Preservation of interlace video from LD capture

[cjdavis83]

Hi there

I have successfully had a LD captured using the Domesday86 project - with the raw RF tapped from the player and then I have had the output software decoded (TBC applied and 2D comb filtering all done via software) using the LD decode process. Audio decoded and ripped as raw PCM from the LD audio tracks.

The resulting output was then transcoded to a AVI container with H264, BUT encoded with MBAFF interlacing (as opposed to preserving the fields) as a MPEG-2 or TS would have done, without the MBAFF interlacing surely???. The frame rate is showing as VARIABLE FRAME RATE mode with 29.97fps however, which i found strange. Secondly, the field order was put as Top field first - i am pretty sure that SD NTSC video should be bottom field first as playback is quite jerky on my android for a start. I am able to play this on my Dell Inspiron Notebook with VLC (most recent player) but on my S9 Samsung, i have to download Video Player for Android to play the files, where playback is quite jerky. The chroma noise is a result of the 2D software comb filter, as 3D comb filter software is still in dev phase.

LD-Decode : https://github.com/happycube/ld-decode/wiki

Ideally, i would be after a 10 bit 444 archival quality video file encoded - where the field rate and interlacing are purely preserved without further encoding to maintain the highest possible picture quality straight from disc. Any thoughts/ideas on the best file container from FFMpeg ?????

Here are the links from the decode/encode:

https://drive.google.com/file/d/1-Trdocl5JlonlpzjfYBTbWkQHxalNB8p/view?usp=sharing

https://drive.google.com/file/d/1pPSFnc_YnjaBByi6oCXkEWYrQ4WBmIjZ/view?usp=sharing

Thanks guys
C Davis

[LigH.de]

Whether using MBAFF or PAFF doesn't matter, the difference is just that MBAFF can be more efficient when there is no full-frame combing all the time. The temporal difference between both fields is preserved either way, just that MBAFF encodes some areas of the weaved frame in progressive mode where that results in better compression efficiency when there is no strong combing (due to very little motion). That's why x264 prefers MBAFF over PAFF: It takes the efforts to optimize efficiency; PAFF is mostly used by hardware encoders with realtime encoding demands.

Seeing variable frame rate mode is nothing to be concerned about; often this flag means only that the output format is able to support it when it is required, but not that the frame rate actually varies a lot. But you should avoid storing H.264 video in an AVI container, because the AVI container does not support some of the features H.264 video uses (e.g. B frames, multiple references). You should prefer the MP4 (or MKV) container.

Whether the field dominance is Top or Bottom Field First, should be analyzed using a Bob filter, comparing both directions. BFF is almost certain for DV tape recordings, DVD Video and analogue TV usually (not always, though) prefer TFF. But I don't know about the preferences of Laser Discs...

The "jerky" playback on a mobile may be caused by different reasons. One may be that the video player does not play interlaced video as separate fields, but just decodes the woven frame and then scales the height regardless of the interlacing, which causes averaged combing; or it may be that the video was encoded for a higher Profile@Level than the mobile decoder is able to decode in time. Or because it does not support H.264 in AVI well.

[jagabo]

h.264 video in AVI is not recommended. You should use a more appropriate container like MP4. If you insist on AVI don't use b-frames. Your encodings have preserved the interlacing properly -- but AVI doesn't really have an interlaced flag. Many players/editors will not recognize the video is interlaced. Your frame aspect ratio is odd, 760x488 and there is no sample aspect ratio flagged. 4:4:4 chroma is overkill for a source that's not even 4:2:2. Most players outside a PC will not be able to deal with h.264 with 4:4:4 chroma.