Opened 13 months ago

Last modified 13 months ago

#10595 open defect

photocd file fails to decode

Reported by: ami_stuff Owned by:
Priority: normal Component: avcodec
Version: git-master Keywords: photocd
Cc: Blocked By:
Blocking: Reproduced by developer: no
Analyzed by developer: no

Description

https://telparia.com/fileFormatSamples/image/pcd/maddog.pcd

./ffmpeg -i maddog.pcd -y out.jpg
ffmpeg version N-112164-gcb26b85953 Copyright (c) 2000-2023 the FFmpeg developers
  built with gcc 11 (Ubuntu 11.4.0-1ubuntu1~22.04)
  configuration: 
  libavutil      58. 25.100 / 58. 25.100
  libavcodec     60. 27.100 / 60. 27.100
  libavformat    60. 13.100 / 60. 13.100
  libavdevice    60.  2.101 / 60.  2.101
  libavfilter     9. 11.100 /  9. 11.100
  libswscale      7.  3.100 /  7.  3.100
  libswresample   4. 11.100 /  4. 11.100
Input #0, image2, from 'maddog.pcd':
  Duration: 00:00:00.04, start: 0.000000, bitrate: 867942 kb/s
  Stream #0:0: Video: photocd, yuv420p(pc, bt709/bt709/iec61966-2-1), 3072x2048, 25 fps, 25 tbr, 25 tbn
Stream mapping:
  Stream #0:0 -> #0:0 (photocd (native) -> mjpeg (native))
Press [q] to stop, [?] for help
[vist#0:0/photocd @ 0x55c65327b900] Decoding error: Invalid data found when processing input
[swscaler @ 0x55c6532b5140] deprecated pixel format used, make sure you did set range correctly
[vost#0:0/mjpeg @ 0x55c653280380] No filtered frames for output stream, trying to initialize anyway.
[vost#0:0/mjpeg @ 0x55c653280380] No information about the input framerate is available. Falling back to a default value of 25fps. Use the -r option if you want a different framerate.
Output #0, image2, to 'out.jpg':
  Metadata:
    encoder         : Lavf60.13.100
  Stream #0:0: Video: mjpeg, yuvj420p(pc, progressive), 3072x2048, q=2-31, 200 kb/s, 25 fps, 25 tbn
    Metadata:
      encoder         : Lavc60.27.100 mjpeg
    Side data:
      cpb: bitrate max/min/avg: 0/0/200000 buffer size: 0 vbv_delay: N/A
[vist#0:0/photocd @ 0x55c65327b900] Decode error rate 1 exceeds maximum 0.666667
[out#0/image2 @ 0x55c65327be80] video:0kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: unknown
[out#0/image2 @ 0x55c65327be80] Output file is empty, nothing was encoded(check -ss / -t / -frames parameters if used)
frame=    0 fps=0.0 q=0.0 Lsize=N/A time=N/A bitrate=N/A speed=N/A    
Conversion failed!

Change History (13)

comment:1 by ami_stuff, 13 months ago

Version: unspecifiedgit-master

comment:2 by Elon Musk, 13 months ago

And what software decodes it?

comment:4 by Balling, 13 months ago

Status: newopen

Yep. Other pcd files work great (except for color management, but whatever).

comment:5 by Elon Musk, 13 months ago

Here it decodes always same colors.

in reply to:  5 comment:6 by Balling, 13 months ago

Replying to Elon Musk:

Here it decodes always same colors.

Which are wrong colors. I told you years ago it uses xvYCC, not sRGB transfer. It is like you cannot read the English language.

https://en.wikipedia.org/wiki/XvYCC

comment:7 by Elon Musk, 13 months ago

MPV displays correct colors, better one then that app that converts it to rgb ppm.

FFplay does not support most of colorspace stuff.

comment:8 by Balling, 13 months ago

MPV displays correct colors

No. You need PQ to correctly show the colors, as the content is HDR and WCG (both outside 100 nits and BT.709 primaries).

https://github.com/FFmpeg/FFmpeg/blob/8eb094adb2ac3b6ea1d2cdd0fdfc7b69b2084db9/libavcodec/photocd.c#L430

Change that to AVCOL_TRC_IEC61966_2_4. Part 4, not 1. Then implement correct quantisation: linear RGB values are from -0.20 to 2.00, while nonlinear RGB' are from -0.43357 to 1.402278, where reference white (perfect, non-fluorescent, white-reflecting diffuser in the original scene) is 1.0, just like in BT.709. That is done by a different MATRIX THAT IS NOT just avctx->colorspace = AVCOL_SPC_BT709; and avctx->color_range = AVCOL_RANGE_JPEG.

There is also a matrix from and to sRGB in Windows XP source code that leaked. Again, I told you all that 3 years ago. Open your eyes: https://trac.ffmpeg.org/ticket/5923#comment:7

comment:9 by Elon Musk, 13 months ago

Your entry is again, wrong, and full of errors.
Your understanding of this subject is pretty invalid.
Otherwise you would post patch(es) to solve this already.

comment:10 by Balling, 13 months ago

Continue denying reality and not seeing a patch in that issue that implements own ff_yuv2rgb_coeffs.

comment:11 by Balling, 13 months ago

https://imgsli.com/MjE1MjQ0

You ARE denying reality, BTW. (On the left image from https://r0k.us/graphics/kodak/kodim23.html on the right original PCD converted to PNG with sRGB chunk https://www.math.purdue.edu/~lucier/PHOTO_CD/IMAGES/.)

Version 0, edited 13 months ago by Balling (next)

comment:12 by Elon Musk, 13 months ago

decoder outputs in yuv not rgb.

comment:13 by Balling, 13 months ago

decoder outputs in yuv not rgb.

That is why the code changes happen in pixfmt.h and https://trac.ffmpeg.org/attachment/ticket/5923/ycc.diff

Last edited 13 months ago by Balling (previous) (diff)
Note: See TracTickets for help on using tickets.