Opened 6 years ago

Last modified 6 years ago

#7258 new defect

Video codec "uncompressed r210 and v210" in Quicktime container produces too dark images

Reported by: wavelet Owned by:
Priority: normal Component: undetermined
Version: git-master Keywords: dnxhd
Cc: Blocked By:
Blocking: Reproduced by developer: no
Analyzed by developer: no

Description

Summary of the bug:
How to reproduce:

% ffmpeg -i input.mov -vcodec r210 -movflags +writecolr output.mov
% ffmpeg -i input.mov -vcodec v210  -movflags +writecolr output.mov
ffmpeg version: N-90169-gf4709f1b7b
Windows: Dll's from zernoe
8bpc and 16bpc exports are working fine: % ffmpeg -i input.mov -vcodec rawvideo -pix_fmt rgb48be  -movflags +writecolr  output.mov

Attachments (1)

SourceVSR210_version2.jpg (98.6 KB ) - added by wavelet 6 years ago.
R210 vs Source file

Download all attachments as: .zip

Change History (12)

comment:1 by wavelet, 6 years ago

In BM Resolve the brightness can be normalized by forcing the color_range to full. There is also a high risk that something with the 32Bit packing is wrong.

comment:2 by Carl Eugen Hoyos, 6 years ago

Keywords: color_range removed

Please provide a short input file and the command line you tested together with the complete, uncut console output to make this a valid ticket.

comment:3 by wavelet, 6 years ago

ffmpeg.exe -i J:\tmp\1980p_test_sequence.mov -vcodec r210 J:\tmp\R210_FFMPEG_1080p.mov
90169-gf4709f1b7b Copyright (c) 2000-2018 the FFmpeg developers
7.3.0 (GCC)
--disable-static --enable-shared --enable-gpl --enable-version3 --enable-sdl2 --enable-bzl
s --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libtheora --enable-libtwo
-amrwbenc --enable-libmysofa --enable-libspeex --enable-libxvid --enable-libmfx --enable-a

  1. 7.101 / 56. 7.101
  2. 13.100 / 58. 13.100
  3. 10.100 / 58. 10.100
  4. 2.100 / 58. 2.100
    1. 12.100 / 7. 12.100
    2. 0.101 / 5. 0.101
    3. 0.101 / 3. 0.101
  5. 0.100 / 55. 0.100

,m4a,3gp,3g2,mj2, from 'J:\tmp\1980p_test_sequence.mov':

: qt
: 512

ands: qt

: Lavf57.25.100

:00.38, start: 0.000000, bitrate: 177741 kb/s
ng): Video: dnxhd (DNXHD) (AVdn / 0x6E645641), yuv422p10le(tv, bt709), 1920x1080, 176160 k

e : Core Media Data Handler

: 01:00:00:00

ng): Audio: pcm_s16le (sowt / 0x74776F73), 48000 Hz, stereo, s16, 1536 kb/s (default)

e : Core Media Data Handler
ng): Data: none (tmcd / 0x64636D74), 0 kb/s

e : Core Media Data Handler

: 01:00:00:00

_FFMPEG_1080p.mov' already exists. Overwrite ? [y/N] y

#0:0 (dnxhd (native) -> r210 (native))
#0:1 (pcm_s16le (native) -> aac (native))
, ? for help
o 'J:\tmp\R210_FFMPEG_1080p.mov':

: qt
: 512

ands: qt

: Lavf58.10.100

ng): Video: r210 (r210 / 0x30313272), rgb48le(progressive), 1920x1080 [SAR 1:1 DAR 16:9],

e : Core Media Data Handler

: 01:00:00:00
: Lavc58.13.100 r210

ng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 128 kb/s (default)

e : Core Media Data Handler

: Lavc58.13.100 aac

.0 q=-0.0 Lsize= 72902kB time=00:00:00.38 bitrate=1555248.3kbits/s speed=1.12x
io:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.002932%
942000] Qavg: 65536.000

comment:4 by wavelet, 6 years ago

Source video: http://Projekte.waveletbeam.com/1980p_test_sequence.mov
Transcoded r210 video (too dark): http://Projekte.waveletbeam.com/R210_FFMPEG_1080p.mov
Source vs r210: http://Projekte.waveletbeam.com/SourceVSR210.jpg

8bpc and 10bpc are working fine! Only the 10bpc exports are showing this behavior.

comment:5 by Carl Eugen Hoyos, 6 years ago

How did you create 1980p_test_sequence.mov? Do you find it likely that it is supposed to look as in the jpg you provided?

comment:6 by wavelet, 6 years ago

The 1980p_test_sequence.mov was created with BM Resolve(1920x1080). I took a tiff sequence and used the DNxHD codec. You can reproduce the same results with other sources like PoRes in mov container too.

The jpg http://projekte.waveletbeam.com/SourceVSR210.jpg is showing the wrong behavior. The left side is the R210 export and the right side is the source material. You can clearly recognize that the brightness ramp on the left side (upper left corner) is much darker as the same area on the right side.

comment:7 by wavelet, 6 years ago

Sorry, I mixed up the left and right image. I will exchange the image

in reply to:  6 ; comment:8 by Carl Eugen Hoyos, 6 years ago

Replying to wavelet:

The 1980p_test_sequence.mov was created with BM Resolve(1920x1080).

Are you sure? If the console output you provided above were readable, it would indicate that the file was created with FFmpeg.

by wavelet, 6 years ago

Attachment: SourceVSR210_version2.jpg added

R210 vs Source file

in reply to:  8 comment:9 by wavelet, 6 years ago

Replying to cehoyos:

Replying to wavelet:

The 1980p_test_sequence.mov was created with BM Resolve(1920x1080).

Are you sure? If the console output you provided above were readable, it would indicate that the file was created with FFmpeg.

Dirk Hildebrandt: Maybe BM is also using FFMPEG. Anyhow the the too dark images are at the R210 export

comment:10 by Carl Eugen Hoyos, 6 years ago

Keywords: dnxhd added
Version: unspecifiedgit-master

I opened the input video with QuickTime player using Avid codecs and it looks much darker than on the jpg you provided.

in reply to:  10 comment:11 by wavelet, 6 years ago

Replying to cehoyos:

I opened the input video with QuickTime player using Avid codecs and it looks much darker than on the jpg you provided.

Dirk Hildebrandt: Carl Eugen you are absolutely right, it is important to check the input first. Also maybe the interpretation of the export format is wrong at the player or postprocessing software. To make it a little bit easier, I generated a ProRes version for the source:

Source: http://Projekte.waveletbeam.com/1080p_test_sequence_ProRes.mov
r210:http://projekte.waveletbeam.com/R210_FFMPEG_1080pFromProRes.mov

I hope this helps.

Note: See TracTickets for help on using tickets.