Guide · corrected 9 October 2026

Why some TikTok videos stay 1080p and others drop to 576p

This guide said the answer was frame rate. It is not.

An earlier version said every 30 fps upload came back at 576p, from eleven measured videos. We then posted one clip twice — same footage, same export, once at 30 fps and once at 60 fps — and both came back at 1080p. One counter-example is enough to retire a rule that used the word every, and we made that one ourselves.

The measurements below are unchanged and still real. What changed is what we think they mean. The old conclusion is kept further down rather than deleted, because a site that quietly edits its mistakes is worth less than one that shows them.

Two creators post from the same kind of phone. One video comes back sharp, the other soft. We measured eleven videos to find what separated them, published the answer, then tested it properly and found we were wrong. Here is what survived.

The short answer

Frame rate does not decide it. What best fits every video we have measured is how much detail the upload itself carried — and that is something neither we nor you can read off someone else’s post. The practical advice barely moves; the reason behind it does.

The test that broke the old answer

One clip, exported twice from the same source, differing in frame rate and nothing else. Both posted to the same account within seconds of each other, then left alone until TikTok finished processing them.

Uploaded at	Served as
30 fps	1080p
60 fps	1080p

The 30 fps upload kept 1080p. Under the old rule it should have come back at 576p, and it is the only kind of result that could have settled the question — eleven videos that happen to agree cannot establish a cause, but one that disagrees can remove it.

A note on how this was read, because it differs from the rest of the page: TikTok serves recent posts only to a signed-in browser, so the checker could not reach these two files. Both were measured from the resolution the player was actually decoding. That is a real reading of a real stream, but it is not the same instrument used for the eleven below.

The eleven videos, read again

Nothing here has been changed since the original version except the order of the rows. They were sorted by frame rate, which is what made frame rate look decisive. Sorted by the bitrate TikTok serves them at, they split more cleanly — and the one video that never fitted the old story falls into place.

Served as	Served bitrate	Uploaded at	Plays
576p	3.5 Mbps	30 fps	29,600,000
576p	3.6 Mbps	30 fps	1,300,000
576p	4.3 Mbps	30 fps	3,500,000
720p	4.3 Mbps	60 fps	6,800,000
1080p	13.2 Mbps	60 fps	338,500
1080p	17.9 Mbps	60 fps	336,900
1080p	39.6 Mbps	60 fps	26,400
1080p	40.0 Mbps	60 fps	2,300,000
1080p	40.1 Mbps	60 fps	347,000
1080p	50.3 Mbps	60 fps	3,242
1080p	107.3 Mbps	60 fps	79,000

Everything at or below 4.3 Mbps is reduced. Everything at or above 13.2 Mbps is 1080p. There is nothing in between — not a gradient, a gap. And the 720p video, which the old version admitted it could not explain, sits exactly where it belongs: 60 fps, and still reduced, because it is down with the low-bitrate group.

Why this is not simply the new answer

That bitrate is what TikTok serves, which is a result of the same decision that set the resolution — a video reduced to 576p has a low bitrate because it was reduced. Reading it as the cause would be running the explanation backwards, which is close to the mistake this page is already correcting once.

What the gap does suggest is that those four files arrived carrying less real detail than the other seven, and that whatever TikTok decides, it decides on the upload rather than on the frame rate. That is a hypothesis about something we cannot see.

What we have ruled out

Three explanations have now died against the data, in the order we tried them.

Popularity. The theory was that TikTok serves cheaper copies of videos that get expensive to deliver. It fits the top of the table and then breaks: a video with 2.3 million plays is served at 1080p while one with 1.3 million comes back at 576p. Both from the same creator.
Age. The theory was that older videos went through an older pipeline. A 20-day-old video is served at 576p while videos of 31, 33 and 40 days are all 1080p. If age drove it, that would run the other way.
Frame rate. It survived the first two and split all eleven videos cleanly, so we published it. Then the controlled test above returned 1080p for a 30 fps upload. It was correlated with the outcome in that sample and was not causing it — which is the ordinary way a finding from eleven observations fails.

The limit on all of this

We can only measure what TikTok publishes. We cannot see the file anyone uploaded. So a video served at 576p might have been uploaded at 576p, in which case TikTok reduced nothing and what we are observing is only that some people export smaller files than others.

No number of extra videos fixes that, because the upload is never visible from outside. The only way past it is the kind of test above, where we control the upload because we made it — and that is also how the old answer here was disproved.

What to do with it

Less of this rests on frame rate than the previous version claimed, and more of it rests on guides where the mechanism is actually understood.

Export at a generous bitrate. The seven videos that kept 1080p were all served above 13 Mbps; the four that were reduced were all under 5. That is an outcome rather than a setting, but nothing we have measured suggests a thin file survives and four cases suggest it does not.
Keep the file off the phone app’s compressor. This is the one mechanism on this site that is actually established: the TikTok apps re-encode on the device above roughly 72 MB on Android and 288 MB on iOS. Why your videos look blurry covers it.
Export at 1080 × 1920, H.264, once. The export settings guide covers why, including what changes when nothing re-encodes your file.
Shoot at whatever frame rate suits the shot. We no longer have a reason to tell you 60 beats 30 for this. If you like 60 fps for motion, use it — just not because of this page.
Check the result. Run your own post through the quality checker and see which column you landed in.

Try it on your own posts

The useful version of this is not our eleven videos, it is your own. The last version of this page asked for a 30 fps video that kept 1080p, on the grounds that a finding this size should be easy to disprove. That turned out to be true, and the counter-example was ours.

The question now is what separates a 576p serve from a 1080p one, and the answer is not in hand. If you have two of your own posts that differ in one thing and landed in different columns, that is worth more than another eleven we cannot see the uploads for — send them over.