Image Ratio Calculator

What an aspect ratio measures

An aspect ratio is width divided by height, written as two numbers with a colon between them. A 1920 by 1080 photo has a ratio of 1920 to 1080, and both numbers divide by 120, so the reduced form is 16:9. The decimal form is the same idea with the division done: 1920 / 1080 = 1.778. The W3C definition of the aspect-ratio property uses that same division, width over height, which is why a ratio below 1 describes a portrait shape.

Two images can share a ratio and have nothing else in common. A 640 by 360 thumbnail and a 3840 by 2160 television frame are both 16:9. The ratio describes shape only. It says nothing about sharpness, file size, or how much detail sits inside the frame, which is why a tiny image can be stretched to the right shape and still look soft.

Reducing the ratio is the first step in any resize, because the reduced form is the shape you have to keep. Divide both sides by their greatest common divisor. For 1920 and 1080 that number is 120: 1920 / 120 = 16 and 1080 / 120 = 9. For a phone photo at 3024 by 4032 the divisor is 1008, so 3024 / 1008 = 3 and 4032 / 1008 = 4, which gives 3:4.

Ratios above 1 are landscape, ratios below 1 are portrait, and exactly 1 is square. That is the whole vocabulary. A 9:16 vertical video has a decimal value of 0.5625, the same number you get when you divide 1080 by 1920, which is not a coincidence: the two shapes are the same rectangle turned on its side.

Scaling multiplies both sides by one factor

To resize an image and keep its shape, multiply the width and the height by the same number. Call that number the scale factor. A factor of 0.5 turns 1920 by 1080 into 960 by 540. A factor of 2 turns it into 3840 by 2160. Because both sides move together, the ratio cannot change.

Changing one side alone is what produces a stretched picture. Set the width of a 16:9 photo to 1080 and leave the height at 1080 and the shape becomes 1:1, with everything in the frame widened by 77.8 percent. Faces get rounder, circles become ovals, and text looks like it was printed on a rubber sheet.

The factor can be found from either side, whichever one you know. Going from 1920 wide to 1080 wide, the factor is 1080 / 1920 = 0.5625. Check it against the height: 1080 × 0.5625 = 607.5. Going the other way, from 1080 wide up to 1920 wide, the factor is 1920 / 1080 = 1.7778, and 607.5 × 1.7778 = 1080.

A factor below 1 shrinks the image, a factor above 1 enlarges it, and a factor of exactly 1 changes nothing but the file name. When you know the factor, you can answer any resize question with one multiplication, including questions about a size you have not measured yet.

Worked example: 1920 by 1080 down to 1080 wide

Start with a 1920 by 1080 photo and a target width of 1080 pixels.

• Factor: 1080 / 1920 = 0.5625
• Height: 1080 × 0.5625 = 607.5 pixels
• Rounded: 1080 by 608 pixels

607.5 is not a whole number of pixels, so some rounding has to happen. Rounding up to 608 makes the file 1080 by 608, a decimal ratio of 1080 / 608 = 1.7763 instead of 1.7778. That is a difference of 0.08 percent, which no viewer will see, but the new file is no longer exactly 16:9. A tool that rounds down to 607 is off by the same small amount in the other direction. Neither answer is wrong. What matters is that the height came from the factor instead of from a guess.

Upscaling runs on the same arithmetic. Going from 1280 by 720 to 2560 wide uses a factor of 2560 / 1280 = 2, so the height is 720 × 2 = 1440 and the new frame is 2560 by 1440. The pixel count goes up by four times, because both dimensions doubled: 2560 × 1440 = 3,686,400 pixels against 1280 × 720 = 921,600 before. The shape is identical, but the file has four times as many pixels to store and four times as much data to push through a network.

Every row keeps the 16:9 shape, and the pixel count grows with the square of the factor. Doubling the width multiplies the pixel count by 4, and tripling it multiplies the count by 9.

Fit and fill give two different answers

A frame and an image rarely share a ratio, so the software has to pick a strategy. Fit, also called contain or letterbox, scales the image until all of it sits inside the frame. Fill, also called cover, scales until the frame is covered and then trims whatever hangs over the edge. Fit uses the smaller of the two possible factors. Fill uses the larger one.

Put the 1920 by 1080 photo into a 1080 by 1080 square. The width factor is 1080 / 1920 = 0.5625 and the height factor is 1080 / 1080 = 1.0. Fit takes 0.5625, so the image lands at 1080 by 607.5 and the leftover 472.5 pixels of height become bars, 236.25 at the top and 236.25 at the bottom. Fill takes 1.0, so the image stays 1920 by 1080 and 840 pixels of width are cropped away, 420 from each side.

Both results keep the 16:9 shape. The difference is what you give up. Fit gives up frame area to empty space. Fill gives up image area to the crop, and the crop is centered by default in most editors, which can cut the top of a head off if the subject sits high in the frame.

The same test on a phone photo of 3024 by 4032 going into a 1080 by 1350 frame gives factors of 0.3571 and 0.3348. Fit uses 0.3348 and lands at 1012.5 by 1350 with 67.5 pixels of side bars. Fill uses 0.3571 and lands at 1080 by 1440, so 90 pixels of height are trimmed.

The third row is the case worth aiming for. When the frame and the image share a ratio, fit and fill produce the same picture and there is nothing to give up, which is why building a composition at the final frame ratio saves a decision later.

Common ratios and the height at 1080 wide

Most work happens at a handful of shapes. The table converts each one to a decimal and shows the height that goes with a 1080 pixel width, which is the width most social uploads settle on.

Read the decimal column as a quick test. If the decimal is above 1, the shape is wider than it is tall. If it is below 1, the shape is taller than it is wide. Two shapes with the same decimal are the same rectangle at different sizes, no matter how the ratio is written.

What the big platforms ask for

Platform specs are not standards, and they change. The safe habit is to keep the master file at the largest size you have and export a copy per platform rather than resizing the same file over and over, because each save with compression loses a little quality.

YouTube's help page for custom thumbnails asks for a 3840 by 2160 pixel image for regular videos, which is 16:9, with a minimum width of 640 pixels. For Shorts it asks for 2160 by 3840, which is 9:16, with a minimum height of 640 pixels. Those two numbers are the same rectangle turned 90 degrees, and the ratio is the part that matters: a 1920 by 1080 file fills the same shape at half the resolution.

Instagram feed posts work in a narrow band around square. The two shapes people upload most are 4:5 portrait at 1080 by 1350 and 1.91:1 landscape at 1080 by 566. Anything much taller or wider gets cropped by the app, and the crop is not something you control after upload. Stories and Reels use the full phone screen at 9:16, which is 1080 by 1920.

The practical rule for a set of posts: build one 1080 by 1350 master, then export 1080 by 1080 from the center for square placements and 1080 by 1920 from the same master for stories. All three keep the same subject in frame because they come from one composition, and each one is a fit or fill decision you make once instead of a stretch you fix later.

Rounding, print sizes and the crop nobody asked for

Print adds a second number to the problem: dots per inch. A file has pixels, paper has inches, and the pixels per inch figure connects them. At 300 pixels per inch, a 1080 pixel width prints 1080 / 300 = 3.6 inches wide. A 1920 by 1080 file at the same density prints 6.4 by 3.6 inches, which is a 16:9 print that no standard frame is built for.

Photo prints come in 4 by 6, 5 by 7 and 8 by 10 inches, and none of those is 16:9. A 3:2 camera file fits 4 by 6 with no crop. An 8 by 10 is 5:4, so a 3:2 file has to lose about 17 percent of its width to fill the paper. That loss is a fill decision made by the print lab unless you make it yourself first.

Rounding is the other quiet source of trouble. Every resize that produces a fractional pixel count has to pick a whole number, and the ratio drifts by a fraction of a percent each time. Resize a file five times in a row and the drift compounds, along with the compression loss from five saves. Resize once from the original and the drift happens once.

This page does the arithmetic, not the image editing. The ratios above are geometry, while platform limits are published by each company and can change without notice, so check the current help page before a paid campaign.