Changes between Version 41 and Version 42 of colorspace


Ignore:
Timestamp:
Oct 21, 2024, 3:28:32 AM (22 months ago)
Author:
MasterQuestionable
Comment:

# ͏"full_chroma_in*" related

Legend:

Unmodified
Added
Removed
Modified
  • colorspace

    v41 v42  
    367367https://github.com/MasterInQuestion/attach/raw/main/@ext/trac.ffmpeg.org/wiki/colorspace/colorspace_yuv444p10le-yuv420p.webp
    368368
    369 === Hint about "[https://ffmpeg.org/ffmpeg-scaler.html#Scaler-Options full_chroma_inp]"[=#Hintaboutfull_chroma_inp] === #full_chroma_inp
    370 
     369=== "full_chroma_in*" related[[span(id=full_chroma_in*)]] === #full_chroma
     370
     371[ haasn @ CE 2024-08-16 09:43:37 UTC:\\
     372https://github.com/FFmpeg/FFmpeg/commit/3e064f52eb368a373ded6e3704fcf29f1db3ff12
     373
     374[ "SWS_FULL_CHR_H_INT"\\
     375͏    Perform full chroma upsampling when upscaling to RGB.
     376
     377͏    For example, when converting 50x50 yuv420p to 100x100 rgba:\\
     378͏    Setting this flag will scale the chroma plane from 25x25 to 100x100 (4:4:4):\\
     379͏    And then convert the 100x100 yuv444p image to rgba in the final output step.\\
     380͏    .\\
     381͏    Without this flag, the chroma plane is instead scaled to 50x100 (4:2:2):\\
     382͏    With a single chroma sample being re-used, for both of the horizontally adjacent RGBA output pixels. ]
     383
     384[ "SWS_FULL_CHR_H_INP"\\
     385͏    Perform full chroma interpolation when downscaling RGB sources.
     386
     387͏    For example, when converting a 100x100 rgba source to 50x50 yuv444p:\\
     388͏    Setting this flag will generate a 100x100 (4:4:4) chroma plane, which is then downscaled to the required 50x50.\\
     389͏    .\\
     390͏    Without this flag, the chroma plane is instead generated at 50x100 (dropping every other pixel):\\
     391͏    Before then being downscaled to the required 50x50 resolution. ] ]
     392
     393͏----
     394
     395[ MasterQuestionable:\\
     396͏    The description is much misleading... if not outright misinformation?\\
     397͏    |1| Both may be RGB irrelevant: occurs during YUV unsubsampling.\\
     398͏    |2| SWS_FULL_CHR_H_INT may involve down-scaling too.\\
     399͏    |3| [\\
     400͏    SWS_FULL_CHR_H_INP asserts full-chroma input; which then forces use of full-chroma interpolation: whether sensible or not.\\
     401͏    Which is actually somewhat useless: as the default handling seems... sensible? ]
     402
     403͏    Hint about [[span( id=full_chroma_inp, [https://ffmpeg.org/ffmpeg-scaler.html#Scaler-Options ͏"full_chroma_inp"] )]]:\\
    371404͏    https://github.com/search?type=code&q=repo:FFmpeg/FFmpeg+%22SWS_FULL_CHR_H_INP%22\\
    372 ͏    https://github.com/FFmpeg/FFmpeg/blob/4ec45aca36ff6c7c7767536ce7f4a855f8b37fb0/libswscale/utils.c#L1530-L1546\\
     405͏    https://github.com/FFmpeg/FFmpeg/blob/08c632ec0f0c36affd2edad2e0add6e046d6c8be/libswscale/utils.c#L1540-L1556\\
    373406͏    https://github.com/FFmpeg/FFmpeg/blob/41ed7ab45fc693f7d7fc35664c0233f4c32d69bb/doc/swscale.txt#L50-L57
     407
     408͏    On "full_chroma_int":\\
     409͏    https://github.com/MasterInQuestion/attach/commit/e18e5cd26b9d27bacefa685ba70d0d41bc7d7cbc#diff-75a0130f1814ba26c23e3a6fc6ce5046620163061d392086bc55e2c13c145bf9L264 ]