logo Sign In

30i to 24p to 25p to 24p?

Author
Time
I have a TV capture which is sourced from an NTSC broadcast of a film, so it's 30i. When I perform an IVTC, it's restored to 24p (or 23.97), but it plays too fast. I need to slow it down around 74% to get to the right speed, which must mean it was originally a 24p film speeded up for PAL, then converted to NTSC for broadcast in North America.

I'm a little confused with "AssumeFPS" and "ConvertFPS" in Avisynth, not sure which applies in this odd situation. Any help would be appreciated. Thanks!
Author
Time
is it the Waterworld from JT?

if it is I will take a look at it sometime today or tomorrow.
what framerate are you wanting in the end?

DJ
Author
Time
hmm, I thought changing FPS was impossible without dropping frames
Author
Time
dark_jedi said:

is it the Waterworld from JT?

if it is I will take a look at it sometime today or tomorrow.
what framerate are you wanting in the end?

DJ


Yeah, it's the Waterworld Theatrical from Jaiman Tuckuh (Not the Extended). I want it to be 24p in the end. The thing is, once I IVTC it from 30i to 24p, it's 24p AS FAR AS AVISYNTH IS CONCERNED. So I can't convert it to 24p becuase I don't know how to tell AviSynth that it's actually 25p PAL.
Author
Time
crocky said:

hmm, I thought changing FPS was impossible without dropping frames


When 24p films are converted to 25p for PAL home video, they fit the 24 frames (the first second) from the film into the first 24 frames of the video, then they take the first frame from the second second on the film and add it as the last frame (frame 25) on the first second of the video. And so on. Thus, PAL runtimes are shorter because they play all the frames, faster.
Author
Time
When I load the .ts into Womble, it says 29 fps (of course).

The pattern is 4 frames of movement. Then one repeated full frame. In some places, though, there is interlace, as if there were an imperfect IVTC. (Doesn't look like a decimated telecine, because the interlacing is pretty rare).

One day I found... 10 years had got behind me. Next day was worse.

 

Download  shows from Cable DVR (Updated! Yes, it needs a rewrite, but it's worth slogging through, anyway).

Author
Time
 (Edited)
mcfly89 said:

dark_jedi said:

is it the Waterworld from JT?

if it is I will take a look at it sometime today or tomorrow.
what framerate are you wanting in the end?

DJ


Yeah, it's the Waterworld Theatrical from Jaiman Tuckuh (Not the Extended). I want it to be 24p in the end. The thing is, once I IVTC it from 30i to 24p, it's 24p AS FAR AS AVISYNTH IS CONCERNED. So I can't convert it to 24p becuase I don't know how to tell AviSynth that it's actually 25p PAL.



in this situation you could use assumeFPS(25).convertFPS(24), which is basically telling avisynth that it is 25 FPS and to convert it to 24 FPS.

However the situation you describe 24p -> 25p -> 30i seems very unlikely and this certainly wouldn't account for a 74% change in speed.

if what Jaiman said is true it sounds like some odd pulldown was applied to go from 24FPS to 30 FPS, but that still wouldn't explain why your getting such a difference in playback speed. maybe you output video is still playing at 30FPS after the IVTC try just adding assumeFPS(24) at the end of your script.
Author
Time
Jaiman Tuckuh said:

When I load the .ts into Womble, it says 29 fps (of course).

The pattern is 4 frames of movement. Then one repeated full frame. In some places, though, there is interlace, as if there were an imperfect IVTC. (Doesn't look like a decimated telecine, because the interlacing is pretty rare).

Sounds like someone has already attempted an IVTC and mucked it up. (I can't believe it was broadcast with every 4th frame repeated, the jerkiness would be unbearable).

If the pattern is constant, the repeated frames can be removed by SelectEvery(5, a, b, c, d), where a, b, c and d are offsets to the 4 frames you want to keep. Decimate() should work if the pattern changes.

You might want to try vinverse() to remove any stray combing that remains.

Guidelines for post content and general behaviour: read announcement here

Max. allowable image sizes in signatures: reminder here

Author
Time
 (Edited)
I spoke too soon - too small of a sampling.


It switches, from repeating every 4th frame, to normal pulldown (3 frames, 2 interlaced). And it often does so over fairly short sequences.

There's always enough movement, that pulldown should always be noticable.

I haven't worked out any sort of pattern, to the switchover, but that's partly because it gives me a headache. :) Weird...


What if the IVTC was set to decode the pulldown sections to extra frames? If that made the repeated frames consistant...


Edit: Wait a minute... I loaded an mpg into VideoDubMod, (which ignores the flags, right?) And now I'm seeing an irregular pattern of pulldown, but I don't notice repeated frames.

GSpot says the mpv is 23.976 pics/sec, and 29.970 frames per second.

So, if the duplicated frames are rid of, then IVTC probably would make it run faster?

One day I found... 10 years had got behind me. Next day was worse.

 

Download  shows from Cable DVR (Updated! Yes, it needs a rewrite, but it's worth slogging through, anyway).

Author
Time
Ok. Whew.

Here's what I did:

I used Womble to fix the GOP errors, which left me with a copy which was already IVTC'd(!) So I simply told Avisynth to assume that the 30i clip was 24p, because it was... up till a point.

From the opening logo to the opening title, this worked, but midway through the first shot, interlacing showed up. So I made another copy of the film, this time IVTC'd to 24p to kill the interlacing, and found the exact spot where the "switch" occured. Then I cut the two clips together.

Fortunately, I won't have to do a lot of this because I only need a half-dozen shots, but it's going to come down to eyeballing it to see which IVTC (ShowFiveVersions) works for each shot, or whether the shot is inexplicably already IVTC'd, or a combination of both!

Argh!

Sorry I thought it was a PAL thing, my bad.