Encoding Video for Blu-Ray using H264/AVC

[shon3i]

Originally Posted by A.Fenderson

I see --mvrange 511 set on a lot of blu-ray profiles: is this strictly necessary, or does it just seem to improve compatibility on some players?

That is because earler version of x264 are not limit mvrange to 511.75, which is now default value, and it's fine for Blu-Ray.

[MasterNobody]

Originally Posted by shon3i

That is because earler version of x264 are not limit mvrange to 511.75, which is now default value, and it's fine for Blu-Ray.

This was changed almost 3 years ago in revision 697 which anyway wasn't Blu-Ray compatible so I don't see the reason why it still appears in Blu-Ray profiles.

[Ulf]

Encoding 720p25 or 720p29.97 with "--pulldown double" and 480p23.976 with "--pulldown 32 --fake-interlaced --pic-struct" will yield Bly-ray compatible streams (assuming the rest of the settings are OK). The question is what the maximum setting for "--keyint" would be in these cases for a 1s GOP? Is "--keyint" referring to the original frame rate or the pulldown framerate, i.e. to 25 or to 50 in the case of 720p25?

[Sharc]

Originally Posted by Ulf

Encoding 720p25 or 720p29.97 with "--pulldown double" .....

Have you been successful with 720p25 --pulldown double? I get jerky playback on my BD standalone unless I encode without b frames.

[Ulf]

Originally Posted by Sharc

I get jerky playback on my BD standalone unless I encode without b frames.

Have you seen a significant difference in playback of a 720p25 clip encoded with "--pulldown double" and the same clip with frame duplication done before encoding, i.e. AssumeFrameBased().ComplementParity().SelectEvery(1,0,0) ?

[kieranrk]

Originally Posted by Ulf

Encoding 720p25 or 720p29.97 with "--pulldown double" and 480p23.976 with "--pulldown 32 --fake-interlaced --pic-struct" will yield Bly-ray compatible streams (assuming the rest of the settings are OK). The question is what the maximum setting for "--keyint" would be in these cases for a 1s GOP? Is "--keyint" referring to the original frame rate or the pulldown framerate, i.e. to 25 or to 50 in the case of 720p25?

I'm not actually 100% sure. I always thought a GOP had to have actual frames in it so --keyint 50 would be correct for 720p25.

[Sharc]

Originally Posted by Ulf

Have you seen a significant difference in playback of a 720p25 clip encoded with "--pulldown double" and the same clip with frame duplication done before encoding, i.e. AssumeFrameBased().ComplementParity().SelectEvery(1,0,0) ?

I did frame duplication like:
AssumeFieldBased().AssumeTFF()
FieldDeinterlace().ChangeFPS(50)
which was ok.
I believe that the problem with b-frames and --pulldown double in my case is that the re-ordering of the b-frames for playback is not done. Maybe a problem of the decoder.

Added:
I tried your script. It works as well. No jerkiness.
Unfortunately the file size for same crf becomes 50% larger with the frame-doubling.

I still have problems with '--pulldown double' and b-frames. It works with MPC-HC but fails (i.e. jerky playback) with VLC and with my standalone - unless I encode without b-frames.

[Midzuki]

Originally Posted by Sharc

I still have problems with '--pulldown double' and b-frames. It works with MPC-HC but fails (i.e. jerky playback) with VLC and with my standalone - unless I encode without b-frames.

Does it make a difference if you set --bframes to 2 or to 1
Or if you use --b-adapt 0

[Sharc]

Originally Posted by Midzuki

Does it make a difference if you set --bframes to 2 or to 1
Or if you use --b-adapt 0

Yes, it makes a difference.
--bframes 2 removes the backward-and-forward jerkiness. The playback is however still choppy in VLC. I didn't test on the standalone yet.
--bframes 1 is about the same
--b-adapt 0 does not make a difference.

[Midzuki]

Thanks for the clarification.
Now, what about:

for n b-frames, --keyint = integer multiple of n+1 ;

N.B., 25 is not an integer multiple of 4, 3, or 2.

--b-adapt 0 does not make a difference.

OK, but it won't hurt (much) if you use it, I guess.

[Sharc]

Originally Posted by Midzuki

Thanks for the clarification.
Now, what about:

for n b-frames, --keyint = integer multiple of n+1 ;

N.B., 25 is not an integer multiple of 4, 3, or 2.



OK, but it won't hurt (much) if you use it, I guess.

I normally use --bframes 3 and --keyint 24, which actually does comply with your proposal: 3+1=4 6x4=24. Similar for --bframes 2: 2+1=3 8x3=24. So --keyint 24 should be ok.

Removing --b-pyramid strict seems to help. No backward/forward jerkyness, but still choppy (means progressive, but not very fluent).

[Ulf]

Originally Posted by kieranrk

I'm not actually 100% sure. I always thought a GOP had to have actual frames in it so --keyint 50 would be correct for 720p25.

If --keyint refers to actual encoded frames, wouldn't --keyint 25 be correct for 720p25 even when applying soft telecine (for a 1s GOP)? Is any x264 programmer around to enlighten us?

[nm]

Originally Posted by Ulf

Is any x264 programmers around to enlighten us?

IIRC, kieranrk is responsible for x264's pulldown support.

[mp3dom]

I have a question for people that have the BD specs on their hands:
I need to encode to BD with (relatively) low max bitrate (about 21 Mbps, but surely not over 24 Mbps) because I've a lot of lossless/PCM audio tracks to use. To gain a little, my idea is to encode as L4.0 to have the possibility to use 1 slice only. The specs says that L4.0 allow a maxrate of 24 Mbps which is perfectly fine for my job. The question is: the BD allows me with a L4.0 video to have a total BD bitrate of 48 Mbps (21 Mbps video + 27 Mbps audio) or with L4.0 my max BD bitrate is more limited?

[shon3i]

Video level bitrate restriction has nothing with overall bitrate which always stay to 48mbps if data rate is least 54mbps. Data rate dictates maximum allowed bitrate for disc. In normal case you always have 48mbps, and lowering video or audio max rate will not change that, only will make more space for other streams.

[mp3dom]

Thank you very much!

[Sharc]

Originally Posted by nm

IIRC, kieranrk is responsible for x264's pulldown support.

His guides are here
(I still have no luck with --pulldown double and b-frames, but this is perhaps just me ....)

[shon3i]

Originally Posted by Sharc

His guides are here
(I still have no luck with --pulldown double and b-frames, but this is perhaps just me ....)

What you use as BD muxer?

EDIT: btw max keyint must be on real fps not pulldowned, so for 720p25 is 25 not 50, 50 is valid for 2 sec GOP

[Sharc]

ok, I may have found my problem:
I have to delete '--b-adapt x' from the command line.
I tried
'--b-adapt 0' => jerky
'--b-adapt 1' => jerky
'--b-adapt 2' => jerky
removing '--b-adapt ... ' helped. Playback ok with MPC-HC and with VLC.
Now I still have to test with my standalone. I will report back.

And thanks shon3i for the clarification about the --keyint with pulldown.

[shon3i]

@Sharc it's Tsmuxer problem. Obviously has a problem with pulldown streams.

I do a simple test, i encode video with pulldown, i muxed it with scenarist and tsmuxer. tsmuxer's stream show jerky playback, while scenarist play smooth.