30p to 60p to 24p conversion example

K

Kholi

Section Moderator
1. Converted 1080/30p H264 in MPEG StreamClip to a 720/59.94p file. Frame Stepping with the 60p and 30p file you can see that the 60p seems to add a repeated frame every even frame. So everyone 2, 4, 6 etc is a frame repeat of 1, 2, 3 etc.


2. Took the 59.94 conversion to After Effects, created a 23.976 720p Composition. Turned on Frame Blending > Pixel Motion then exported Quicktime Animation (I think it was). For a 12 second 720p 60p clip it took 1 hour and 2 minutes. I'm on my laptop with the following specs:

Q9000 Quad Core CPU @ 2.2ghz
4Gb DDR2 800 memory (Going to double this in the coming days)

My After Effects hasn't been optimized, so this time could have probably been better.


3. File result was 700mb, then took it back to MPEGStreamClip to compress a smaller H264 for upload.

I dunno how this worked or if it was worth the extra 60p step because I haven't converted the 30p to 24p to compare, but here are the results.

(right Click and Download)
www.kholi.net/35mmFloatMKii30.mov
www.kholi.net/35mmFloatMKii60.mov
www.kholi.net/35mmFloatMKii24.mov

EDIT: Added this file

30p to 24p straight
www.kholi.net/35mmFloatMKii30to24.mov


I can see more artifacting if you step frame by frame but that's only frame by frame so is it worth it? Probably not! Would like to try a good 60P convert like the MCR_FRC and see what happens.

EDIT: hyalinejim over at the 5d forums did this test

http://rapidshare.com/files/262998227/35mmFloatMKii24_TWIXTOR.mov -- Twixtor 30p to 24p. Took Five minutes and you can download the results there. There is definitely much more artifacting going on, but is it a deal breaker @ a five minute conversion?

I think I can live with the Pixel Motion 24p. I chose this video because it had a LOT of motion in it. Not steadicam, but it was plenty of motion to see if the Pixel Motion method held up enough. Takes forever to freakin' render but I think I can figure out a better method to render, especially when I try it with the 8 Core 24GB memory machines.

Comparing the 30 to 24, there doesn't seem to be any skipped frames and I can't see any odd artifacts, especially no ghosting, but I wanted to get some outside opinions.

Anyway, thoughts? I am going to try 30 to 24 when I get a chance then compare.
 
Last edited:
Not a steadicam or anything by the way, which should be obvious. Just walking with the camera.
 
There are some artifacts when the girl in white passes close to the camera... right when she leaves frame..

The ultimate test is if you can do a shot of someone punching or kicking, or just waving their arms around very quickly... always a 24p interpolation deal-breaker for me.... (I'm using twixtor in AE mostly..)
 
Matthew Bennett said:
There are some artifacts when the girl in white passes close to the camera... right when she leaves frame..

The ultimate test is if you can do a shot of someone punching or kicking, or just waving their arms around very quickly... always a 24p interpolation deal-breaker for me.... (I'm using twixtor in AE mostly..)
Click to expand...

I just saw them. Also, when the guy's passing frame, every ten or twelve or so you see a strange outline.

Do you see them without frame stepping?

EDIT HERE: Lee Wilson says this idea will not work, so don't bother with it. Proceed.

It's extra work but I suppose that if problem frames bug you then that might be the only way.

For me, personally, I hate that there are any artifacts but if I can only see them when I step through frame-by-frame then I need to just back of and let well enough alone, just gotta hope what's being watched is interesting enough to others to not have to frame step. lol
 
Last edited:
Matthew Bennett said:
There are some artifacts when the girl in white passes close to the camera... right when she leaves frame..

The ultimate test is if you can do a shot of someone punching or kicking, or just waving their arms around very quickly... always a 24p interpolation deal-breaker for me.... (I'm using twixtor in AE mostly..)
Click to expand...
Are you going well to 1080/24p with Twixtor?
 
Kholi said:
Anyway, thoughts?
Click to expand...

I did all these tests with 5DII footage late last year - all of it in After Effects.

I am not sure why you initially doubled each frame to produce a 59.94p file - what was the thinking behind this ?



I went from . . . .


30p > 60p > 25p

&

30p > 25p

All conversions done with Pixel Motion switched on - and without a doubt there was no qualitative difference between a straightforward 30p > 25p and sticking a 60p stage in-between.

I tried lots of varied footage and cold not see any advantages in having the 60p stage.

http://www.vimeo.com/2069928
 
I dunno if it's the best free means, but it's pretty effective I'd suppose. At the expense of time. Someone on the C5D boards just uploaded a 30p to 24p via Twixtor for me, of the same file.

I'll upload it to my site and share, but he said the 30p to 24p in Twixtor took like 5 minutes!

I also did a straight 30p to 24p just now and am compressing to upload it.
 
Kholi said:
I wonder if the best course of action would be to render out an image sequence of choice, compile, step and locate the problem frames then re-render out those specific frames, plug them back into the sequence.
Click to expand...


How would this help ?

If there is artefacting on a frame - how would re-rendering it help ?
 
Lee Wilson said:
I did all these tests with 5DII footage late last year - all of it in After Effects.

I am not sure why you initially doubled each frame to produce a 59.94p file - what was the thinking behind this ?



I went from . . . .


30p > 60p > 25p

&

30p > 25p

All conversions done with Pixel Motion switched on - and without a doubt there was no qualitative difference between a straightforward 30p > 25p and sticking a 60p stage in-between.

I tried lots of varied footage and cold not see any advantages in having the 60p stage.

http://www.vimeo.com/2069928
Click to expand...

I actually didn't know that MPEGStreamclip would just double frames. I was hoping for a result like MSU_FRC which created new frames in between the others (see here: http://www.youtube.com/watch?v=ZVtWIe62HA4 -- Avisynth + MSU_FRC plug-in) but that wasn't the case.

If you can get something like the MSU_FRC to work first, then I would wager that the 30 to 60 to 24 will work better. Kinda like how 60P to 24p timeline drops are pretty nice with GH-1 footage, or the 60p to 24p conversions workout to look a bit smoother.

Unfortunately, I dunno how to get something as clean as MSU_FRC.
 
Lee Wilson said:
How would this help ?

If there is artefacting on a frame - how would re-rendering it help ?
Click to expand...

Re-render that specific set of frames then plug it into what you already have rendered that's fine. It's a theory, though, and I clearly stated maybe that would work. Maybe doesn't mean it DOES, just that it MIGHT. I haven't had to deal with trying to get to 24P so this frame rate conversion is foreign to me.
 
Kholi said:
Re-render that specific set of frames then plug it into what you already have rendered that's fine.
Click to expand...

I am still not clear what it is you are saying !?



Pixel motion produces artefacts on frame 17 of a 100 frame sequence.

You re-render frame 17 and it somehow works better the second time you apply pixel motion !??


Sorry if I am being stupid here, but it's not entirely clear to my tired brain.
 
I dunno if it would work better. Again, I was just guessing at it. I haven't tried it yet and I don't know what would happen.

This is the first time I've had a camera that doesn't shoot 24p natively. I'm trying to figure out what'll work and what won't on a PC.
 
Kholi said:
I dunno if it would work better. Again, I was just guessing at it. I haven't tried it yet and I don't know what would happen.
Click to expand...


Like I said - I don't really know what 'it' is. I am not sure why you think simply re-rendering it would change anything ? (if I understand you correctly)

It sounds like you are suggesting that you render out one converted sequence (lets say 30p to 24p).

The you find a couple of bad frames here and there - so you re-render just those frames and drop them into the previous render.

What I am saying is that Pixel Motion will not render the frames out any differently the second time - in fact you could keep re-rendering the problem frames and will find that re-render 453 is identical to the first render (and identical to every other render).
 
What happens if one just slows down the 30p footage by 80% and render that out as 24p.

Then reload that exported footage back in After Effects, speed that up by 125%, and keep it as 24p?
 
BandanaDan said:
What happens if one just slows down the 30p footage by 80% and render that out as 24p.

Then reload that exported footage back in After Effects, speed that up by 125%, and keep it as 24p?
Click to expand...

If you speed 24p up by 125% it will be 30p footage.

How can you speed something up and keep it at the same speed - at the same time ?
 
You must log in or register to reply here.
Back
Top