Page 1 of 1

Question about FPS variations

Posted: Thu Jan 24, 2019 11:09 am
by roadrider
I have a video that is 10:00 minutes in duration. It is recorded as follows:
FPS = 29.97
Bitrate = 1200 kbps
screen size=640x480

if I just increase the fps from 30 to 60, what is actually happening internally to the video. ( I tried it and the video duration was the same and the size on disk increased a bit). there was not noticeable difference in the video. what I thought was that the video will speed up by 2x and the duration would be 5:00 minutes. but this did not happen. can you please explain what this means in technical terms.

Re: Question about FPS variations

Posted: Sat Feb 09, 2019 8:30 am
by mirrorman
I'm not an expert on ffmpeg, but I think I know something about frame rate - and as no one else has replied...

There are three basic possibilities if you double the frame rate of a video:
1. You duplicate each existing frame and play them in sequence twice as fast. As you might imagine, this would result in no visible change compared to the original, but it might make the video technically compatible with some system that requires all footage to be at 60 fps.
2. Instead of duplicating the frames, you make every other frame a blend (i.e. a 'dissolve') of the two frames on either side. The visible effect of this is not much different from the original video.
3. You use an algorithm to track the motion of moving areas within the image from frame to frame, and synthesise an intermediate frame where those areas have moved half as far. The effect of this is to go from 'film effect' to 'soap opera effect'. i.e. the result looks like TV video, news or sports footage rather than (close to) cinema footage. The effect is striking and can add smoothness and immediacy to material that would benefit from that. But used inappropriately, you get this: ... ai-5969817

With ffmpeg you can achieve (2) and (3) using the minterpolate filter and the appropriate settings. I am guessing that simply doubling the frame rate without minterpolate will give you (1).