introduce a special quantization function for WHT

WHT is somewhat a special case: no sharpen[] bias, etc.
Will be useful in a later CL when precision of input is changed.

Change-Id: I851b06deb94abdfc1ef00acafb8aa731801b4299
This commit is contained in:
skal
2013-12-10 14:21:47 +01:00
parent 41c0cc4b9a
commit 73b731fb42
4 changed files with 41 additions and 1 deletions

View File

@ -930,6 +930,11 @@ static int QuantizeBlockSSE2(int16_t in[16], int16_t out[16],
}
}
static int QuantizeBlockWHTSSE2(int16_t in[16], int16_t out[16],
const VP8Matrix* const mtx) {
return QuantizeBlockSSE2(in, out, 0, mtx);
}
#endif // WEBP_USE_SSE2
//------------------------------------------------------------------------------
@ -941,6 +946,7 @@ void VP8EncDspInitSSE2(void) {
#if defined(WEBP_USE_SSE2)
VP8CollectHistogram = CollectHistogramSSE2;
VP8EncQuantizeBlock = QuantizeBlockSSE2;
VP8EncQuantizeBlockWHT = QuantizeBlockWHTSSE2;
VP8ITransform = ITransformSSE2;
VP8FTransform = FTransformSSE2;
VP8FTransformWHT = FTransformWHTSSE2;