Outside of initially unlocking the TV-led and some reworking of the mel/fel detection (which itself now seems unsure if needed anyway) - which are all in the mainline, I was just exploring some nice to have features, what you describe is all there in the nightly for a DV display.
The features caused some additional momentum and with it more eyes and expectations, maybe it is now a lull until another phase can take this to the new kernels - which obviously needs resource - and most likely in the form of another outside dev with time.
The HDR10+ conversion alone is a biiiiiiiiig win to me because it means I no longer have to care which movies are HDR10+ or Dolby Vision. It saves me and I suspect others a lot of time spent manually converting our movies to be in a format which is acceptable for our current TV. The fact I can use this to effectively keep a single MKV as an archival copy of my physical movies means a lot to me and I suspect it would others.
Like you say there isn’t really a good solution to keeping a digital collection of movies and being able to enjoy them as they were intended to be. I did so much research figuring out how I can buy and watch 4K blu-rays anywhere in my home and for me having automatic HDR10+ conversion beats every other option available imo.
Yes this was a key driver why I did it (longer term was looking at loading alternative RPU dynamically as well - for similar reason)
This conversion is purely on the Kodi side, just C++ mainly based on quietvoid’s Rust code, not amlogic specific - if can find a friendly dev in Kodi project (maybe 22) then better to have it lifted into Kodi itself rather than in the CE fork.
It needs someone who has the time and wants to do it, again I can answer questions but do not have the time to make that happen myself.
On a related note I also understand @doppingkoala has a new approach to conversion that by initial accounts is providing nice results I think with time that could also be brought in (assuming that is ok with the author )
All good with me. Wouldn’t be much effort to do - just uses the distribution values.
Haven’t actually put the code for the improvement up yet though, still playing with it - but any further changes are looking to be a lot less significant.
For now, if anyone wants to give it a try, the dovi_tool.exe I provided earlier can be used (does required conversion on a PC first though). My latest build does do it on-the-fly.
Yes, little less than 100 CE team devs is not capable. Sorry.
And if I read such statement one more time I will just delete whole post. Enough was written in the past and should be clear the position of everyone involved!
Just FYI for everyone else I spent about 4 hours investigating and looking at the player process and exposing the values - had a bank holiday here - and they were not too happy, they thought I had already trimmed down my involvement with this “silly” video stuff and would not be locked away trying to get things working all day - trouble is I cannot leave an idea alone until get something working.
I thought I could just do some minor stuff still but not the case.
Yesterday I spend time to fix the issues with CE settings addon after last Python bump. And no one even noticed that feature were broken. But now at least it works again as it should.
What I’m trying to say is: I do this with not obligatory reasons. Sometimes I do more, sometimes less, sometimes nothing.
I’m not a C++ developer at all but I started looking and thought maybe even a small contribution could be meaningful for the project. Your contributions will always be welcome by the team (at least that’s my understanding).
I have learnt so much about the gcc compiler flags, CPU registers, etc. I have even skimmed about 900 page document of the S922X CPU application guide to understanding how it all works. Though I have no meaningful contribution to make to the community I am happy with the learnings along the way.
I didn’t know about ARM vs Thumb vs Thumb 2 before I read but now I do. I’ve always been fascinated by SIMD so now I’m diving deep into that. Who knows what this project teaches me next.
Would not over think it, lots of considerations from many angles just better to close it here, tomorrow is another day and something else will undoubtedly come in time.
Gents, We cannot find any words to describe how grateful we are for all your efforts in providing and supporting CoreELEC OS, we appreciate you and thank you so much.
Why do people really pressure the devs into incorporating the features into mainline as a priority. Doesn’t the cpm build do everything you guys need out of CE right now.
Some of us genuinely appreciate what the developers and cpm have done here and apologise for the people who think they are owed this.A development approach targeting multiple devices rather than one perfect device benefits more use case scenarios and more people. corelec would not be what it is today if development was focused on one device.It was only your efforts cpm was able to get his features built upon.I think it is the pressure of us users that has led to this situation rather than anything else.
@cpm I respect your decision.I think family comes first for everyone.Take care.I think you have accomplished what I assume you set out to do with these builds already(except the new hdr10+ conversion approach) as that was the last thing on your list of personal goals as you stated that you are not looking to provide updates/builds as it completes what you need.I just hope you can help out the devs if and when they decide to incorporate these features in the mainline.
CE devs should should be free to continue with efforts for the long run without being pressured to incorporate it and just close threads or delete the requests pressuring them to do so and continue working towards what CE set out to be.
Thank you again to all developers and cpm for your contributions.
Did you think about contacting Kodi developers on their Kodi forum, since there are many more of them who might want to tackle this task and include your work in future Kodi releases?
This would also make CE developers happy, to let someone else resolve this. The only thing is that then they could not boast how they were the first to do it, as they falsely did with CE being first to implement FEL support, several moths after it was available in Android Kodi on AM6B+
)
All good with me. Wouldn’t be much effort to do - just uses the distribution values.