How video editing softwares change the FPS from ~29.87 FPS to 30 FPS. Where do the extra frames come from?
In: 39
The video is sped up by an *imperceptible* amount, or slowed down to get the opposite. The fractional frame rate is an artifact from analogue television, where the numbers were chosen to reduce interference between the color and sound subcarriers.
Now we are stuck with rates such as 30000/1001 and 24000/1001 for TV and film respectively.
The easiest way is that the video simply becomes shorter. The audio has to be sped up very slightly but that’s easy to do. The speed change is exactly in the ratio 1001:1000 which is too small of a change in pitch to be noticed (less than 2 cents) and means the video is only 3.6 seconds shorter per hour.
There are no extra frames. Imagine a video has 1000 frames. At 10fps, it’d be 100 seconds long; at 30fps it’d be 33.3 seconds long; at 29.87fps it’d be 33.48s long. The difference between 10fps and 30fps is very noticeable, but the 0.18s difference between 29.87 and 30 isn’t noticeable.
Depending on software. But they can simply interpolate one frame from two neighbouring frames every 1000 frames.
This is done for many 24 to 30 frames conversions. Every other frame gets interpolated from it’s two neighbours.
Ok, first, one thing to clarify.
You have 29.87 in your question, but that’s not a usual framerate. 29.97 is though. however, if you’re looking at a video you’ve recorded on a phone it may indeed show 29.87 as the framerate. Thats something a little different.
if its 29.97, the switch to 30 is really tiny and mostly unnoticeable. its usually just a tiny retiming of the end of each second. If you’re working with multiple cameras it can be slightly noticeable, but usually not. And the framerate is an artifact of film anyway, as back then, you could have .97 of a frame pulled through, and after 33 seconds, its now 1 frame behind a 30fps file (and would need about 17 minutes to drift by a second). With digital its just slightly adjusting the timing.
OTOH, since you put `~29.87` rather than `29.97`, I’m thinking cellphone and the rest of this is with that in mind. and the reason I’m thinking cellphone is that all cellphones (that I’m aware of) do this really annoying thing when recording video called ‘variable framerate’. The phone will change the framerate depending on the situation, if there’s little motion it’s going to reduce the framerate slightly, which can make for a smaller file, and it can up it if there’s movement. It can also reduce it if it’s kinda dark, as it increases the shutter time to give a longer exposure (just like in photography, After all, whats video but lots of photos being taken one after another?) – I have one camcorder that will drop to 5 frames a second in very low light mode, but unlike a phone it still records at the same frame rate all the time – it may be using a shutter speed of 1/7s, but the 5fps at the sensor is recorded as 30fps with 6 copies of each frame. The framerate on a phone though is not constant at all, a graph of the framerate over time looks like a polygraph result – so the 29.87 is just an average value. Ive had files from a cellphone set in a tripod for a 2 hour show that gave me files ranging from 27fps to 32 fps, and all were supposed to be 29.97.
So you can convert the footage from variable framerate, to constant framerate, which you have to do if you’re going to match the footage with any other camera (else it’ll go out of sync every few seconds) – most people use either handbrake or shutter encoder. To date no editors that i’m aware of handles VFR, so it has to be done.
A spring in my room is 10 feet long and has 30 rings per foot. If I shorten it slightly it becomes 29 springs per foot or 29.5 or 29.87 but it is no longer 10 feet long. The video still has the same frames but the difference in speed has the consequence of a difference in length. Slow it down it gets longer. Speed it up it gets shorter.
If you simply need to play back at little higher rate, you can simply speed up the video — a 1% speed up is going to be imperceptible.
However, for video editing, it’s a tad bit more complicated. You might be taking video of different frame rates and combining them to make a video of yet another rate.
In this case, the software simply maps a frame to a time (maybe as millisecond resolution) in the timeline. When you go to any point in the timeline, it figures what frame is displayed at that specific time and uses that. The frame rate is pretty much ignored.
For example, imagine two videos: A with a frame rate of 3 fps, and B that’s 4 fps, and we are going to add them to a project that will be 10 fps. Each frame in A is 333.33 ms and in B 250 ms. So in video A frame 0 is 0-333.33 ms, 1 is 333.34-666.66 ms, etc. For video B, frame 0 is 0 to 250 ms, 1 is 250.01 to 500ms, etc. For the 10 fps timeline, it has frames at 0, 100, 200, 300, 400, 500, 600 ms … etc. We can simply figure out which frame is displayed at a given time. If we write it (Time, Frame A, Frame B): (0, 0, 0), (100, 0, 0), (200, 0, 0), (300, 0, 1), (400, 1, 1), (500, 1, 1), (600, 1, 2), … and so on.
I used to work on video editing software in a previous job, for all framerate changes it’s done via image/frame interop. (Combining 2 frames to produce a merged in-between frame)
It’s more complicated than just adding 13 additional frames for every 100 frames of the video because that would just cause a weird speed up/slow down situation as the original 29.87 frames would be playing at a slightly faster 30fps rate while the inserted 13 frames would cause a slight gap/delay.
What ends up happening is each frame is rebuilt, with a percentage based formula of how much of the previous and next frame should be used, over the course of this process new extra frames are also introduced but generally speaking most frames of the video have been altered slightly to keep the timing correct.
Latest Answers