Commit 39c6d082 authored by Loren Merritt's avatar Loren Merritt
Browse files

remove TODO. most of it is done, and the rest is out of date.

git-svn-id: svn:// df754926-b1dd-0310-bc7b-ec298dee348c
parent 918a791b
It is far from complete, anyway :
* CABAC: check if adaptive model is really working. (I didn't see any improvments)
* Field support : no and I probably won't do it.
* Slice A/B/C (partion): is there any interest doing it ? (Shouldn't be hard).
- extend x264_t
- review x264_macroblock_write_cavlc
* Intra encoding:
- in I_4x4 mode, some predict mode aren't tested and not supported :
when some pixels around are unavailble but could be predicted from
others. (see the norm)
* Inter coding:
- D_4x8 D_8x4 and D_4x4 ME P block -> done but too slow.
- B_SUB8x8 types.
- better SKIP detection.
- long terme ?
- ...
* B frame: B_L0/L1/BI/SKIP/DIRECT work (need more testing).
-> adaptive placement of B-frames
-> look at weighted prediction (should give better result)
-> better analyse algo (as always ;)
-> joint motion estimation of BI types
* Speed issue (oprofile is your friend)
- mc.c:30% and pixel.c:20% (mc is used by ME)
- Motion Estimation -> try better/faster algos.
- loop filter
- stream writing (bs)
- ...
* Time spent: (test file: 720x576, mmx, mmxext)
CAVLC: analyse=73% encode=15% write=4% filter=6%
CABAC: analyse=69% encode=16% write=8% filter=5%
* Limitations:
- frame width/height %16 == 0 only.
* ...
* Currently decoder/* won't even compile, and anyway is unusable.
* Finish to port gcc inlined asm to nasm file.
Coding issue:
* table : somes are duplicated -> find a solution (easy).
* documentations ? (mouaaaarfff ;)
* ...
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment