View Full Version : PAL to Film (highest possible quality)
Ranguvar
28th March 2008, 03:13
Using solely AviSynth, I'm searching for a way to do PAL to Film conversion for ripping PAL DVDs... assuming they used the ~4% speedup technique.
My first question. Should corrected PAL be 24fps, or 24000/1001fps?
And the second, is using AssumeFPS with the True parameter for also changing the audio free of the slight artifacts produced by TimeStretch? If not, is there another way to do the audio changing?
:thanks:
2Bdecided
28th March 2008, 11:03
Your title is misleading. It sounds like you want to take some PAL footage and create a 35mm film print from it!
What I think you want to do is take 25fps material and watch it at 24fps, but maybe you can explain more clearly.
Cheers,
David.
zee944
28th March 2008, 11:14
And the second, is using AssumeFPS with the True parameter for also changing the audio free of the slight artifacts produced by TimeStretch? If not, is there another way to do the audio changing?
If the source is DVD, then the audio stream is probably in AC3 format. I doubt you can timestretch it without uncompressing. I'd extract the AC3 to WAV with tools like Azid or Besweet then timestretch it with any audio editor. I prefer GoldWave, or to name a command-line tool, sox. Usually the ~4% PAL speedup is true for the audio pitch as well, so converting back you need to resample it (affect the pitch) again, normally this kind of timestretching is free of artifacts.
Ranguvar
28th March 2008, 18:14
I am doing this to make a correct DVD rip/compress from sped up PAL DVD material.
zee944, BeSweet uses TimeStretch(). I assume Azid does too. By artifacts, I do not refer to having to re-encode the audio... look up how TimeStretch works.
Thank you :)
sixide
28th March 2008, 20:59
I believe BeSweet can use something like TimeStretch(), but you want to do a rate adjustment. It will affect pitch and tempo, but you will probably be just reversing what was done in the FILM->PAL conversion.
I really haven't had much luck with BeSweet for timestretch, I have used Audacity (which is painfully slow) to do a stretch and normalize with 32-bit float precision.
Comatose
29th March 2008, 00:00
I don't think you need to actually go and bother modifying the audio manually.
If all they did was really do the 4% speed up (24->25?) then you can just do it like this:
assumefps(24,sync_audio=true)
Though I've never actually done anything this way. sync_audio=true will change the sampling rate of the audio so it syncs.
Once you've done that, I recommend downloading SoundOut.dll (from http://avisynth.org/warpenterprises/) and saving it to WAV with that, then encoding to whatever from the WAV.
Note that what I just told you (about using assumefps) is based on the assumption that they go from 24 to 25 fps, and not 4%, since 4% from 24 is actually 24.96 fps.
Perhaps somebody else who's experienced can give you more information.
Ranguvar
29th March 2008, 00:45
Thank you, Comatose.
I did find that little function... what I want to know is whether I need to go to 24fps, or 23.976fps. And it's definitely not 4%, it's around 4%, which is why I said ~4%.
sixide, I used to use that. I need to know if there is a higher quality method rather than using TimeStretch. Thanks for the Audacity suggestion though.
So, is PAL 23.976fps (24000/1001) truly, like NTSC? Or is it 24fps (24000/1000)?
sixide
29th March 2008, 01:05
I guess I should have said that BeSweet can do a rate adjustment, using the -soundtouch( -rate -4.xxx) command as opposed to -r, -pitch, and -tempo, all which do the sort of distortion I think you don't want.
zee944
29th March 2008, 09:59
So, is PAL 23.976fps (24000/1001) truly, like NTSC? Or is it 24fps (24000/1000)?
PAL is 25fps. You mean the film? AFAIK it's 24 fps all over the world. Though it's really isn't a noticable difference between 23.976 or 24 fps.
I still don't know Timestretch, but doing it in Goldwave ("Rate" method) will be high quality conversion without any kind of artifacts I'm sure.
2Bdecided
31st March 2008, 12:06
You'd only need to do a pitch-preserved time stretch down if the original transfer was done with a pitch-preserved time stretch up.
Given that both processes can introduce artefacts, I think it's something to avoid if possible.
(I talked to the guy who did the UK/PAL LOTR encoding, and he's using slightly better software than Goldwave, and he still cringed at what it did to the strings in the score)
Cheers,
David.
Ranguvar
31st March 2008, 19:59
Alright... guess I'll be keeping it 25fps then.
:thanks:
2Bdecided
31st March 2008, 21:39
No, you can do to 24fps if the original was simply resampled, because it's easy enough to resample it back.
You can also do 23.976fps if you can get a sound track from an NTSC version to match up to your PAL version. Before we had HD available, that was the ultimate quality for obsessives - the correct pitch track from NTSC, and the higher resolution from PAL. I'm not sure anyone ever bothered with this though.
Cheers,
David.
shon3i
31st March 2008, 21:39
I am doing some NTSC (24) to PAL (25) conversions :D, and from my knowledge, AssumeFPS ("pal_film", true) doing very well job.
You should use AssumeFPS ("ntsc_film", true), for reverse conversion.
Encoding audio should be simply and quality if you use BeHappy. BeHappy can use AVS as input and you can always use TimeStretch function instead AssumeFPS, because is safer, with 25.000->23.976 preset and Aften AC3, should make transparency job
If you don't beleve in these stretch functions you can use some audio editor and do this manualy.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.