Every image that arrives stretched, squashed, or awkwardly cropped comes from the same tiny mistake: someone changed one dimension and let the other stay where it was. The whole subject reduces to one rule — aspect ratio is width divided by height, and if you change one number you must multiply the other by the same factor. Everything below is that rule applied to the four jobs people actually have: naming a ratio, resizing, fitting a frame, and cropping to a new shape.
A single photo may now need to appear as a square, a 4:5 portrait, a 9:16 vertical, and a 16:9 thumbnail — and each of those is a different shape, not a different size. Knowing which of the two operations you are doing, scaling or reshaping, is what stops the results looking wrong.
What an aspect ratio actually is
An aspect ratio describes shape, not size. It is the width compared to the height, written either as two numbers with a colon (16:9) or as a single decimal (1.7778).
The decimal form is just the division:
ratio = width ÷ height
A 1920 × 1080 frame gives 1920 ÷ 1080 = 1.7778, rounded to 4 decimal places. A 3840 × 2160 frame gives 3840 ÷ 2160 = 1.7778 as well. Different sizes, identical shape — which is exactly why a 4K video fills a 1080p screen edge to edge with nothing left over.
Two pixel dimensions have the same aspect ratio if, and only if, their decimals match. That single check settles most "will this fit?" questions in about three seconds.
How to reduce pixels to a named ratio
To get from 1920 × 1080 to the name 16:9, divide both numbers by their greatest common divisor (GCD) — the largest whole number that divides both exactly.
The reliable way to find it is Euclid's method: divide the larger by the smaller, keep the remainder, repeat until the remainder is zero. The last non-zero remainder is the GCD.
For 1920 and 1080:
- 1920 ÷ 1080 leaves remainder 840
- 1080 ÷ 840 leaves remainder 240
- 840 ÷ 240 leaves remainder 120
- 240 ÷ 120 leaves remainder 0
So GCD = 120. Then 1920 ÷ 120 = 16 and 1080 ÷ 120 = 9. The ratio is 16:9.
Try it on a photo straight out of a camera at 4000 × 3000. GCD(4000, 3000) = 1000, so the ratio is 4000 ÷ 1000 : 3000 ÷ 1000 = 4:3.
The ratios worth recognising on sight
| Pixel size | True ratio | Decimal | Where you meet it |
|---|---|---|---|
| 1920 × 1080 | 16:9 | 1.7778 | HD video, most TVs and monitors |
| 3840 × 2160 | 16:9 | 1.7778 | UHD "4K" — same shape as 1080p |
| 4096 × 2160 | 256:135 | 1.8963 | DCI 4K — slightly wider than UHD |
| 4000 × 3000 | 4:3 | 1.3333 | Many phone and compact-camera photos |
| 6000 × 4000 | 3:2 | 1.5000 | Most interchangeable-lens cameras |
| 1080 × 1080 | 1:1 | 1.0000 | Square posts |
| 1080 × 1350 | 4:5 | 0.8000 | Portrait posts |
| 1080 × 1920 | 9:16 | 0.5625 | Vertical video and stories |
| 2560 × 1080 | 64:27 | 2.3704 | "21:9" ultrawide monitor |
| 3440 × 1440 | 43:18 | 2.3889 | "21:9" ultrawide monitor |
Two honest notes from that table. "4K" is not one shape: UHD (3840 × 2160) is 16:9, while DCI 4K (4096 × 2160) is wider, so footage from the two sometimes needs a thin crop to match. And "21:9" is a marketing label, not a reduced ratio — 2560 × 1080 truly reduces to 64:27 and 3440 × 1440 to 43:18, which is why two "21:9" monitors are not quite the same shape.
Formula 1: resizing without distortion
This is the everyday one. You know the new width and need the height that keeps the shape:
new height = new width × (original height ÷ original width)
Or, the same thing from the other direction:
new width = new height × (original width ÷ original height)
Worked example. A 4000 × 3000 photo needs to be 1200 px wide.
- original height ÷ original width = 3000 ÷ 4000 = 0.75
- new height = 1200 × 0.75 = 900
The result is 1200 × 900, still exactly 4:3. Check it: 1200 ÷ 900 = 1.3333, matching 4 ÷ 3 = 1.3333.
There is an equally valid shortcut. Work out the scale factor once and apply it to both numbers: 1200 ÷ 4000 = 0.30, so the height is 3000 × 0.30 = 900. Same answer, and it makes the idea obvious — resizing multiplies both sides by one number.
Note what that does to pixel count. 4000 × 3000 = 12,000,000 pixels, or 12 MP. At 1200 × 900 you have 1,080,000 — about 1.08 MP, roughly a ninth, because area scales with the square of the scale factor (0.30² = 0.09).
Formula 2: fitting a shape into a different shape
When the source and the destination are different shapes, you must choose. There are only two options, and they are opposites:
- Fit (letterbox/pillarbox) — keep the whole image, add blank bars. Scale by the smaller of the two ratios
frame width ÷ image widthandframe height ÷ image height. - Fill (crop) — cover the whole frame, lose the overflow. Scale by the larger of the two.
Worked example, fitting. A 6000 × 4000 photo (3:2) into a 1080 × 1080 square, losing nothing.
- 1080 ÷ 6000 = 0.18; 1080 ÷ 4000 = 0.27
- take the smaller, 0.18 → 6000 × 0.18 = 1080 wide, 4000 × 0.18 = 720 tall
- leftover height = 1080 − 720 = 360, so bars of 360 ÷ 2 = 180 px top and bottom
Worked example, filling. Same photo, same square, but the square must be covered.
- take the larger, 0.27 → 6000 × 0.27 = 1620 wide, 4000 × 0.27 = 1080 tall
- overflow = 1620 − 1080 = 540, so you cut 270 px from each side if you centre it
Same photo, same target, two completely different outcomes. Deciding before you drag a handle is the entire skill.
The cropping cost nobody calculates first
Reshaping a horizontal frame into a vertical one is the most expensive crop there is, and it is worth seeing the number.
Take a 1920 × 1080 frame (16:9) and crop it to 9:16 while keeping the full 1080 height. The width you can keep is:
1080 × 9 ÷ 16 = 607.5 px
Out of 1920. That is 607.5 ÷ 1920 = 0.3164, so you keep about 31.6% of the original width and discard about 68.4%. Anything outside that narrow central strip is gone — which is the honest argument for keeping the subject in the middle third whenever you know a vertical version is coming.
Three traps that produce squashed images
Rounding one side and not the other. Round only the final pixel values, never the ratio you are multiplying by. Using a rounded 1.78 instead of 1.7778 pushes a 1080-tall frame to 1922.4 rather than 1920 — small, but enough to leave a one-pixel seam.
Half pixels. In the letterbox example above, 1080 × 9 ÷ 16 = 607.5, and pixels do not come in halves. Round to a whole number, and for video prefer an even number: many common encoders require even width and height and will refuse or silently adjust an odd dimension.
Assuming a name means a shape. "1440p" describes a height, not a ratio: 2560 × 1440 is 16:9, while 1440 × 1080 reduces to 4:3. And a 4:3 photo printed at 6 × 4 inches is not 4:3 at all — 6 ÷ 4 = 1.5, which is 3:2, which is why standard prints clip the long edge. Reduce the actual numbers rather than trusting the label, the same discipline that keeps unit conversions honest.
FAQ
How do I calculate aspect ratio from pixel dimensions?
Divide width by height for the decimal form, then divide both numbers by their greatest common divisor for the named form. 1920 × 1080 gives 1.7778 and, after dividing both by 120, 16:9.
What height do I need for a given width?
Multiply the new width by the original height divided by the original width. For a 4000 × 3000 image at 1200 px wide: 1200 × (3000 ÷ 4000) = 900 px.
Is 4K the same shape as 1080p?
UHD 4K (3840 × 2160) is, because both reduce to 16:9. DCI 4K (4096 × 2160) is not — it reduces to 256:135, slightly wider, so it needs a small crop or thin bars to match 16:9.
Should I letterbox or crop?
Letterbox when losing content is unacceptable — diagrams, full-body shots, anything with edge detail. Crop when filling the frame matters more than the edges, and check first how much you are sacrificing: reshaping 16:9 to 9:16 discards roughly 68% of the width.
Why does my photo print with the edges cut off?
Because the print size is a different ratio from the file. A 4:3 photo (1.3333) on 6 × 4 inch paper (1.5) has to be cropped along its long edge to fill the sheet. Compare the two decimals before ordering and either accept the crop or pick paper closer to the file's shape.
Get the number without the arithmetic
Reduce with the GCD, scale both sides by the same factor, and decide between fit and fill before you touch a handle — simple maths that punishes one careless step, exactly like the errors in our guide to everyday calculations.
When you would rather not do it by hand, the free aspect-ratio calculator on Wapuula Tools reduces any pixel pair to its true ratio and returns the matching width or height, with the formula shown beside the result — so you are checking the number rather than trusting it. It sits alongside the rest of the everyday calculators worth bookmarking.
