How to export for YouTube, TikTok and everywhere else

The export dialog asks six questions, and most people answer them the same way every time without knowing what any of them do. Three of the six genuinely matter, two matter only in specific cases, and one is decided entirely by where the video is going.

Before any of them, the fact that settles most arguments: every platform re-encodes whatever you upload. YouTube, TikTok, Instagram and LinkedIn all take your file, throw it away, and make their own from it. Nothing you upload is what anybody watches.

That single fact has a clear consequence. Your job is not to guess the platform's settings — it is to hand it the best master you reasonably can, so the quality is destroyed once rather than twice.

Resolution: the only setting that can lose detail permanently

Exports run from 480p to 4K, plus Match source, which mirrors your largest source file so nothing is downscaled.

Two rules cover almost every case:

• Never export smaller than you intend to deliver. A 1080p export of 4K footage throws away three quarters of the pixels, and no upload setting brings them back.
• Exporting bigger than your sources does not add detail. Rendering 4K from 1080p footage produces a large file full of the same information, slightly softer. If you actually need more resolution, that is an upscaling job on the footage, and it has real limits worth understanding before you rely on it.

If a finished export looks soft, this is nearly always the reason: a 1080p render from 4K sources when Match source was what you wanted.

Frame rate: match the source, then leave it

24, 30 or 60. Pick what your footage is, and if your footage is mixed, pick what most of it is.

Exporting 24 fps material at 60 does not make it smoother — the renderer duplicates frames to fill the gap, and duplicated frames look exactly like the original judder in a bigger file. Genuinely adding frames is a different operation, and there is a whole post on doing it properly.

Use 60 when the source is 60 and the motion is fast: gameplay, sport, screen recordings with a lot of scrolling. Use 24 for anything that wants to look filmic. 30 is the safe default for talking heads and social.

Codec and quality

H.264 is the universal answer. Every player, every platform, every phone, every browser.

H.265 halves the file size at the same quality, which is a real benefit for a 4K master you are moving around — but it previews poorly in some browsers, so an H.265 file that looks broken in a preview window is usually fine in a normal player. Use it when file size is your problem and you know what will open it.

WebM (VP9) is the one that keeps real transparency: timeline gaps and keyed areas export as alpha. That is the format for an overlay, a logo animation or a cut-out subject destined for another edit. It is not a delivery format for social.

The quality preset — High, Balanced or Small file — trades size against fidelity. Given that the platform re-encodes anyway, use High for anything being uploaded. Small file is for email, a Slack message or an embed you are hosting yourself, where the megabytes are actually yours to pay for.

Two more formats, for completeness. GIF is capped at 15 fps, has no audio, and produces very large files on anything longer than a few seconds — it is for short loops, not for video. MP3 exports the audio mix alone, at 320, 192 or 128 kbps depending on the quality preset, which is how you get a podcast cut or a voiceover out of an edit.

Shape: decide it before you edit, not at export

The export offers 16:9, 9:16, 1:1 and 4:5, and the aspect chip above the preview sets the same thing for the project.

Set the shape at the start and frame the edit for it. Exporting a vertical file from an edit that was composed wide gives you a centre crop of every shot, and centre is rarely where the subject is.

Converting a finished wide cut to vertical is a different job with a proper tool: reframing re-crops shot by shot and keeps the subject in frame instead of throwing the sides away.

The settings, per platform

YouTube (landscape). 16:9, H.264 MP4, High, at the highest resolution your footage supports — 1080p, or 4K if you have it. There is a widely observed effect worth exploiting: YouTube's encoding ladder treats high-resolution uploads more generously, and a 4K master often ends up looking better at 1080p than a native 1080p upload does. Frame rate matches your source.

Shorts, Reels and TikTok. 9:16 at 1080×1920, H.264, 30 fps unless the source is 60, High. Keep the important part of the frame out of the bottom fifth and the right third, where the interface sits.

Instagram feed. 4:5 takes up more vertical space than 1:1 in the feed, which is free attention. Same codec, same quality.

LinkedIn and X. 1080p H.264, and burn the captions in — autoplay is muted and neither platform's own captioning is reliable enough to depend on.

A web page or an email. 720p, Small file, H.264. Nobody waits for a 200 MB hero video.

An overlay for another edit. WebM, Match source, so the transparency survives.

In all of these, export one file per destination. A single "good enough for everything" file is a vertical video letterboxed on YouTube, or a wide one pillarboxed on a phone.

Audio, which nobody checks

Platforms normalise loudness. YouTube quietly turns loud uploads down to around −14 LUFS; the others do their own version of the same thing. Exporting everything as loud as possible therefore achieves nothing except a squashed mix that gets turned down anyway.

Use the export's cleanup options — loudness normalisation and noise reduction are free switches — and mix so the voice is clearly above the music rather than so the whole thing is loud. Then listen to the exported file on phone speakers, which is the only listening test that predicts anything.

Practical notes on the render itself

Exports render on the server, so you can close the tab or keep editing while one runs; progress shows in the nav pill and refreshing the page picks the tracking back up.

The finished file lands in your library as a reusable clip and appears under Recent exports — which means an export can be dropped straight back onto a timeline. That is the trick behind making a vertical cutdown of a finished video: export the master, then edit the export rather than rebuilding the project.

Renders are queued, and your plan's allowance is shown before you submit rather than sprung on you afterwards. There is no watermark on any format, at any plan, in any resolution.

The order that works

1. Set the shape before you edit, not at export.
2. Match source, or the highest resolution your footage supports.
3. Match the source frame rate — never "upgrade" it.
4. H.264 and High for anything being uploaded; H.265 or Small file only when size is the actual problem.
5. WebM when you need transparency, MP3 when you need the audio alone.
6. Export one file per destination, reframing rather than cropping.
7. Normalise the audio and check it on a phone before it goes anywhere.