert_viaFormat(Data+ syval + ((s32)sx)*BytesPerPixel, Format, 1, ((u8*)target)+ yval + (x*bpp), format);