[Ffmpeg-devel] [PATCH] DV weights (original) (raw)
Roman Shaposhnik Roman.Shaposhnick
Mon Feb 27 17:05:46 CET 2006
- Previous message: [Ffmpeg-devel] [PATCH] DV weights
- Next message: [Ffmpeg-devel] [PATCH] DV image quality improvement
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Sat, Feb 25, 2006 at 02:41:33PM -0500, Dan Maas wrote:
This patch implements AC coefficient weighing (per SMPTE 314M) for the DV encoder and decoder.
I have also attached the Python script I used to make the tables. Weighing is implemented as fixed-point multiplication. I have chosen the number of bits of precision that gives the smallest error without causing 32-bit overflows. Weighing is inherently a lossy process because weighted coefficients have fewer bits than unweighted ones. But, in no case is the error greater than one least-significant bit. (the Python script checks this) Performance impact is less than 1%.
Applied. With one minor nitpick -- next time, please include updated regression golden files as well.
Thanks, Roman.
- Previous message: [Ffmpeg-devel] [PATCH] DV weights
- Next message: [Ffmpeg-devel] [PATCH] DV image quality improvement
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]