Virtualdub 23.976 fps to 29.97 fps

[luciofulci]

Hi All,

I've got a capture file (AVC) of an NTSC laserdisc that has been converted back to 23.976 fps, but I want to put this onto a blu ray so I need to convert it to 29.97 fps as 23.976 fps isn't an accepted standard for SD video.

What is the best way to add the duplicate frames to do this? I've tested with virtualdub, under frame rate conversion, convert to fps - is there a better method?

Obviously the better method would be to capture it again without changing the frame rate but unfortunately that's not possible.

Many thanks

[LMotlow]

What does "converted back to" 23.976 fps mean? Do you mean it was inverse telecined?


In most case, people would add 3:2 pulldown (telecine) when encoding your video into BluRay. Most DVD and BluRay encoders have an instructions to do this.

[luciofulci]

Yes I believe it was inverse telecined, exactly what method or program was used to do it I am not sure of though. My usual encoder of choice is ripbot, but I cannot find any option to add 3:2 pulldown.

The virtualdub option seems to work ok, the running time is the same before and after and the audio is in sync, but I wonder if there is a better way of doing it?

[LMotlow]

The best and cleanest way is to add pulldown playback flags (known as "soft telecine) rather than adding or interpolating extra frames. Anyway, pulldown doesn't add frames it simulates periodic duplicate fields during playback, another way of saying it is that it plays the video as if it's partially interlaced. Actually creating duplicate, phsyical frames is known as a version of "hard pulldown" and is entirely progressive, not allowed for SD BluRay.


If Ripbot can't add soft-coded pulldown flags like most encoders why do you prefer it and why do you think it's so great?
If the VirtualDub method works, what do you mean by "better"?

[luciofulci]

I usually work with HD projects and prefer ripbot because it has a target file size setting which I find useful. I'm not saying that it's so great, it's just what I am used to using. However if you feel another program is better suited feel free to suggest one and I will take a look at that instead.

In terms of virtualdub I don't know if the method I described is actually doing the right thing to do, that was the point of my post. From looking at the resulting virtualdub file it appears that three frames are unique, then the 4th and 5th are identical, then it repeats. This is not my area of expertise at all, I don't know if this is soft or hard, and from what you hard said hard is not allowed.

Another program I have tried is Premiere Pro, placing my 23.976 video into a 29.97 timeline, that resulting files has three unique frames, then the 4th and 5th frame are different and both have horizontal interlaced lines.

So from what you have said and what limited knowledge I have already, I think the virtualdub method is hard telecine i.e. duplicating frames, whereas the premiere method is soft telecine, in which case I am best going with premiere, or figuring out how to soft telecine in virtualdub.

If it makes any difference the source of this was an NTSC laserdisc

Edit - the other idea I am toying with is up-scaling my existing file to 720p, then I don't have to worry about pulldown or the frame rate etc.

[poisondeathray]

In terms of virtualdub I don't know if the method I described is actually doing the right thing to do, that was the point of my post. From looking at the resulting virtualdub file it appears that three frames are unique, then the 4th and 5th are identical, then it repeats. This is not my area of expertise at all, I don't know if this is soft or hard, and from what you hard said hard is not allowed.

This is incorrect pulldown with duplicate frames. It's progressive 23.976 with duplicates to make up 29.97p . SD BD requires 59.94 fields / second interlaced signal

Another program I have tried is Premiere Pro, placing my 23.976 video into a 29.97 timeline, that resulting files has three unique frames, then the 4th and 5th frame are different and both have horizontal interlaced lines.

On a 29.97i timeline, that indicates hard pulldown 3:2. Actual fields are repeated (not frames)

It's not "soft" because actual fields are repeated

I usually work with HD projects and prefer ripbot because it has a target file size setting which I find useful. I'm not saying that it's so great, it's just what I am used to using. However if you feel another program is better suited feel free to suggest one and I will take a look at that instead.

Not sure if ripbot has user editable command line box, but to add soft pulldown using x264 , use --pulldown 32 . Not all BD authoring tools are compatible with this for AVC . Or use MPEG2 instead

Another option is to upscale to 720p23.976 . Native progressive

[jagabo]

I've got a capture file (AVC) of an NTSC laserdisc that has been converted back to 23.976 fps, but I want to put this onto a blu ray so I need to convert it to 29.97 fps as 23.976 fps isn't an accepted standard for SD video.

What is the best way to add the duplicate frames to do this? I've tested with virtualdub, under frame rate conversion, convert to fps - is there a better method?

That will leave you with a duplicate frame every 5 frames. If you watch a panning shot the 6 little jerks every second will be obvious. The best method is to encode 24.976 progressive MPEG 2 and apply 3:2 pulldown flags.

[luciofulci]

Thanks jagabo and poisondeathray, I have considered using MPEG2 but I was under the impression that x264 is a more efficient encoder and it will allow me to get higher bit rates for the same file size?

Also yes poisondeathray there is a command line box in ripbot so will try --pulldown 32. I think what I'll do is try x264 SD, MPEG2 SD and x264 720p upscaled, then compare the results.

Thanks all

[poisondeathray]

I have considered using MPEG2 but I was under the impression that x264 is a more efficient encoder and it will allow me to get higher bit rates for the same file size?

The filesize will be the same at a given bitrate, because:

filesize = bitrate x running time

x264 is more efficient than any MPEG2 encoder - it just means at a given bitrate you get higher quality

Just like "soft" will be better than "hard" because you're encoding 1.25x more data for nothing with "hard" because of the field repeats . (you can think of it as the same bitrate is "spread" over more fields or frames, so the image quality per field or frame is lower)

And progressive is more efficient than interlaced encoding in terms of motion vectors

Also yes poisondeathray there is a command line box in ripbot so will try --pulldown 32.

You need other switches to make it BD compliant as well
https://sites.google.com/site/x264bluray/home/480p-ntsc

But this soft pulldown AVC stream has problems in some BD authoring tools. Works fine in pro authoring tools

[luciofulci]

Great, thanks for all your help and I'm using Blu Disc Studio so should be fine with the soft pulldown I imagine but I'll run a small test file first.

[jagabo]

I may be wrong but I thought only MPEG 2 was officially supported for SD resolutions on BD.

[johns0]

I just checked with tmpgenc authoring works 6 and it accepted a sd avc encode as blu-ray compliant for authoring.

[Sharc]

Blu-ray 'Secondary Video' allows 480p23.976 and 480p24 mpeg4, level 3.2.
Because a certified blu-ray player (carrying the blu-ray logo) should support secondary streams it shouldn't be required to hard- or soft telecine the 23.976 or 24fps progressive source to 29.976fps, I think.

[luciofulci]

I've used a few SD AVC files with blu ray and they have always worked, but then again they might work but not be officially supported. However when I added the 480p 23.976 fps file it ceased the authoring process immediately and gave an error.

[Sharc]

If soft pulldown fails you could use this avisynth script for hard telecining 23.976 -> 29.97fps

...your source filter for the 23.976p source…..
SeparateFields()
SelectEvery(8, 0,1, 2,3,2, 5,4, 7,6,7)
weave()

OR:

...your source filter for the 23.976p source…..
SeparateFields()
ChangeFPS(59.94)
weave()

and encode as interlaced (TFF).