DV Tonemapping

[CaseyP 17]

I know in previous threads related to HDR tonemapping, like this post for instance, it had been said that DV tonemapping would be a challenge given the nature of the tech, either the proprietary aspect of it, or the complicated nature of the tech.

 

Our open-sourced friendly rivals at JF have released a new server update that has added, "Dolby Vision Profile 5 and 7 tone-mapping"

After testing a DV file on a non-DV capable tv, the tonemapping worked just as it should.

 

Given that DV tonemapping is now feasibly possible, could this be a pursuit of the Emby dev team please? 

 

(Here is the video section of mediainfo of the DV file that JF was able to tonemap and deliver to a non-dv capable device)

Video
ID                                       : 1
Format                                   : HEVC
Format/Info                              : High Efficiency Video Coding
Format profile                           : Main 10@L5@High
HDR format                               : Dolby Vision, Version 1.0, dvhe.05.06, BL+RPU
Codec ID                                 : dvhe
Codec ID/Info                            : High Efficiency Video Coding with Dolby Vision
Duration                                 : 1 h 36 min
Bit rate                                 : 16.4 Mb/s
Width                                    : 3 840 pixels
Height                                   : 2 160 pixels
Display aspect ratio                     : 16:9
Frame rate mode                          : Constant
Frame rate                               : 24.000 FPS
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 10 bits
Bits/(Pixel*Frame)                       : 0.082
Stream size                              : 11.1 GiB (95%)
Default                                  : Yes
Alternate group                          : 1
Color range                              : Full
Codec configuration box                  : hvcC+dvcC

 

 

[rbjtech 5284]

Interesting.  Emby has had proper performant tone mapping for a while now (at least 12 months?) - but only on HDR Items.

Thus DV7 has always been covered - as that contains an HDR10 layer anyway - but DV5/DV8 have always been missing as there has not been any publicly available tone mapping algorithms for DV as it's not open source.

If one has become available (and is 'legal' to use), then Emby could probably easily add into the 3 currently available for HDR.

A prerequisite of this however, is for ffmpeg to identify the DV stream type - otherwise it is not going to know which tonemapping algorithm to use.

I'm sure @softworkz is aware of these developments.

[softworkz 5073]

I am closely following this development. I know who as figured out the algorithms for this (in December 2021) and the first available implementation was for Vulkan. Though, I didn't want to go into that direction because it would have added once another complication to our transcoding pipelines (i.e. something that may work or fail..).

Now, it seems JF have ported the Vulkan implementation to CUDA and OpenCL. It's a nice achievement, but it has limitations, which means that it works in certain scenarios only, requiring certain decoders and there's no software implementation. 
For our side, it's important to provide a consistent experience rather than "islandic solutions", which means that differences between sw and various hw transcodings should be as minimal as possible. 

What's pending on the roadmap with regards to tone mapping is being able to apply dynamic (scene-specific) HDR parameters (like HDR10+) and probably we'll look into handling those profiles as well, once we get to it.