What YouTube, Meta, X and TikTok actually publish for video aspect ratios
· 9 min read
Every video aspect ratio table is an uncited paraphrase. What YouTube, Meta, X and TikTok actually publish, and how to run the arithmetic yourself.

Search for video aspect ratios and you get a table: Instagram 9:16, TikTok 9:16, YouTube 16:9, done. The table is almost never linked to anything. Its numbers are copied out of vendor documentation by someone who did not read the documentation, and the copies have drifted, because the sources say different things on different pages of the same website.
Read the sources and a different picture appears. YouTube's own rule for what counts as a Shorts names no ratio at all. It says square or vertical, and stops there. Meta publishes a six-column matrix where a tick means supported and only one cell in each row means recommended, and in that matrix Instagram Feed recommends 2:3 for video while Threads recommends nothing at all. X tells you 9:16 fills the screen, then tells you 16:9 still runs.
None of that is pedantry. A recommended ratio is a statement about composition and screen real estate. The crop, the padding and the overlays belong to the feed. This post quotes the vendor documents that were read in full, names what they disagree about, and hands you the arithmetic for whatever number your own platform states.
YouTube's Shorts rule names a video aspect ratio and no number
Here is the sharpest single fact in this subject, and it comes from the platform everyone copies 9:16 into. YouTube's help page on three-minute Shorts states the eligibility rule in words and declines to give a number: "Any videos uploaded on or after this date with a square or vertical aspect ratio up to three minutes in length will be categorized as Shorts on YouTube."
The date it names is October 15, 2024. It names two orientations, square and vertical, and not one ratio, not one pixel pair. The page then offers the escape hatch in the same register: if you do not want a video classified as a Shorts, "use a wider aspect ratio such as 16:9 for long-form videos". A qualitative rule, and a qualitative exception.
Official Artist Channels get a near-identical rule phrased "square or taller in aspect ratio", with a cut-off of December 8, 2025. Even the exception to the exception is described as geometry rather than a figure.
So the one platform with a hard mechanical consequence attached to a ratio (reclassify my upload into a different product) describes the ratio in adjectives. If a pixel count were load-bearing it would be printed. It is not, and the guidance that does exist points the other way: YouTube's video resolution and aspect ratio page for the desktop player calls 16:9 the standard aspect ratio on a computer, then adds that if your video has a different ratio "the player will automatically change to the ideal size to match your video and the viewer's device". The player makes the decision.
"Recommended" is not "required": what three vendors say
TikTok's in-feed ad specifications, on a page stamped "Last updated: June 2026", print the distinction explicitly and structurally:
Vertical (recommended): 9:16, more than or equal to 540960px. Horizontal: 16:9, more than or equal to 960540px. Square: 1:1, more than or equal to 640*640px
Three ratios accepted. The word "(recommended)" appears next to one of them and is absent next to the other two. No third-party table reproduces that asymmetry, because every table turns recommended into required on the way through.
X is blunter, on its Vertical Video Ads page: "Use videos with 9:16 aspect ratios. This allows you to maximize the screen real estate of your ads and fully immerse the user in your content." Then, in the next sentence: "While you can run 16:9 or 1:1 aspect ratio videos in the Immersive Media Viewer, they are not ideal for this surface." Nine-sixteen is preferred. Horizontal still plays. X's own wording is also the best available answer to why the vertical shape gets recommended at all. Maximizing screen real estate is a claim about a player filling a phone screen, not about what your file must be.
The arithmetic behind that shape needs no source: 9 ÷ 16 is 0.5625 and 16 ÷ 9 is 1.7778, so 9:16 is 16:9 turned on its side, the same relationship transposed. What that shape was adopted for, and when, is a history question this research pass could not answer from a vendor document, so there is no origin story here.
Note also the minimums TikTok attaches: 540 × 960, not 1080 × 1920. The shape is 9:16 either way. The pixel count in every listicle belongs to somebody else's arithmetic.
One company, six ratios, four surfaces deep
Meta's Aspect Ratios Supported by Placements in Meta Ads Manager is the closest thing to a spec sheet any of these companies publishes, and it is the counter-example to every ratio table ever posted. The page is a six-column grid (1.91:1, 16:9, 1:1, 4:5, 2:3, 9:16) with rows for placements. A tick means supported. Only some cells carry a "Recommended" label, and the gap between the two is the entire information.
| Placement | Recommended for video | Also supported |
|---|---|---|
| Instagram Feed | 2:3 | 1.91:1, 16:9, 1:1, 4:5 (recommended for images), 9:16 |
| Threads feed | none | 1.91:1, 16:9, 1:1, 4:5, 2:3, 9:16 |
| Facebook in-stream reels | 4:5 | 1.91:1, 1:1 (recommended for images) |
| Ads on Facebook Reels | none | 1.91:1, 1:1, 9:16 |
| Instagram Reels | 9:16 | 1.91:1, 16:9, 1:1, 4:5, 2:3 |
| Facebook Reels | 9:16 | 1.91:1, 16:9, 1:1, 4:5, 2:3 |
| WhatsApp Status | 9:16 | 1.91:1, 16:9, 1:1, 4:5 |
| Facebook Feed | 4:5 | 1.91:1, 16:9, 1:1 |
| Facebook Marketplace | 1:1 | 16:9, 9:16 |
| Facebook right column | none | 1.91:1 (recommended for both images and videos), 1:1 |
Three findings fall straight out of it. Instagram Feed, the grid surface, recommends 2:3 for video and reserves 9:16 for a tick. Threads supports all six ratios and recommends none of them. Facebook in-stream video, universally repeated as 16:9, is recommended at 4:5, with 1:1 recommended for stills instead. One company, six ratios, four surfaces deep, and no single number to carry away. Every listicle compresses this into "Instagram is 9:16" because a listicle has one row per company and this needs one row per placement.
One caveat that has to ride along with the table: these are advertising specifications, drawn from Ads Manager. YouTube's ratios above are player and upload specifications, and X's come from an ads product page. An ad spec is not a posting spec, and a ratio you read in an ads guide does not automatically become the requirement for an organic upload.
Ratio and resolution drift apart in Meta's own numbers
The same company, two documents, two different resolutions for the same vertical surface. Meta's Instagram Reels awareness ad specs give a file type of MP4 or MOV, a ratio of 9:16, and a resolution of "1440 x 2560 pixels". Every ratio table says 1080 × 1920 for Reels. Meta's own design recommendations say 1440 × 2560.
Neither is a typo and neither is wrong, and the arithmetic settles it in one line. Divide the first pair by 160 and the second by 120: 1440 ÷ 9 = 160 and 2560 ÷ 16 = 160, while 1080 ÷ 9 = 120 and 1920 ÷ 16 = 120. Same shape, more pixels in one and fewer in the other. This is the aspect ratio pillar argument in miniature: the ratio and the resolution are separate quantities, and a vendor document can agree with you on one while drifting from you on the other.
Drift shows up inside Meta too. The Instagram Stories design requirements recommend "1080 x 1920" with a "Minimum: 600 x 1067" for the same 9:16 recommendation, a pair that divides to 0.5623 rather than 0.5625, a near miss on Meta's own stated shape. Worth one sentence rather than a section, because nothing breaks for the uploader.
That Stories page also varies its advice by objective: 9:16 for a normal ad, but 1.91:1 for Meta Advantage+ catalog ads. Same surface, two recommendations, selected by campaign type, which a one-row-per-platform table cannot hold.
What a feed does to a mismatched file
Strip out the guesses and six behaviours are actually documented. YouTube adds themed padding: for "9:16 vertical videos on computer browsers", "the padding is white by default, and dark gray when Dark theme is turned on." YouTube's player also resizes itself to the file, and advises against baking your own bars in: "avoid adding padding or black bars directly to your video", because the padding "interferes with YouTube's ability to change the player dynamically to your video size and the viewer's device". Facebook does something similar on the desktop: in Facebook Feed, "the video will be shown with a 1:1 aspect ratio, which may cause black bars to appear on either side of the video."
Then there are overlays. X asks you to "be mindful of the various overlays in the Vertical Video viewer", noting that the overlay "disappears after two seconds of your video is watched", so the first two seconds are the ones to compose around. Meta frames the same problem as one of design, warning about elements "cropped, covered by the profile icon or call to action". And TikTok warns that "there might be slight differences between the preview and the live version since the ad preview is not specific to a device", which is a useful thing to know before you trust a preview as proof of anything.
The widely repeated belief is that a feed silently reframes a landscape upload into vertical and keeps the important parts in shot. Nothing in the vendor documents read for this post says that. Not the YouTube pages, not the Meta matrix, not TikTok's in-feed specifications. The belief is widespread and the documentation is silent about the mechanism. Silence is not proof either way, but it does mean no vendor statement can be quoted to justify re-exporting one landscape cut for every platform. What is documented is padding, overlays and dynamic resizing, and all three move your framing without asking.
If you would rather decide the crop yourself than hand it to a viewer, that is a rendering decision, and CSS is where it belongs. aspect-ratio and object-fit covers the property that fixes the box and the one that decides what is left of the picture inside it.
The safe-margin number nobody can cite
Here is the sharpest line in this post. No vendor document in this pass publishes a universal "keep your text this many pixels from the edge" figure, because the number is not universal, and the one that circulates anyway is unsourced.
Meta publishes numbers, on one surface, in percentages: "at least 14% of the top, 35% of the bottom and 6% on each side" of an Instagram Reels ad should be left free of text, logos or important creative elements. Read the reason clause and it explains the whole problem: the margins exist so key parts are not "placed too close to the edges of devices with screens taller than a 9:16 aspect ratio". A 9:16 asset on a taller phone loses its edges to the interface. It is scoped to Reels ads, not to Instagram as a whole.
TikTok publishes no numbers at all. Its safe zone "is determined by the dimension (vertical, horizontal, or square), ad caption length, and any additional formats used", and what it ships instead is four downloadable ZIP templates: left-to-right standard, left-to-right with anchor, right-to-left Arabic, right-to-left with anchor. The caption length alone is enough to make a single fixed figure impossible.
X confirms the overlays exist and points at an annotated screenshot for the measurements, which are pixels inside a raster rather than numbers in prose. YouTube offers padding guidance with no percentages at all. One company gives you arithmetic you can check, two give you a template file or a diagram, and none of them gives you the round number the listicles keep quoting.
Run the arithmetic against your own number
The aspect ratio calculator does the arithmetic and nothing else. That is a deliberate limit, not a missing feature: a grep of the module that drives it turns up no platform names and no preset table. Its orientation vocabulary is portrait, landscape and square, derived by comparing the two ratio numbers, and its resolution ladder is anchored on width (640 up to 7680) because the familiar 720p and 1080p labels are height conventions. Output names, not platform names.
The live tool page hedges its own numbers as carefully as this post does, telling readers to treat them as collected guidance rather than platform law because specifications drift. That hedge is now demonstrably correct: the page's vertical short-form figure is 1080 × 1920, and Meta's own current recommendation for the same surface is 1440 × 2560.
So the workflow is: read your platform's page, take the ratio it states, then run the arithmetic against a pixel dimension you actually have. Once the numbers are settled the rest is separate work: resizing to the pair you computed, then cutting file weight, where the published ceilings are real ceilings rather than folklore: 4 GB for an Instagram Reels ad, 500 MB for a TikTok in-feed ad.
One honesty note about scope. This post rests on five vendor documents across four companies, four surfaces deep on Meta; nine platforms or surfaces were not obtained, among them Pinterest, Snapchat, Discord, Slack, LinkedIn, TikTok's own consumer upload flow and Facebook's organic placements. Nothing here fills those gaps from memory, and no search-volume figure appears because demand for these terms was not measured for this piece.
Related Tools & Further Reading
- Aspect Ratio Calculator — the tool this post is about: give it a ratio and one pixel dimension and it returns the other, with a tested ladder.
- Image Resizer — you have the ratio and now the pixel pair; changing the dimensions is the separate operation.
- How to Compress Images for Web — the file-weight lever to pull before anything gets uploaded.
- Reduce Image File Size Without Losing Quality — what to do when the file meets a published ceiling but should not be near it.
- The Complete Image Tools Guide — the whole toolkit, when you are building a pipeline rather than making one decision.

Written by
Sunil Kapoor
My attention goes to the two decisions that are made once and then applied to a hundred files. Aspect ratio is the first. A resize that forces a source image into a fixed frame either crops or distorts, and both are visible. The choice of which depends on the destination: a profile photo wants a centre crop, a banner wants letterboxing or a regenerated design, and a product photo wants neither.
Deciding once and applying it to everything is how a batch job produces two hundred files with the subject cut off. Upscaling is the second. A larger image cannot be recovered from a smaller one, so enlarging produces a file with more pixels and no more detail.
It can still be the right choice for a fixed layout, and I cover when it is acceptable versus when it is a way of pretending a missing asset exists. Resampling method matters more at large reductions than people expect. Averaging and similar methods avoid the aliasing that nearest-neighbour produces, and aliasing shows up as jagged edges that look like a sharpening problem and are not.
Sharpening should come after resizing, never before, because sharpening a small image amplifies the very artefacts resizing introduces. Colour profiles need deciding once too. Resampling in the wrong space changes the appearance of the result, so a profile-aware resize and a naive one produce visibly different images from the same source. I end every page on output naming, because a batch that overwrites its own source has no undo.