Best method for time lapse video

[jairovital]

I've been seen to many time lapse videos nowadays that I wonder the best method to do them. Is it just adjusting the frame rate? Or are there a better tool around? Avidemux only allow to increase fps to max 200.

The question is: how to make a good time lapse video? A frame rate issue? Or is there another tool that collect few frames and goes ahead many others to collect more?

[transporterfan]

What is your definition of 'time lapse'? Can you link to an example?
Time lapse usually means taking a frame and letting a set time period go by (lapse) before taking the next frame.
At 200 fps what you would get is 'slow motion' when viewed at normal speed, say 30 fps.
Time-lapse means taking 1 (ONE!) frame every 10 or 15 or whatever-you-decide SECONDS apart!

Here is a brilliant example of time lapse video:

http://vimeo.com/9679622

And an introduction on how to do it here:

http://www.hongkiat.com/blog/how-to-make-time-lapse-video-ultimate-guide/

[jairovital]

At 200 fps what you would get is 'slow motion' when viewed at normal speed, say 30 fps.

Yeah, you're right! I had mentioned 200 fps just because this is the limit for Avidemux. I'd like to input 2000 or even much more.

I know what time lapse is, but I'd like to apply that effect to a video already made normally, not to adjust the camera to take frame by frame along a entire day, but to steal many frames from an already made video. Got it?

Intervalometer and D-Software Cam Control takes control of the shutter interval. But, as I said, the video is alredy done.

How to make automatic the retreat of many frames from a video?

[transporterfan]

So, what you mean is, 'I already have footage from a camcorder taken at 60fps. What program do I use to automatically extract frames @60 frames apart (1 second interval), put them on a time line, let me edit them in anyway I like, and rebuild a video from these selections?'

If that is what you mean then my answer is: I haven't a clue.

Anyone else know?

[poisondeathray]

avisynth 's selectevery() or vdub

see
https://forum.videohelp.com/threads/317543-Trying-to-make-a-time-lapse-video-%28speedin...footage%29-How

[jairovital]

avisynth 's selectevery() or vdub

Poisondeathray, as usual, you have my answer!
Of course, Avisynth!

I was thinking about some tool, some GUI. But I can always use Avspmod.

Except Avisynth, any tool with graphic interface out there to do this task?

[poisondeathray]

Except Avisynth, any tool with graphic interface out there to do this task?

vdub , scroll down in that link for instructions

https://forum.videohelp.com/threads/317543-Trying-to-make-a-time-lapse-video-%28speedin...=1#post2002122

or any NLE

[jairovital]

Great, poisondeathray! That thread has everything I wanted to know!

[OrrinH]

I did this recently using time remapping in After Effects. Here's a good write up about it: http://help.adobe.com/en_US/aftereffects/cs/using/WS3878526689cb91655866c1103906c6dea-7d4da.html

[jairovital]

I did this recently using time remapping in After Effects. Here's a good write up about it: http://help.adobe.com/en_US/aftereffects/cs/using/WS3878526689cb91655866c1103906c6dea-7d4da.html

Another good method. Could you post a small sample of what you did here?

[OrrinH]

Ahh not at the moment sorry, it's a segment of a longer clip which is still a WIP.

It was a 10 minute recording which we shrunk down to about 15 seconds with 10 seconds of normal speed play at either end. Took no more than 5 minutes to setup, which I was really happy about but the rendering speed took a massive hit because of it.

[budwzr]

Yeah, that would be a simple thing any NLE should be able to do. I suppose you could buy a $1000 software like AE to do it, but not necessary. Save AE for more important things like smoke puffs, comets, and exploding planets.

[jairovital]

avisynth 's selectevery() or vdub

Poisondeathray, if you still around, how I do the same for the audio?
I mean, how to change the speed of the audio to match the video?

Reading SelectEvery I know I can use SelectRangeEvery with audio false or true. But it doesn't work with DirectShowSource or ffVideoSource/ffAudioSource. Audio is always the same (original audio, not speeded up).

DirectShowSource("D:\work\test1.mp4")
#A = FFAudioSource("D:\work\test1.mp4")
#V = FFVideoSource("D:\work\test1.mp4")
selectevery(last, 10)
#selectrangeevery(V, 10, audio=true)
AssumeFPS(30000,1001)
#AudioDub(V, A)

BTW, using ffmpegsource, selectevery/selectrangeevery dosn't work. I must use DirectShowSource to get the fast foward effect. Why?

My goal:
Learn how to put into an avisynth script the right lines to speed up audio and video, with synchronism.

Other questions: what's the difference between DirectShowSource and ffmegSource when the similarity and distance (in quality) from the original is important? Which are the better if I want to preserve the most of the original?

[poisondeathray]

I don't know if SelectEvery processes audio, but SelectRangeEvery apparently does according to the documentation

The reason why DirectShowSource works, but FFVideoSource doesn't, is SelectRangeEvery must come after audiodub (otherwise you are using video only) . DirectShowSource has audio=true by default, so includes video+audio by default

Or you can use FFMpegSource("video.ext", atrack=-1) . It's a helper function that simplifes FFAudioSource and FFVideoSource with AudioDub

DirectshowSource is inconsistent and may be not frame accurate in some situations (e.g. temporal filtering) , but FFMS2 has some quirks as well (sometimes frame rate is a bit off, but you can fix with AssumeFPS, or fpsden, fpsnum in the source filter)

[jagabo]

SelectEvery() doesn't process the audio because it doesn't change the video length. Ie, a 10 second 30 fps video after SelectEvery(10) is still 10 seconds long, but only 3 fps. If you want to make it play faster you have to AssumeFPS(30) after. But that doesn't change the audio either. So you'll have 3 seconds of video with 10 seconds of audio.

You can change the audio duration with:

#assuming 48000 Hz audio
ResampleAudio(4800).AssumeSampleRate(48000)

But that will give you very high pitched audio. I don't know a way to keep only the audio from every 10th frame. That's not to say there isn't a way. I don't do much audio processing in AviSynth.

[jairovital]

The reason why DirectShowSource works, but FFVideoSource doesn't, is SelectRangeEvery must come after audiodub (otherwise you are using video only)

Done, but ffVideoSource/ffAudioSource doesn't work. Video and audio play in normal speed, they don't speed up. Audio true or false in SelectRangeEvery doesn't change the speed.

Or you can use FFMpegSource("video.ext", atrack=-1).

Done, but "there is no function named ffmpegsource" error message...

[jairovital]

SelectEvery() doesn't process the audio because it doesn't change the video length. Ie, a 10 second 30 fps video after SelectEvery(10) is still 10 seconds long, but only 3 fps. If you want to make it play faster you have to AssumeFPS(30) after. But that doesn't change the audio either. So you'll have 3 seconds of video with 10 seconds of audio.

Good explanation.

But there is no way to speed up the audio in avisynth?

[poisondeathray]

The reason why DirectShowSource works, but FFVideoSource doesn't, is SelectRangeEvery must come after audiodub (otherwise you are using video only)

Done, but ffVideoSource/ffAudioSource doesn't work. Video and audio play in normal speed, they don't speed up. Audio true or false in SelectRangeEvery doesn't change the speed.

Not sure. I'll have a look. Normally you don't use original audio for this sort of thing (glitchy audio because you are sampling every x frames)

Note sampling the audio every x frames isn't the same thing as time compressing the audio

Or you can use FFMpegSource("video.ext", atrack=-1).

Done, but "there is no function named ffmpegsource" error message...

Typo. It should be FFMpegSource2 . You need the .avsi included with the ffms2 package

[jagabo]

SelectEvery() doesn't process the audio because it doesn't change the video length. Ie, a 10 second 30 fps video after SelectEvery(10) is still 10 seconds long, but only 3 fps. If you want to make it play faster you have to AssumeFPS(30) after. But that doesn't change the audio either. So you'll have 3 seconds of video with 10 seconds of audio.

Good explanation.

But there is no way to speed up the audio in avisynth?

I added one way in that post.

[jairovital]

You can change the audio duration with:
#assuming 48000 Hz audio
ResampleAudio(4800).AssumeSampleRate(48000)

This will match the audio with selectEvery?

But that will give you very high pitched audio.

I know. But my purpose is just to learn.

Typo. It should be FFMpegSource2 . You need the .avsi included with the ffms2 package

It's fine, here.

[jairovital]

ResampleAudio(4800).AssumeSampleRate(48000)
I'm testing and getting real pitched audio here. Not sure if it matches the video speed. I'll try later with some explosions, because it's easier to check the syncronism.

[poisondeathray]

I don't think resampling the audio will help for what you want to do because you are selecting every "x" video frames . The other frames are discarded. So audio won't match exactly

Resampling audio means you keep all original audio samples, just speed it up.

SelectRangeEvery(audio=true) does keep the audio from every "x" frames, so it matches the video, but then you get glitchy audio (the video drops match the audio drops)

[jagabo]

For example, a 2x speedup:

AviSource("SuntoryBeer2.avi") 
SelectEvery(2).AssumeFPS(23.976)
ResampleAudio(24000).AssumeSampleRate(48000)

[jairovital]

I don't think resampling the audio will help for what you want to do because you are selecting every "x" video frames . The other frames are discarded. So audio won't match exactly

You are right. Testing here, resampling audio gives the same result no matter selectEvery is.

SelectRangeEvery(audio=true) does keep the audio from every "x" frames, so it matches the video, but then you get glitchy audio (the video drops match the audio drops)

Good, but my SelectRangeEvery here doesn't work... It plays the same speed, audio and video...

[jairovital]

For example, a 2x speedup:

AviSource("SuntoryBeer2.avi") 
SelectEvery(2).AssumeFPS(23.976)
ResampleAudio(24000).AssumeSampleRate(48000)

Thank you, jagabo. But if you change SelectEvery to 5, the audio will be the same as 2. The resampling keeps the audio at same speed, regardless SelectEvery.

[poisondeathray]

I don't think resampling the audio will help for what you want to do because you are selecting every "x" video frames . The other frames are discarded. So audio won't match exactly

You are right. Testing here, resampling audio gives the same result no matter selectEvery is.

SelectRangeEvery(audio=true) does keep the audio from every "x" frames, so it matches the video, but then you get glitchy audio (the video drops match the audio drops)

Good, but my SelectRangeEvery here doesn't work... It plays the same speed, audio and video...

Seems to work ok here. The frame rate is the same, but it's sampled every "x" frames for video & audio

eg.

ffmpegsource2("video.ext", atrack=-1)
selectrangeevery(4, 1, audio=true)

would give you every 4nd frame (video & audio)

Here is an example - Ironman movie trailer

[jairovital]

Seems to work ok here. The frame rate is the same, but it's sampled every "x" frames for video & audio

ffmpegsource2("video.ext", atrack=-1)
selectrangeevery(4, 1, audio=true)

Great! Now it's working.

My mistake:

selectrangeevery(last, 10, audio=true)

No!

The right line:

selectrangeevery(last, 10, 1, audio=true)

Big thanks!

[Gavino]

An alternative way to handle the audio (possibly better, depending on the effect you want to achieve) would be to use TimeStretch().

SelectEvery(10).AssumeFPS(last)
TimeStretch(tempo=10)

This speeds up the audio while preserving the original pitch, but uses information from all the original audio samples
(rather than just discarding chunks as SelectRangeEvery does).

[jairovital]

This speeds up the audio while preserving the original pitch, but uses information from all the original audio samples (rather than just discarding chunks as SelectRangeEvery does).

Gavino, your solution is even better! preserves all the audio samples.
Cool.

[jagabo]

In the version of AviSynth I have TimeStreth's tempo parameter is percent. Ie, tempo=10 makes the tempo 1/10 of the original (very slow). If you want it twice as fast you specify tempo=200. 10 times as fast, tempo=1000.