Page Comparison

FFmpeg is frequently used by different studios for encoding their media, however the documentation for ffmpeg is often poor, or cryptic so its often harder than it should be to come up with a good starting point. We are aiming to come up with recommendations for different scenarios as well as document what the different flags are doing with the aim to make this easier to get to a good baseline.

Overview

We are looking for recommendations for the following:

...

Where ffmpeg arguments, it would be great to document why we are using them, rather than ending up with a recipe. 

Presentation Links

Overall workflow

For this paper, we are assuming that we are encoding from a file-sequence of frames into a movie (rather than re-encoding a movie), but we are also assuming almost all of the colorspace work would be done outside of ffmpeg, with tools using the OCIO library. Examples could include nuke and oiio. Once we get to ffmpeg, the goal being that the pixel data we get in, should be as close as possible to the data we get out of ffmpeg. However, there are still quite a few areas that a ffmpeg user could go wrong, which we break down below. 

TODO: Find examples of overall workflow. 

Codec Selection.

h264

Key flags (see https://trac.ffmpeg.org/wiki/Encode/H.264 ) 

• -crf 23 - This is the constant rate factor, controlling the default quality (see: https://slhck.info/video/2017/02/24/crf-guide.html ) where -crf 0 is uncompressed. By default this is set to 23, which is probably good enough for our needs.
• -qp 23 - Quantization Parameter - it is recommended that you do not use this, in preference to -crf above (see: https://slhck.info/video/2017/03/01/rate-control.html )
• -preset slower - https://trac.ffmpeg.org/wiki/Encode/H.264#FAQ
• -tune film - Optionally use the tune option to change settings based on specific inputs - https://trac.ffmpeg.org/wiki/Encode/H.264#FAQ - qscale:v 1 - Generic quality scale flag:see also: https://www.ffmpeg.org/ffmpeg.html#toc-Main-options - not sure if its needed?

-preset slow -crf 18   level 4 profile high.

RV - 10bit YUV444.

Better to stick with yuv – 

FFMPEG.

TODO:

• Suggestions for max-bitrate?
• Suggestions for preset - ? slow
• Suggestions for tune

h265/HVEC

...

• superuser.com/questions/564402/explanation-of-x264-tune I suspect that we would want to use one of:
  1. -tune film    good for live action content.
  2. -tune animation good for animated content with areas of flat colors.
  3. -tune grain   good for live action content where you want to preserve the grain as much as possible.
• -qscale:v 1 - Generic quality scale flag: https://

...

• www.ffmpeg.

...

links:

• https://codecalamity.com/org/ffmpeg.html#toc-Main-options - not sure if its needed?

-preset slow -crf 18   level 4 profile high.

RV - 10bit YUV444.

Better to stick with yuv – 

FFMPEG.

TODO:

• Suggestions for max-bitrate?
• Suggestions for preset - ? slow
• Suggestions for tune

h265/HVEC

Support: https://caniuse.com/hevc (or https://www.chromium.org/audio-video ) currently no support for h265 on chrome or chromium based browsers.

links:

• https://codecalamity.com/encoding-uhd-4k-hdr10-videos-with-ffmpeg/
• https://codecalamity.com/encoding-settings-for-hdr-4k-videos-using-10-bit-x265/
• https://support.frame.io/en/articles/4305241-creating-hdr-files-for-frame-io

VMAF

I did explore using VMAF - Video Multi-Method Assessment Fusion as a way to quantify the compression, the notes for setting this up are below, however I think we are going with a fairly high compression factor , so I think this is probably not really going to help us much.

...

• https://

...

• stackoverflow.com/

...

• questions/69251960/how-can-i-encode-rgb-images-into-hdr10-videos-in-ffmpeg-command-line
• https://brandur.org/fragments/ffmpeg-h265

HDR

• https://

...

• developer.apple.com/

...

• av-

...

• foundation/High-

...

• Dynamic-

...

• Range-

...

• Metadata-

...

• for-

...

• Apple-

...

• Devices.pdf
• https://

...

• www.avsforum.com/

...

Color Preservation

Testing Methodology

Converting SMPTE color bars to the compressed movie, using ffmpeg to expand and then compare with OIIO. NOTE, for compression schemes that are not 444 we may need to mask the transitions.

...

• threads/open-source-video-testing-calibration-patterns.2944378/
• https://github.com/test-full-band/tfb-video/releases
• https://trev16.hatenablog.com/entry/2021/07/23/145725 – good site for HDR test sites.

ProRes

There are four Prores encoders, Prores, Prores_ks, Prores_aw and now with ffmpeg 5 VideoToolBox Prores, which is a hardware based OSX M1 encoder/decoder. 

From https://trac.ffmpeg.org/wiki/Encode/VFX the recommendation is to use Prores_ks with -profile:v 3 and the qscale of 11

Options that can be used include:

• -profile:v values can be one of.
  • proxy (0)
  • lt (1)
  • standard (2)
  • hq (3)
  • 4444 (4)
  • 4444xq (5)
• -qscale:v between values of 9 - 13 give a good result, 0 being best.
• -vendor apl0 - tricks the codec into believing its from an Apple codec.

Using this with the usual color space flags, seems to work well with the exception of ffmpeg itself, which needs the flags:-vf scale=in_color_matrix=bt709:out_color_matrix=bt709 added to the command to ensure the right input colorspace is recognised, e.g.:

ffmpeg.exe -i INPUTFILE.mov -compression_level 10 -pred mixed -pix_fmt rgba64be -sws_flags spline+accurate_rnd+full_chroma_int -vframes 1 -vf scale=in_color_matrix=bt709:out_color_matrix=bt709 OUTPUTFILE.png

However, other encoders seem to be recognised correctly, so there is clearly some metadata missing. I did try using the prores_metadata filter to try adding some additional parameters, but it didnt seem to help.

ffmpeg.exe -i ./chip-chart-yuvconvert\basicnclc.mov -c copy -bsf:v prores_metadata=color_primaries=bt709:color_trc=bt709:colorspace=bt709 chip-chart-yuvconvert\basicnclcmetadata.mov

TODO:

• Figure out the missing metadata.
• Wedge qscale values
• Do some colorspace tests with different qscale values to see where color breaks down.

VMAF

I did explore using VMAF - Video Multi-Method Assessment Fusion as a way to quantify the compression, the notes for setting this up are below, however I think we are going with a fairly high compression factor , so I think this is probably not really going to help us much.

https://github.com/Netflix/vmaf

https://jina-liu.medium.com/a-practical-guide-for-vmaf-481b4d420d9c

https://netflixtechblog.com/toward-a-practical-perceptual-video-quality-metric-653f208b9652

https://ottverse.com/vmaf-ffmpeg-ubuntu-compilation-installation-usage-guide/ - building VMAF on ubuntu.

Color Preservation

Testing Methodology

Converting SMPTE color bars to the compressed movie, using ffmpeg to expand and then compare with OIIO. NOTE, for compression schemes that are not 444 we may need to mask the transitions.

Testing loading the compressed movie in to RV, Firefox, VLC, Avid, resolve, , to compare the resulting color transformation - not sure if there is a procedural way to run this?

...

Q: Currently focusing just on color matching in vs. out, but should also do EXR ACEScg in to resulting movie. Feels like we should also bless full pipeline, e.g.: Reference "Dailies script" https://github.com/jedypod/generate-dailies

Test Sources

SMPTE test chart: https://commons.wikimedia.org/wiki/File:SMPTE_Color_Bars_16x9.svg 

...

C: Do as clean a conversion from RGB to YUV as possible.

RGB/YCrCb Colorspace Conversion  

As a rule of thumb, we would like ffmpeg to do as little as possible in terms of color space conversion. i.e. what comes in goes out. The problem is that most of the codecs are doing some sort of RGB to YUV conversion (technically YCrCb). The notable exception is x264rgb (see above). For more information, see: https://trac.ffmpeg.org/wiki/colorspace 

For examples comparing these see: https://richardssam.github.io/ffmpeg-tests/tests/chip-chart-yuvconvert/compare.html

colormatrix filter

ffmpeg -y -i ../sourceimages/chip-chart-1080-noicc.png -sws_flags spline+accurate_rnd+full_chroma_int -vf "colormatrix=bt470bg:bt709" -c:v libx264 -preset placebo -qp 0 -x264-params "keyint=15:no-deblock=1" -pix_fmt yuv444p10le -qscale:v 1 -color_range 1 -colorspace 1 -color_primaries 1 -color_trc 1 ./chip-chart-yuvconvert/spline444colormatrix2.mp4

This is the most basic colorspace filtering. bt470bg is essentially part of the bt601 spec.  See: https://www.ffmpeg.org/ffmpeg-filters.html#colormatrix

colorspace filter

ffmpeg -y -i ../sourceimages/chip-chart-1080-noicc.png -sws_flags spline+accurate_rnd+full_chroma_int -vf "colorspace=bt709:iall=bt601-6-625:fast=1" -c:v libx264 -preset placebo -qp 0 -x264-params "keyint=15:no-deblock=1" -pix_fmt yuv444p10le -qscale:v 1 -color_range 1 -colorspace 1 -color_primaries 1 -color_trc 1 ./chip-chart-yuvconvert/spline444colorspace.mp4

Using colorspace filter, better quality filter, SIMD so faster too, can support 10-bit too.  The second part -vf "colorspace=bt709:iall=bt601-6-625:fast=1" encodes for the output being bt709, rather than the default bt601 matrix. iall=bt601-6-625 says to treat all the input (colorspace, primaries and transfer function) with the bt601-6-625 label). fast=1 skips gamma/primary conversion in a mathematically correct way.  See:  https://ffmpeg.org/ffmpeg-filters.html#colorspace

libswscale filter

ffmpeg -y -i ../sourceimages/chip-chart-1080-noicc.png -sws_flags spline+accurate_rnd+full_chroma_int+full_chroma_inp -vf "scale=in_range=full:in_color_matrix=bt709:out_range=tv:out_color_matrix=bt709" -c:v libx264 -preset placebo -qp 0 -x264-params "keyint=15:no-deblock=1" -pix_fmt yuv444p10le -qscale:v 1 -color_range 1 -colorspace 1 -color_primaries 1

...

 -color_trc 1 ./chip-chart-yuvconvert/spline444out_color_matrix.mp4

Using the libswscale library. Seems similar to colorspace, but with image resizing, and levels built in.  https://www.ffmpeg.org/ffmpeg-filters.html#scale-1

This is the recommended filter.

Color Metadata NCLC/NCLX

The above gets the underlying data stored correctly, but there are additional metadata flags that can be set that are interpreted by some players, these are the NCLC color tags for color primaries, transfer function and conversion matrix.  This is defined as a ISO spec here (see https://www.ffmpegiso.org/ffmpeg-filters.html#scale-1

This is the recommended filter.

Color Metadata NCLC

The above gets the underlying data stored correctly, but there are additional metadata flags that can be set that are interpreted by some players, these are the NCLC color tags for color primaries, transfer function and conversion matrix.  This is defined as a ISO spec here (see https://www.iso.org/standard/73412.html). The numbers below are part of the definition.standard/73412.html). The numbers below are part of the definition.

NCLC stands for Non-Consistent Luminance Coding, a brief overview of its history is here. For MP4 files, its also known as NCLX. Additionally this metadata can also be represented in the h264 metadata stream in the video usability Information (VUI) block. 

You can read the metadata using mp4box.js which is a visual browser of the mp4 metadata, and look at moov/trak/mdia/minf/stbl/stsd/avc1/colr

NOTE: None of the flags below affect the encoding of the source imagery, they are meant to be used to guide how the mp4 file is decoded.

...

The docs are pretty sparse for this, some of the better info is FFmpeg/pixfmt.h at master · FFmpeg/FFmpeg · GitHub

links:

Color Range

Uses the flag -color_range e.g. -color_range 1 or -color_range tv

Color Space

This defines the YUV colorspace type, as defined by ISO/IEC 23091-2_2019 subclause 8.3

...

Color Primaries

Chromaticity coordinates of the source primaries. These values match the ones defined by ISO/IEC 23091-2_2019 subclause 8.1 and ITU-T H.273.

...

Color Transfer Characteristic aka color_trc

These values match the ones defined by ISO/IEC 23091-2_2019 subclause 8.2.

...