VS10 Engine / VSVDB Injection Addition Into Mainline CE

I was wondering if a good framework could be created to break up these changes.

As far as I see, the greatest challenge would be porting these changes over to Kernel 5.15 and Kernel 5.4. This would mainly affect:

  1. the media_module-aml code for the ATEME parsing which CPM already submitted as a PR.
  2. There are six patches for linux-amlogic which would require porting,

There there is a batch of patches for XBMC which should not require porting.

I think I see two realistic pathways to getting this merged,

  1. Require the breaking down of this PR into its parts: VS10; HDR10+ conversion; improved mode switching; etc. I think this would be extremely hard because it would require quite a bit of dev time.
  2. Port the remaining six linux-amlogic to 5.4. Then do a round of community testing to ensure that all the features work in 5.4 and merge this as one monster PR. This would require less dev time, but such a large merge carries a larger risk of bugs. However, this would make it much more feasible to merge. I’m excluding porting to 5.15 because no dovi.ko module exists for this kernel version.

I really think that option 2 is the only reasonable pathway forward. However, I would welcome input from the devs so that a realistic path for merging can be created.

Could be OK, because now, at this development stage, it’s absolutely unusable because of subtitle stutter.

yeah, this subtitle stutter is a weird issue, it happens to me on occasion even on the ng/4.9 kernel (probably not as severe as in no), sadly haven’t found a reproducible pattern so far…

-no does not have a dovi.ko module so developing these features would be a moot point.

-ne might be useful for the the S905X4

1 Like

We just don’t have the manpower to do this at this point. We were hoping that someone knowledgeable from the community could start working on organising cpm’s work and join the team to add the features and then we could try to make it ne+ng (if there’s no advance in NE front that brings FEL to it.)
So is anyone up to the task? We can only offer some fun times and the occasional vendor sample to work on!
Something like the very beginning of CE:

1 Like

on YadaYada thead where he’s got FEL working we found test files and he wanted to track down a potential author to give credit
markswift2003, anyway I just did a google search for
“VS10 engine / VSVDB injection / HDR10+ developer”

and third link was " [

It is possible using the vs10 engine with HDR+? - Zidoo forum

](It is possible using the vs10 engine with HDR+? | Zidoo forum)"

and markswift2003 seems to be answering the questions. I’ve signed up on that forum to ask if he’s the author. I’ll ask if he’d like to help here and join the CoreElec team, he seems like he’s been working on similar stuff himself and seems perfect for this! plus he’s last active 4 days ago

He just answered a basic question in that thread, not sure how that helps.

He’s also well-aware of the CoreElec project, its capabilities, and still chooses to use a Zidoo and insists on watching content with MEL and fake TV-led since it’s come up multiple times on the Zidoo forum.

That alone is a huge redflag…

3 Likes

that’s a shame.

I think CPM’s idea of hiring a developer to merge his changes is a solid one.

I don’t know what the policy is for the CE development team regarding paid development or the implications regarding tax or or other elements but here is an idea :

It’d be nice to have some sort of “project plan” outlined, including tasks, timelines and associated costs. Based on this, a total amount could be calculated for the cost of the developer. This could be shared with the community here. This would allow us to see the tasks and timelines and associated costs as a group.

We could then organize a poll to see if people are willing to support the cause and if so, with how much money. If the monetary goals are met by the offerings in the poll, we could start a crowdfunding campaign (Kickstarter, Indiegogo, Fundable, Patreon) to get the money.

Just an idea

Unfortunately this didn’t went the way we thought it would go. This is not a commercial project. The Team spends its free time trying to bring the best support for all the devices. If no one in the community can bring this past the goal line we have to put it on the back burner and proceed with life as usual. Thank you all for your insights.

1 Like

Just as finish up to show it’s not just a click/merge job: Workaround for ATEME Titan created HEVC ES - where existing logic cannot determine the SEI Mastering Display Colour Volume and Content Light Level by cpm-code · Pull Request #3 · CoreELEC/media_modules-aml · GitHub

media_modules-aml-efc5cebbdd3d9a5ca9be2d912a07c314e8bbc345/drivers/frame_provider/decoder/h265_fb/vh265_fb.c:9406:13: error: 'struct hevc_state_s' has no member named 'sei_present_flag'
 9406 |         hevc->sei_present_flag |= SEI_MASTER_DISPLAY_COLOR_MASK;
      |             ^~

It’s a PR for a 4.9 EOL system. We merged it, than did uplift to NE, 5.4. What does at least compile (untested if it works at all). Then the lift to NO, 5.15, and it does not even compile. So we need to invest minimum a few days to check, compile, test and confirm the uplift.

You might still remember: it’s a hobby, not a commercial product!

This is why we did ask the author @cpm to handle this by himself. He did decline so team CE is in touch to handle the changes. But as several times written, CE is out of power and time.

So to through just something on the table and then lean back and say: “don’t ask me!” is not the correct attitude for a open source project.

Some user did write team CE does not take @cpm changes because of ego. Such people are just wasting time on free projects what do think it’s a “one click” job. Merge just everything and break everything…

Team CE is one of the projects over others what try to take care about users and stability of the system. Sure when other not named projects are broken anyway what should go wrong to merge something broken. It only can get better…

It’s not just to get one device perfect running, it’s to make all thousands other device happy to.

5 Likes

I presume this is directed at me, had I known this up front I would not have shared to start with, I apologies to everyone, I have always been open to questions and answer them all, they were just ideas I was trying nothing more / nothing less tried my best to make that work with my limited time.

Please feel free to close all the community topics I created and everyone can continue about your normal process, it was not my intent to sow division or make people upset.

2 Likes

Speaking as a member of the community, we all appreciate your work. You’ve really done an amazing job improving CoreELEC and added many needed features as well as bug fixes.

I think some of the developers might be frustrated because they want to merge the features, but they are having a hard time seeing how to do it. As you know, it’s a lot of work. But, you’re build is beloved because it’s awesome.

4 Likes

It’s beyond question the code change are very qualified and will be a improvement for CE. But it’s not just a one time merge job. This is why we were asking if you are able to handle this at CE directly by give access to our GH repos. This was declined and this is what I mean about “no help”.

Now we are in a quandary as user do push to get your changes merged in “mainline” - now! But we aren’t able (yet) to handle it - to be tested and confirmed for all CE devices.

Maybe @TheCoolest is trying in future to handle the changes - but this is no promise. So please @cpm keep in touch with CE if there is any question about the changes or if you decide one day to help us out by handle these changes by yourself.

thx

3 Likes

Oh come-on, please don’t let some miscommunication or misunderstanding drive you off from this great/desired achievement. You know how many users benefited from your work, and would be sad if you just give up now, when it’s finally on the way to benefit all CE users.

2 Likes

With all due respect, CE team spent what, two months? to bring 3D compatibility to a handful of users, but this is now so challenging that it’s not worth your free time to bring these changes to all CE users?

1 Like

Yes, but also remember to finally worked on and implemented did take 6 years?
Again, as CE is not a commercial product it depends on time and the “want to” factor to…

About the idea to hire someone: our income is about €500/Year. About €350/year are need to cover our outcome like server, DNS and other items.

So with the left €150/year we can hire a independent software developer for one meeting/year, maybe? :rofl:

2 Likes

I did not mean it that way. As a long XBMC/Kodi user I have witnessed the development of LE and how, at the point when they decided to drop Amlogic SoC from further development, it took the team led by @Adam to walk away and started CE development. I know how the story with @Adam finished and how the team kept on going to present day of success.
I just don’t want to se another such split in ideas to cause such harsh proceedings, because I’m well aware how egos influence people behavior.

1 Like