Skip to content
navigationLanguage

Which frame rate to use: fps 24, 25 or 30 for editing and delivery

The source sets the frame rate, not your wish. We look at when a conversion makes sense, what it really does to the file, and what the platforms require — with measurements you can repeat.

Which frame rate to use: fps 24, 25 or 30 for editing and delivery — the answer is simple, the one the material was shot at. Conversion is worth it only when a platform or the client requires it. Below is what happens to the file when you convert it, with our own measurements: they give a result opposite to the expected one.

Which frame rate to choose

The rule is simple: take the rate of the source and do not change it without a reason. 24 fps is cinema, 25 is the European television standard (PAL, 25 and 50 fps), 30 is the North American one (NTSC, 30 and 60 fps). YouTube accepts 24, 25, 30, 48, 50 and 60 fps. The timecode counter counts by the nominal value: at 29.97 fps as 30, at 23.976 fps as 24. That is how the format itself works, not a decision of the platform.

What the platforms themselves say

YouTube requires video to be encoded and uploaded at the same rate it was recorded at, and warns: “Please do not use resampling methods — they cause jitter in the picture and often a reduction in quality”. The formatting guide on the same site says: “the video must be in the original frame rate, without resampling”, and for cinema sources 24 or 25 fps are recommended. So a conversion on the platform side is not needed and does harm. VK Clips is softer: there, Full HD and 60 fps are supported in the vertical format 9:16 and 1080×1920, the running time runs from 3 seconds to 3 minutes, and the maximum file size is 1 GB.

The difference between 24, 25 and 30 fps

Parameter24 fps25 fps30 fps
Frame rate242530
Where it is usedcinema, author shootingPAL, European TV and lightingNTSC, internet, studio shooting
Frames in a 10-second span240250300
Over 10 secondsthe fewest framesmore than 2425 % more than 24
Between 24 and 25 fps the difference in rate is one frame per second, between 24 and 30 it is already 25 %. In practice this means that a conversion changes both the number of frames and their distribution over time. How visible that difference is to the eye, we did not measure: image quality was not scored in our measurements. The practical conclusion: if the output rate is known in advance, it is more sensible to shoot at it right away and not to convert finished material.

What happens during a conversion

We took a test clip of 1920×1080 at 25 fps, 30 seconds long (750 frames), and converted it to other rates in three ways. The source weighed 29 796 554 bytes after encoding, that is 7.95 Mbit/s. Below are the results; the rows are named exactly as in our measurement protocol, so each of them can be checked against the protocol file.
VariantFramesFile sizeBitrateSize change
Source 25 fps75029 796 554 bytes7.95 Mbit/s—
→ 30 fps, frame duplication90029 073 502 bytes7.75 Mbit/s2.4 % smaller
→ 30 fps, motion interpolation89832 373 785 bytes8.65 Mbit/s8.6 % bigger
→ 24 fps72028 681 717 bytes7.65 Mbit/s3.7 % smaller
→ 23.976 fps71928 708 837 bytes7.66 Mbit/s3.7 % smaller
→ 60 fps, motion interpolation179637 956 427 bytes10.14 Mbit/s27.4 % bigger

Why duplication does not increase the file

This is the most unexpected result of our measurements: when going from 25 to 30 fps with ordinary frame duplication, the file became smaller — 29 073 502 bytes against 29 796 554 for the source. The number of frames grew by 20 % at the same time.
The reason is how encoding works. A repeated frame carries almost no new information, and the codec spends a minimum of bits on it. The result is that the file shrinks while the picture does not: 150 frames appear in the edit sequence that show nothing new, that is 16.7 % of the running time. A repeated frame is perceived visually as a pause — how visible that exactly is, we did not measure.
Motion interpolation solves this differently: it draws the intermediate frames, and the movement becomes smoother. But the file pays for it — 32 373 785 bytes, 8.6 % more than the source. The synthesized frames contain pixels that were not in the source, and the codec has to encode them. The higher the target rate, the more expensive the interpolation: converting the same material to 60 fps with interpolation gave 37 956 427 bytes and 10.14 Mbit/s — 27.4 % more than the source, and that is with the frame count more than doubling.

What the platforms require

YouTube accepts any of the standard rates: 24, 25, 30, 48, 50 and 60 fps, but asks you not to resample. In its encoding guide the platform gives the bitrate separately for the normal rates (24, 25, 30) and the increased ones (48, 50, 60): for 1080p — 8 and 12 Mbit/s, for 1440p — 16 and 24 Mbit/s, for 2160p (4K) — 35–45 and 53–68 Mbit/s. So the platform expects the higher rate to cost more data and sets a limit for it.
VK Clips has its own requirements: the vertical format 9:16 and 1080×1920, from 3 seconds to 3 minutes, Full HD and 60 fps are supported, the maximum size is 1 GB. Here 30 or 60 are perfectly reasonable, and the video is usually shot vertically from the start — and then no conversion is needed at all.
The rest is a question of the order of work: choose the bitrate and the codec for the fps, not the other way round. It is also worth keeping the timecode in mind: at 29.97 fps a timecode without drop-frame loses 3.6 seconds per hour of real time, while the drop-frame variant matches real time. For syncing sound and picture that is the difference between an even edit and a permanent fix.

Frequently asked questions

Do you need to convert 25 fps to 30 for YouTube?
No. YouTube accepts 25 fps without conversion and warns directly that resampling causes jitter in the picture and a reduction in quality. Conversion is needed only if the client or the platform requires a specific rate. If the video is already finished in 25 — leave it in 25.
What is better for vertical video — 30 or 60 fps?
It depends on the source. If it was shot at 30, there is no need to convert to 60: you get 60 fps, half of which are duplicates. We measured this: the 30 → 60 fps conversion with duplication gave 1800 frames instead of 900, of which 900 are duplicates — 50 % of the timeline carries no new information, and the file at the same time became smaller: 30 093 816 bytes (8.03 Mbit/s) against 32 521 385 bytes (8.67 Mbit/s) of the 30 fps source, that is 7.5 % smaller. The picture did not become smoother. 60 fps makes sense when the source was shot at that rate.
What is 23.976 fps and why is it needed?
It is 24 frames per second, exactly 24000/1001, the cinema standard that is slightly slower than 24 because of the fractional division. In our measurement the conversion from 25 to 23.976 gave 719 frames in 29.988 seconds: the duration barely changed, and the rate became exactly 24000/1001. The drop-frame scheme does not apply to it: it is needed only for the whole 30 and 60, where the counter counts by the nominal value. And the NDF counter itself lags at 23.976 — by about 3.6 seconds per hour, exactly as at 29.97; our timecode calculator shows it.

What to read next

Need this done as a project?

We take on projects like this — with the same settings we walk through in this piece.