Video Aspect Ratio Calculator

Aspect ratio, resolution and pixel aspect ratio are three different measurements

They get used interchangeably and they are not interchangeable. Getting them apart is what stops a conversation about "making it 9:16" from producing a file nobody can use.

Aspect ratio is shape. It is width divided by height, written as two numbers with a colon (16:9) or as one decimal against 1 (1.78:1). It carries no information about size: 1280x720, 1920x1080 and 3840x2160 are all 16:9. Two files with the same ratio will play in the same frame without bars regardless of how different their pixel counts are.

Resolution is size. It is the literal count of pixels stored, written width x height. It determines detail and file weight, and it implies a ratio, but the implication only runs one way — a ratio never tells you a resolution. This is why "what resolution is 9:16?" has no single answer, and why the useful answer is a convention rather than a fact: 1080x1920, because that is what the platforms recommend and transcode toward.

Pixel aspect ratio is the shape of each individual pixel, and it is the one that produces files whose numbers appear to lie. Modern web video is square-pixel: one pixel is as wide as it is tall, so the stored dimensions and the displayed shape agree. Broadcast and tape formats were not. HDV stores 1440x1080 — arithmetically 4:3 — with a pixel aspect ratio of 4:3, and displays as 1920x1080 16:9. DV NTSC stores 720x480 with a pixel aspect ratio of 10:11 and displays as 4:3. FFmpeg names these separately: SAR is the sample (pixel) aspect ratio and DAR is the display aspect ratio, with DAR = (width ÷ height) × SAR. If a file measures one shape in a calculator and plays as another, a non-square pixel aspect ratio is almost always why.

The calculator above works in square pixels, which is correct for anything you would post today. When it reads dimensions from a file it takes them from the browser's decoder, which reports display dimensions — so an anamorphic file comes back already corrected.

Why 1080x1920 is the vertical standard

9:16 is the exact inverse of 16:9, which makes it the shape of a phone held upright — and the shape every full-screen mobile feed was built around. 1080x1920 is the resolution the industry settled on inside that shape, and the reason is not aesthetic.

It is where the platforms' own transcoders sit. TikTok, Instagram Reels, YouTube Shorts and Facebook Reels all name 1080x1920 as the recommended upload. Several accept more — YouTube will take 2160x3840 for a Short, X accepts up to 1920x1200 — but every one of them re-encodes on receipt, and uploading above the target resolution hands the decision about how to throw those pixels away to a stranger's encoder rather than yours.

There is a second reason, and it is the one people miss: 1080 x 1920 = 2,073,600 pixels, which is exactly the same count as 1920 x 1080. A vertical master is not a smaller file or a lower-quality one. It is the same pixel budget turned ninety degrees. What makes vertical exports look soft is almost never the canvas — it is that the picture inside it was cropped out of a landscape frame and then stretched to fill the canvas, which is a different problem with a different fix.

What converting 16:9 to 9:16 actually costs

Start from 1920x1080 and keep the full height. The widest 9:16 window that fits inside 1080px of height is 1080 × (9/16) = 607.5, rounded up to 608 pixels. So the crop window is 608 x 1080. That window is 31.7% of the original width: converting a landscape frame to full-screen vertical by cropping discards 68.3% of the picture you shot. Two thirds of the frame, gone permanently.

For comparison, the same 1920x1080 frame cropped to 4:5 keeps 864px — 45% of the width. Cropped to 1:1 it keeps 1080px — 56%. 9:16 is by a wide margin the most expensive conversion in ordinary use, which is why it deserves an editorial decision rather than an export preset.

The consequences are all about framing, and they are concrete:

• Two people cannot share the frame if they are more than 608px apart. A standard two-shot with subjects on the one-third and two-thirds lines of a 1920px frame puts them 640px apart. No crop position holds both. Centre it and you cut both at the shoulder; aim at one and the other leaves the video.
• Anything at the edges leaves first. Lower thirds, logos, a game HUD, the toolbar of a screen recording, the second half of a slide. In a 16:9 composition those regions are where information deliberately lives, because the centre was reserved for the subject.
• A static crop cannot follow a moving subject. If the presenter walks, or the edit cuts between two chairs, no single horizontal offset is right for the whole video. That is the subject-tracking problem, and it is why reframing is properly a per-shot decision rather than a per-file setting.
• Delivering a 608x1080 crop at 1080x1920 is a 1.78x upscale. It fills more of the phone screen with a less detailed image. That is often the right trade, but it is a trade — no interpolation invents detail the sensor never recorded.

When to pad or blur-pad instead of cropping

Padding fits the whole frame inside the target canvas and fills the rest. A 1920x1080 shot placed in a 1080x1920 canvas renders as a 1080x608 picture with 656px of empty canvas above and below it. Every source pixel survives — and your picture is 31.7% of what the viewer sees, which is the same 31.7% cropping was throwing away. The arithmetic does not offer a free option. It offers a choice about which loss you take.

Crop when the content is centre-weighted

One person talking. A single subject with clear space around them. Footage where the edges carry context rather than content. A vertical crop of a talking head is not a compromise — it is what vertical video is, and cropping is the conversion that respects the format.

Pad when the content runs to the edges

Screen recordings, software demos, gameplay with a HUD in three corners, wide establishing shots, charts, slides, two-shots. Crop these and you have not reframed them, you have truncated them, and the parts removed were the parts carrying information. Fitting the whole frame in looks like the lazier option and is usually the honest one.

Blur-pad when you want it to read full-bleed

Blurred padding fills the bars with a scaled, heavily blurred copy of your own frame instead of black. It preserves the whole picture like a plain pad, but the result reads as a full-screen video rather than a letterboxed one — which matters on platforms where black slabs signal "reposted from somewhere else". It has a practical benefit too: those 656px bands are somewhere captions, a title or a hook line can live without covering a single pixel of the footage.

There is a hybrid worth knowing. Fit the 16:9 shot into the top of a 9:16 canvas rather than the centre, and use the remaining 1312px for text. That is why so much interview and podcast content on vertical platforms looks that way. It is not a style — it is the arithmetic on this page, with the leftover space put to work.

Valmera reframes by all three methods — crop, pad or blurred pad — and you pick by describing the result you want. It never upscales: a 1280x720 source reframed to 9:16 renders 720x1280, not a stretched 1080x1920. See resize video for any platform for the conversion itself, or the reframing docs for how the crop window is placed.

Platform aspect ratios, resolutions and safe areas (2026)

The ratio gets your video into the right shape. The safe area is what is left after the app draws its own interface on top of it — username, caption, sound label, action rail, follow button. A composition can be perfect in 16:9 and broken in 9:16 without a single pixel of the picture changing, because what changed is the furniture sitting over it.

Two honest caveats. These pixel figures are measurements of shipping app interfaces, not published standards, so they move without notice and differ between devices, app versions, and a platform's organic feed versus its ad placements — treat every number here, including ours, as current and decaying. And they are floors, not targets: inside a 1080x1920 frame, a horizontally centred band 800px wide running from y≈260 to y≈1440 — 140px clear on each side, 260px clear at the top, 480px at the bottom — sits inside every layout in this table at once, which is what to design against if one master has to serve TikTok, Reels and Shorts.

Why the calculator rounds to even numbers

H.264 in 4:2:0 — the codec and chroma layout every platform wants — stores colour at half resolution on both axes, so each chroma sample covers a 2x2 block of luma pixels. A frame with an odd width or height contains no whole number of those blocks, and encoders refuse it outright. This is why 1080 × (9/16) = 607.5 becomes 608 rather than 607, and why an export can fail on a size that was arithmetically perfect. Rounding at the point you do the maths is cheaper than discovering it at the point you render.

How to Convert a Video to Another Aspect Ratio

1. 1
   Work out what the conversion costs

   Put your source dimensions into the calculator above and pick the target shape. The crop panel tells you how much of the picture you lose; the pad panel tells you how much of the output frame becomes bars. Decide which loss the footage can afford before you open an editor.
2. 2
   Upload the footage and describe the reframe

   In Valmera, upload the video and type the outcome — "reframe this to 9:16 for Reels", or "fit the whole screen recording with a blurred background". Choose crop when the content is already centre-weighted, and pad or blurred pad when it is not: the crop window is centred on the frame, and there is no automatic subject tracking to follow a presenter who moves.
3. 3
   Watch the preview and change the fill if the crop lost something

   The agent renders a preview and looks at the frames it produced. If the centred crop took out something that mattered, say so and switch — "use a blurred background instead so the whole screen stays visible" — and it re-renders with every pixel intact. Export when it is right: the final render comes from your original file at source quality, and Valmera never upscales.

The one-time index runs once per upload with visible progress; every reframe and correction after that is a chat message.

Ratio maths is the easy half

A calculator can tell you that a 9:16 crop of 1920x1080 is a 608px window. It cannot tell you whether that window can hold your shot, and that is the entire difficulty of reframing. If your subject sits in the middle and stays there, a centred crop is the whole answer. If they are off to one side, or they move, or the edit cuts between two speakers, no fixed window works and the honest conversion is a pad or a blurred pad that keeps the frame whole.

Valmera is explicit about which of those you are getting. Its crop is centred and it does not track a moving subject — it says so rather than claiming a smart reframe it cannot perform. What it does do is index the footage up front (word-level transcript, shot boundaries, labeled frame tiles it reads directly), render a preview, and look at the frames it produced, so you can see what the conversion cost and switch fill mode in one message. The full detail is in the aspect ratio glossary entry, and the tool itself is resize video. Everything else the agent can do is in the tools hub.

Frequently Asked Questions

Do the Conversion, Not Just the Arithmetic

Upload the footage, say which shape you need, and pick crop, pad or a blurred pad in plain English. 50 free credits on every account, no card.