| 1 |
todo |
| 2 |
|
| 3 |
|
| 4 |
*** build environment *** |
| 5 |
- portable makefile [DONE] |
| 6 |
more testing needed |
| 7 |
|
| 8 |
|
| 9 |
*** api *** |
| 10 |
- decore/encore2 wrappers; [DONE] |
| 11 |
|
| 12 |
- xvid_init() [DONE] |
| 13 |
handle cpu detection, yuv/dct lookup table init |
| 14 |
return api version number |
| 15 |
return core build number [build number should be incremented automatically, how?] |
| 16 |
|
| 17 |
|
| 18 |
*** common *** |
| 19 |
|
| 20 |
- mpeg-4 quantizer (quant4) testing [DONE] |
| 21 |
- consder moving a modified mbprediction/mbdeprediction here [DONE] |
| 22 |
|
| 23 |
- add [-2048,2048] to mmx dequant [DONE] |
| 24 |
|
| 25 |
|
| 26 |
- mmx quant4 [DONE] |
| 27 |
- sse2: convert intel's idct/sad/comp apnotes to nasm [IN PROGRESS] |
| 28 |
- mlib |
| 29 |
|
| 30 |
|
| 31 |
*** image *** |
| 32 |
|
| 33 |
- sort out in/out & conversion func naming issues [DONE] |
| 34 |
maybe, in_yuv, in_yuyv, in_uyuv, in_rgbXX |
| 35 |
out_yuv, out_yuyv, out_uyuv, out_rgbXX |
| 36 |
|
| 37 |
- input flipping support |
| 38 |
output flipping support is not 'perfect'. |
| 39 |
using negative height is bad, best to have seperate vflip parameter |
| 40 |
|
| 41 |
- rgb24/32_mmx cause yellow dots to appear on left hand size [DONE] |
| 42 |
(only visibile with high contrast colors) |
| 43 |
|
| 44 |
- rgb16. [DONE] |
| 45 |
can be done generically using r,g,b masks |
| 46 |
or just code for the popular 16bit formats 555, 565 |
| 47 |
|
| 48 |
- diff between basic-yuv and iso-yuv |
| 49 |
|
| 50 |
|
| 51 |
*** decoder *** |
| 52 |
|
| 53 |
- quant=1 encoder material appears "soft" [FIXED] |
| 54 |
- block based interpolation [DONE] |
| 55 |
|
| 56 |
- msmpeg-4 support... dont hold your breath [IN PROGRESS] |
| 57 |
best to write an entirely new 'decoder' than bolt stuff onto |
| 58 |
the existing one. |
| 59 |
|
| 60 |
- post processing |
| 61 |
|
| 62 |
- add basic mpeg4 stuff: |
| 63 |
- custom mpeg quant matrixes [DONE] |
| 64 |
- intra_dc_vlc_threshold [DONE] |
| 65 |
- quarterpel [in progress] |
| 66 |
- resync? |
| 67 |
|
| 68 |
- b-vops [IN PROGRESS] |
| 69 |
- figure out what todo with time codes, and fatal error |
| 70 |
|
| 71 |
|
| 72 |
|
| 73 |
*** encoder *** |
| 74 |
|
| 75 |
- purple mb block (rare) [FIXED] |
| 76 |
- coeff range checks need to be verified against mpeg-4 spec [DONE/PARTIAL] |
| 77 |
- dequants now clamp to [-2048,2047]. [DONE] |
| 78 |
- fdct_mmx clamps to [-2048,2047], it doesnt have to (but prevents overflows in mmx quantization) |
| 79 |
- quants _should_ clamp to [-127,127], but this introduces errors at low quant levels so we ignore it |
| 80 |
- idct _should_ clamp [-256,255], but its really unnessary overhead. |
| 81 |
|
| 82 |
- pmvfast bug fixing [DONE] |
| 83 |
|
| 84 |
- quater pixel motion search |
| 85 |
|
| 86 |
- 'smearing bug' when motion occurs in straight vertical direction ( |
| 87 |
http://www.videocoding.de/forum/viewtopic.php?topic=71&forum=2&6 ) |
| 88 |
coefficient thresholding problems? [FIXED/REMOVED] |
| 89 |
|
| 90 |
- better luminance masking [DONE] |
| 91 |
|
| 92 |
- use N_VOP when a interframe is entirely composed of ublks |
| 93 |
& frame dropping. eg. simple ublk threshold |
| 94 |
|
| 95 |
|
| 96 |
|
| 97 |
*** gpl conversion *** |
| 98 |
|
| 99 |
mbdecoding.c |
| 100 |
|
| 101 |
encoder.c [REPLACED] |
| 102 |
mbmotionestcomp.c [REPLACED] |
| 103 |
mbcoding [REPLACED] |
| 104 |
putvlc.c [REPLACED] |
| 105 |
ratecontol.c |
