mirror of
https://github.com/webmproject/libwebp.git
synced 2025-01-02 16:58:23 +01:00
6a37a2aaa9
When FANCY_UPSCALING is defined, use a smoothing filter for upscaling the U/V chroma fields. The filter used is a separable t[1 3 3 1] x [1 3 3 1] filter. It can be easily changed in macros MIX_*. The upscaling code reside on the thing shell between user and core decoding (in webp.c), and not in the core decoder. As such, this smoothing process can still be offloaded to GPU in some future and is not integral part of the decoding process. Coincidentaly: changed the way data is tranfered to user. For profile 2 (no filtering), it used to be on a per-block basis. Now, for all profiles, we emit rows of pixels (between 8 and 24 in height) when they are ready. This makes the upscaling code much easier. Will update the test vectors MD5 sums soon (as they'll be broken after this change) Change-Id: I2640ff12596cb8b843a4a376d7347447d9b9f778 |
||
---|---|---|
.. | ||
decode_vp8.h | ||
decode.h |