3D Frame Packed output

@tanio99 do you now have all of the sample videos and log entries you need for all so-far-encountered issues? Let us know if there’s anything you haven’t got.

Thanks Nicholas.

We should have everything needed now.
A sysfs poke might allow us to use existing infrastructure to achieve the eye fix

Sam

So happy with this update! Thank you for your support in this :-).

I have looked a small piece of Gravity and observed some small issues with macro blocking also, which someone else reported also. Also some issues with stretched subtitles, but also reported by someone else. This is a Kodi issue?

So everything looking nice :slight_smile:

Unfortunately I haven’t been able to test more, because I am not at home for a few days. Will certainly test some more in the coming week.

It looks that way, sadly. You see the same issue on the Raspberry Pi, under both OSMC and LibreELEC.

Can we determine the difference between the subs on Avatar that are stretched agsinst those that aren’t?

Avatar’s subtitles are picture based. That’s probably the type of subtitle affected.

It’s not every subtitle that’s affected, though, there’s some other factor as well - it happens consistently to the same subtitle each time, but only to a minority.

It does seem to affect PGS tracks but not SRT tracks.

Right eye first is now fixed and will come in the next build.
But there are a couple of other things we want to fix first before doing another release.

Cheers

Sam

2 Likes

Great news thanks Sam

A minor thing I hadn’t noticed before: while playing a bit of Hugo, if I press the Back button during playback (as the first step of getting into Settings) there’s some slight glitching in the first couple of seconds - some skipped frames, I would guess.

Logs: https://paste.osmc.tv/tameleyuce

I think I know what’s going on here … or at least I’ve an idea :slight_smile: . I’ll try to fix it in the next build, too.

1 Like

At the risk of sounding like like an annoying child (“ARE WE NEARLY THERE YET?!”) do you have a rough idea of when that be? Days, weeks, or months…?

The most succinct answer possible is, when it’s ready.

1 Like

@ActionA, I know you think your job description is to hang around this forum abusing and insulting customers, but could you possibly find another thread to annoy people in and leave this one alone?

I’ve provided you the best answer that will be given. Even you yourself suspected you would be seen as an impatient child, your words, not mine. Please exhibit a modicum of patience as a significant and time consuming bit of development is undertaken by our team, who works without financial compensation in their free time. The issue is clearly being worked on.

2 Likes

I don’t see how @ActionA has done anything wrong here. I’ve just had a chance to check this thread, and my answer would have been the same ‘when it’s ready’. We don’t like to give firm ETAs anymore, as when we give estimates they are often held against us.

We made progress yesterday, as well as today. I’ve been speaking to @tanio99. There was a bit of a shortcut in how MVC decoding was initially implemented, and I was surprised it worked. But it seems that this shortcut is inhibiting the correct functionality with regards to ISO support.

Macroblocking is caused by the bitstream converter the eliminating mvcC atom. It is then sent to the decoder after a few seconds and the issue disappears. So this will also be solved soon.

Then we found out that SSIF demuxing is a bit more complex than expected.

PGS subtitle positioning is broken in Kodi on the whole, not just for 3D. That may take longer to fix.

We now know what needs to be done.

Things take time and I will keep updating the thread as progress is made. The only important takeaway should be that we are working on it.

giphy

5 Likes

I’m currently looking at that issue. Can you give me a timecode when that macro-blocking occurs?

If you have ‘A Christmas Carol’ it does macroblocking almost right from the start.

In Gravity at about 00:04:56 there’s a consistent pulse of macro-blocking. It’s roughly where Houston says “Engineering is recommending a vis-check for component damage” - maybe a second or two after that line.

EDIT: And there’s a second burst at around 00:07:04.

Thanks! Tomorrow I’ll get a copy of Gravity. I hope I can reproduce it.