Cine Encoder 2021 (Windows/Linux)

[Helg1980]

Program description

Cine Encoder is an application, uses the FFmpeg, MKVToolNix and MediaInfo utilities, that allows to convert media files while preserving HDR metadata. Supported hardware encoding NVENC for codecs H265, H264 and Intel QSV (for Windows and experimental for Linux) for codecs H264, MPEG-2. The following encoding modes are implemented: H265, H264, VP9, MPEG-2, XDCAM, DNxHR, ProRes. The program can also be used to change the HDR metadata, such as Master Display, maxLum, minLum, and other parameters.


Download

https://github.com/CineEncoder/cine-encoder/releases

Linux versions:

https://www.pling.com/p/1406740/


Report a bug

https://github.com/CineEncoder/cine-encoder/issues

[Selur]

The first program for Linux for encoding HDR videos appeared.

What about all programs which support x265 or NVEncC, which both support HDR encoding fine for quite a while?

I want to know your opinion about the program.

Personally I don't like the black theme.
Looks like a simple Qt based GUI which uses MediaInfo to analyse video and ffmpeg for decoding and encoding.
So another ffmpeg gui.

Problems I see are:

• afaik ffmpeg doesn't allow to set light levels, master displaying information and chroma sample location.
  You only set "-sei hdr -color_primaries {colorprim} -color_trc {transfer} -colorspace {colormatrix} -color_range tv" so 'while preserving HDR signals' also seems a big far fetched.
• you hardcode tv luma range
• only support HDR10, no HDR10+, HLG or DolbyVision.
• rather limited usage due to the restricted HDR support and no additional filtering&co

Also why Linux only? Seeing that you base your code on Qt, why not support Windows and MacOS?

-> Which you the best with your tool! Linux can always use some more coders.
Sadly it but it doesn't seem to offer anything to me which wasn't available before so for me it's a pass.

Cu Selur

Ps.: Why the mediainfo dependency? Why not use ffmpeg to analyse the source if you don't set stuff like light levels&co?

[Helg1980]

Hi!

As far as I know, there is no support for HDR signals in Linux programs yet (we are only talking about Linux programs).
I tried running Hybrid, it has an experimental mode, but it doesn't work for me.

To add brightness signals I used the mkvmerge utility as recommended by Google:
https://developers.google.com/media/vp9/hdr-encoding

HLG mode is also supported, you can see it in the Advanced section of program.
Other modes are in development.

I know about your Hybrid program for a long time, it is a very good program,
but it works stably only for Windows for me (I haven't tried it under MacOS)
and an excessive number of settings, as it seems to me.
So I was forced to use my own program.

The version of my program under Windows and MacOS I don't know when it will appear, it is still only in my plans.

[Selur]

To add brightness signals I used the mkvmerge utility as recommended by Google

Good catch, I totally forgot about that method, but this does limit the output to mkv(webm) only right?

[Helg1980]

Yes, this is the only problem when using this method.

[Selur]

Small suggestion, you could:
a. encode the video with x265 (decode with ffmpeg pipe to x265)
b. encoder audio and add passthrough video using ffmpeg
this way at least for H.265 you would have proper HDR signaling and you wouldn't be bound to mkv as container.
Yes, it would add one additional step to the processing, but this way the stream would have the proper signaling.

I'm wondering if you add the signaling through mkvmerge why not add all (colorprime, transfer, colormatrix, luma range) the signaling this way?

Cu Selur

[Helg1980]

why not add all (colorprime, transfer, colormatrix, luma range)

I also add these signals using mkvmerge


mkvmerge \
-o HDR_strobe_scientist_18Mbps.mkv\
--colour-matrix 0:9 \
--colour-range 0:1 \
--colour-transfer-characteristics 0:16 \
--colour-primaries 0:9 \
--max-content-light 0:1000 \
--max-frame-light 0:300 \
--max-luminance 0:1000 \
--min-luminance 0:0.01 \
--chromaticity-coordinates 0:0.68,0.32,0.265,0.690,0.15,0.06 \
--white-colour-coordinates 0:0.3127,0.3290 \
HDR_strobe_scientist_18Mbps.webm

[Selur]

ARGH!
I fust looked at your code:

cmd = f'mkvmerge -o "{output_file}" --colour-matrix 0:9 --colour-range 0:1 ' \
                          f'--colour-transfer-characteristics 0:{trc} --colour-primaries 0:9 --max-content-light 0:1000 ' \
                          f'--max-frame-light 0:300 --max-luminance 0:1000 --min-luminance 0:0.01 ' \
                          f'--chromaticity-coordinates 0:0.68,0.32,0.265,0.690,0.15,0.06 --white-colour-coordinates ' \
                          f'0:0.3127,0.3290 "{temp_file}" '

WTF? Do you even know what you are doing?

--colour-primaries 0:9 --max-content-light 0:1000 ' \
                          f'--max-frame-light 0:300 --max-luminance 0:1000 --min-luminance 0:0.01 ' \
                          f'--chromaticity-coordinates 0:0.68,0.32,0.265,0.690,0.15,0.06 --white-colour-coordinates ' \
                          f'0:0.3127,0.3290 "

are source dependent, those should not be hardcoded!
You should either let the user decide their values and/or take those values from the source.

For a source with for example values like:

colour_description_present               : Yes
colour_description_present_Source        : Stream
Color range                              : Limited
colour_range_Source                      : Stream
Color primaries                          : BT.2020
colour_primaries_Source                  : Stream
Transfer characteristics                 : PQ
transfer_characteristics_Source          : Stream
Matrix coefficients                      : BT.2020 non-constant
matrix_coefficients_Source               : Stream
Mastering display color primaries        : BT.2020
MasteringDisplay_ColorPrimaries_Source   : Stream
Mastering display luminance              : min: 0.0020 cd/m2, max: 1000 cd/m2
MasteringDisplay_Luminance_Source        : Stream
Maximum Content Light Level              : 992 cd/m2
MaxCLL_Source                            : Stream
Maximum Frame-Average Light Level        : 518 cd/m2
MaxFALL_Source                           : Stream

you would totally change the HDR signaling!
Those values on the google VP9 page were an example not a 'always use these'.
If those values were always the same signaling them would make no sense.

Cu Selur

[Selur]

When reading the MediaInfo data:

MediaInfo-Data -> mkvmerge parameters
'Mastering display luminance' -> gives you 'min-luminance' and 'max-luminance'
'Maximum Content Light Level' -> gives you 'max-content-light'
'Maximum Frame-Average Light Level' -> gives you 'max-frame-light'
'Color primaries' -> gives you 'colour-primaries'
'Color range' -> gives you 'colour-range'
'Mastering display color primaries' -> gives you 'chromaticity-coordinates' and 'white-colour-coordinates'

=> with those values you should be able to fix your tool

For help interpreting the 'Mastering display luminance' data, here's a short code snippet from Hybrid source code:

  if (line == QString("Display P3")) {
    line = QString("R(x=0.680, y=0.320), G(x=0.265, y=0.690), B(x=0.150, y=0.060), White point(x=0.3127, y=0.3290)"); // DCI-P3 D65
  } else if (line == QString("DCI P3")) {
    line = QString("R(x=0.680, y=0.320), G(x=0.265, y=0.690), B(x=0.150, y=0.060), White point(x=0.314, y=0.3510)"); // DCI-P3 Theater
  } else if (line == QString("BT.2020")) {
    line = QString("R(x=0.708, y=0.292), G(x=0.170, y=0.797), B(x=0.131, y=0.046), White point(x=0.3127, y=0.3290)"); // Rec. ITU-R BT.2020-2
  }
  QRegExp reg("^R\\(x=(.*), y=(.*)\\), G\\(x=(.*), y=(.*)\\), B\\(x=(.*), y=(.*)\\), White point\\(x=(.*), y=(.*)\\)$");
  if (!reg.exactMatch(line)) {
    reg = QRegExp("^R: x=(.*) y=(.*), G: x=(.*) y=(.*), B: x=(.*) y=(.*), White point: x=(.*) y=(.*)$");
    if (!reg.exactMatch(line)) {
      sendMessage(HANALYSE | HPOPUP, QString("Unknown 'Mastering display color primaries'-format, please report to Selur.Thanks!\n%1").arg(line));
      return;
    }
  }

The info about how to parse the 'Mastering display luminance' data from MediaInfo can also be found somewhere over at doom9s in the MediaInfo-thread iirc.


Cu Selur

[Helg1980]

Yes, you are absolutely right. I'll need to fix it.

[Helg1980]

Hi,

I completely rewritten the program. Now everything clearly works.

[Selur]

Good for you.
Looking at the code and since you use ffmpeg for decoding you might want to support gpu decoding.

Cu Selur

[Helg1980]

After several months of work, a new version of the Cine Encoder v3.0 was released.
In the new version:

- changed the design,
- added batch mode,
- added 2-pass mode,
- added a mode for working with presets,
- added a pause function during conversion.

There are packages for the main distributions: Debian, Ubuntu 20.04, Fedora 32, CentOS 7.8, Arch Linux, Manjaro Linux.

[Helg1980]

In new version 3.1:

- released a package for Windows,
- fixed some bugs,
- improved design.

[Helg1980]

In new version 3.2:

- added new features: metadata editing, audio track selection, subtitle selection, clip cropping, clip sorting,
- added presets,
- fixed some bugs,
- improved design.

[Helg1980]

In new version 3.3:

- the interface has been redesigned into dock widgets,
- added new features: additional settings for presets.
- added presets for Intel QSV, XDCAM,
- fixed some bugs.

[Helg1980]

In new version 3.4:

- added log window,
- added a light theme,
- added the ability to minimize the program to the tray,
- apply the preset with a double left click,
- change a preset's names in the preset window,
- added the ability to convert the color primary, color matrix, transfer,
- fixed some bugs.

[ricardouk]

Can you help?
https://forum.videohelp.com/threads/405889-Trying-to-install-cine-encoder-in-Ubuntu