How do you convert a 50fps video to 60fps with AviSynth?

[VideoFanatic]

I don't want to do any interpolation I just want the extra frames to be duplicate frames as it's just a video for YouTube.

[buzz1891]

Why bother? YT can handle 50 fps in exactly the same way as 60 ..

[themaster1]

Changefps(60) i believe

[johnmeyer]

buzz is correct: it is a pointless exercise because YouTube will display ANY framerate (e.g., I've uploaded 12 fps material).

themaster is also correct: ChangeFPS will do what you want. However, the result will look like garbage.

[VideoFanatic]

It's not pointless if I want to combine 60fps footage with 50fps footage.

How will ChangeFPS look like garbage if it's just duplicating frames?

[themaster1]

changefps will give you a jerky video but you clearly said you don't want frame interpolation so there you go case solved

[Cornucopia]

Assumefps so it changes the speed. Then similarly adjust the sound to keep in sync time-compress w pitch correction). That is the way the pros do it and is usually the best option. Exceptions would be motion-critical/speed-sensitive work. Assume fps keeps same original frames, just plays them at the assumed (different in this case) rate.

Scott

[VideoFanatic]

When you use Yadif(mode=1, order=1) on something that's already 1080p 25fps for example then it will double the framerate to 50fps but it will simply duplicate every frame so there's no glitches when watching it back. The footage looks exactly the same as the original. Is there no way to do something like that but for converting 1080i 25fps to 60fps?

[poisondeathray]

When you use Yadif(mode=1, order=1) on something that's already 1080p 25fps for example then it will double the framerate to 50fps but it will simply duplicate every frame so there's no glitches when watching it back. The footage looks exactly the same as the original. Is there no way to do something like that but for converting 1080i 25fps to 60fps?

For 25p content, there will be loss in effective resolution when using Yadif(mode=1, order=1), because yadif will discard half of the the lines per frame, then attempt to interpolate the data it just dropped

In terms of cadence, however, yadif will appear the same - 25p content with duplicates to make 50p, it's just that the quality of each frame will be lower (essentially 1/2 effective resolution)



25fps is not evenly divisible into 60fps- so you cannot do the same thing in terms of cadence unless you use motion interpolation (synthesize new inbetween frames). But motion interpolation has risk of artifacts

ChangeFPS (inserting duplicates) is the "cleanest" way to do 25fps to 60fps, but it will be jerky because the inserted frames are not at some even,constant interval. For 25p to 50p , that would be ideal


Another option is "VFR" , but it's slightly more complicated. Each section would play at it's correct framerate, in sync

[VideoFanatic]

What if on a 1080i 25fps video I do this which gives 100fps then use ChangeFPS to get 60fps, will I still get jerky video?

QTGMC(Preset="Super Fast")
QTGMC(Preset="Super Fast")
Changefps(60)

[poisondeathray]

What if on a 1080i 25fps video I do this which gives 100fps then use ChangeFPS to get 60fps, will I still get jerky video?

QTGMC(Preset="Super Fast")
QTGMC(Preset="Super Fast")
Changefps(60)

Still jerky, because 60 is not evenly divisible into 100. The dropped frames from 100 to 60 using changefps are not "evenly spaced in time"

[Selur]

Another option is "VFR" , but it's slightly more complicated. Each section would play at it's correct framerate, in sync

out of curiosity: How does YouTube handle vfr content?

[VideoFanatic]

I would have 1080i 25fps, 1080i 29.97fps and 1080p 60fps videos which I combine into 1 video in a video editor. Are video editors such as Adobe Premiere any good at 1080i 50fps to 60fps. Do they do the same thing as AssumeFPS by also changing the sound speed and does it give the same results as in Avisynth?

[poisondeathray]

Another option is "VFR" , but it's slightly more complicated. Each section would play at it's correct framerate, in sync

out of curiosity: How does YouTube handle vfr content?

ok for the most part - Think of all the mobile phone videos uploaded directly - they are almost all VFR

As long as the min FPS sections were not <6 for a long time, YT essentially copied the timecodes(timestamps). But if you had long sections with low FPS <6, it triggered a conversion to CFR (inserting duplicates like changefps). 6 was the cutoff (where people wanted to use 1 frame for say, a song + 1 static picture, and didn't care about seek points).

I would have 1080i 25fps, 1080i 29.97fps and 1080p 60fps videos which I combine into 1 video in a video editor. Are video editors such as Adobe Premiere any good at 1080i 50fps to 60fps. Do they do the same thing as AssumeFPS by also changing the sound speed and does it give the same results as in Avisynth?

Premiere is a CFR timeline. All NLE's use CFR timelines. Everything is converted to the sequence settings in PP by duplicating or dropping frames. Think of it like ChangeFPS(sequence setting) . The other setting you can use for the sequence (and most NLE's have this option too) is "blending" frames. That would be like ConvertFPS in avisynth

AssumeFPS is keeping the same frames but speeding up or slowing down. It's like interpreting the footage in Adobe, assigning a playback framerate . Speedup would not work well for 50fps to 60fps , because it's 1.2x (assuming you wanted "normal" playback , not sped up on purpose) . The magnitude is too large. For something like 23.976p to 25p, it's only ~1.0427x - That difference is less noticable

The equivalent function used in PP or avisynth would give the same results.

[Selur]

@posiondeathray: but what happens if I have a video that starts with 1fps then mainly is 30fps with a few portions being 120fps? and the frame rate of the clip indicates 17.3 fps.
What will YouTube do?

[poisondeathray]

@posiondeathray: but what happens if I have a video that starts with 1fps then mainly is 30fps with a few portions being 120fps? and the frame rate of the clip indicates 17.3 fps.
What will YouTube do?

It's far from perfect , and it tends to convert anything "not simple" to CFR .

Things might have changed too - my tests were a few years ago

[VideoFanatic]

Does CFR mean contant frame rate?

[Selur]

yes
cfr = constant frame rate
vfr = variable frame rate

[davexnet]

Here's an example of 25 > 29.97 using AssumeFPS and Avisynth Timestretch for audio. May be feasible depending on the material

[Bwaak]

I would deinterlace and drop 50p onto 60p timeline, modern NLEs are pretty good with handling different frame rates and are smart enough to employ appropriate frame repeat pattern. I also prefer turning off frame blending.

[johnmeyer]

It's not pointless if I want to combine 60fps footage with 50fps footage.

Ah, it would have helped if you had included that in your OP. I'll answer your revised question after first answering your question below.

How will ChangeFPS look like garbage if it's just duplicating frames?

As others have now said, duplicating frames, even when used for 3:2 pulldown to convert 24p to 60i, results in a "stuttering" or "juddery" (pick your adjective) playback. This will be most noticeable on fast motion and especially on horizontal camera pans. However, the 3:2 pulldown has the advantage of having the dups appear more frequently. The pattern of duplicating just one out of every five frames is much more likely to call attention to itself.

Back to your revised question: how to create video which combines 50 fps with 60 fps. First of all, you have to tell us whether the footage is interlaced or progressive. Second, you should tell us what NLE you plan to use. Depending on your NLE, there are other options besides interpolating frames (or fields, if interlaced), or simply duplicating fields or frames. (You should duplicate fields, not frames, if your video is interlaced.) As one example, my NLE, Vegas, has an option to do frame rate conversion by blending adjacent fields or frames. Like every approach, this has its own set of pros and cons. On the pros side, you do not get any jerkiness. You also don't get the occasional horrible artifacts which motion-estimated interpolation can create. The one con is that the new intermediate frames may look a little soft during high motion. You should do a test, if you have access to this technology, and see what you think.

Simply speeding up the 50 fps is an intriguing idea but, as poinsondeathray (a person you should always listen to) points out, the 20% speed increase is too much of a change, whereas the 4% speed up to get 24p film to play at 25fps PAL speed is acceptable.

But, consider this: what if you changed both the 50 fps and the 60 fps footage? If you changed them both to 54.5 fps, that would only be a 9% change for each one. If YouTube only accepts integer FPS (I would try a test to see), then use 55 fps. Since I do know for a fact that YouTube will accept any framerate up to 60 fps, it will be quite happy with 55 fps. As long as there is no music which must be matched between the two sources at the cut point, this might work pretty well.

[edit]My last statement was stupid: you can put the audio through a conversion which maintains the pitch.

[Skiller]

How will ChangeFPS look like garbage if it's just duplicating frames?

Because the pattern 50—>60 is very uneven and there is no room to distribute that pattern nicely. For example, 25 fps to 60 fps using ChangeFPS looks quite alright (arguably the next best thing after not changing the frame rate) because there is plenty of room to "hide" the pattern.

[jagabo]

50 to 60 fps via ChangeFPS() (duplicate frames) doesn't look too bad. Many people won't even notice the 10 little jerks per second. Another thing to keep in mind: if you're viewing on a 60 Hz monitor or TV you're going to get those same 10 jerks per second encoding at 50 fps.

[_Al_]

I also would not be avoiding using 50 to 60 via changeFPS(), for internet it is just fine, most "economic, fastest and usable solution". It is not some BD delivery. If doing it in NLE it is blurred abit if blend modes are used and also do not forget this detail: graphic are added on top of videos, so it messes up things, better just preprocess it automatically (human just starts it, loads it, done) . Then load only videos with specs. So it could be automatized for anyone who wants to post or broadcast different frame rates, or mixing different specs, to just use QTGMC(if needed) and resize (if needed) or changeFPS (if needed) or both. Think of it like any new video outside of specs goes thru this and only then edited. Because in theory anyone can come up with best solution for particular scenario, but in real life, there is no time. Content itself is 10x more valuable.

[jagabo]

If you want blended frames with AviSynth you can use ConvertFPS() instead of ChangeFPS().

[_Al_]

ok, well, I just revisited scripts and realized ConvertFPS is used instead of ChangeFPS. It was put together years, years ago, and I'd swear, ChangeFPS was used. But I was wrong. Everything was tested with comparisons and overall look. So if changing 50i to 60p, ConvertFPS sort of looked better than ChangeFPS after QTGMC. Sorry, for some reason I thought it was the other way.