What an aspect ratio actually fixes
An aspect ratio is the proportion between width and height, written either as a pair of whole numbers (16:9) or as a single decimal (1.7778). It carries no information about size. A 640×360 thumbnail, a 1920×1080 broadcast frame and a 7680×4320 8K master are all 16:9; they differ only in how many pixels are spent filling that shape.
That separation is the whole point. When you resize an image or a video proportionally, you hold the ratio constant and let both dimensions move together. The moment the ratio changes, the picture stretches: circles become ovals, faces get wide or thin, and no amount of later correction undoes it. So every resize is really two decisions — what ratio do I keep, and how many pixels do I want along one side — and this calculator handles the arithmetic that connects them.
You reduce a ratio the same way you reduce a fraction: divide both numbers by their greatest common divisor. For 1920 and 1080 the gcd is 120, and 1920÷120 : 1080÷120 gives 16:9. For 3440 and 1440 the gcd is 80, giving 43:18 — which is why an ultrawide monitor advertised as “21:9” is not exactly 21:9 at all. The marketing number is a family label, not a measurement.
The international broadcast formats are defined by name. ITU-R BT.709 specifies the 16:9 HDTV image with 1920×1080 square-sampled pixels, and ITU-R BT.2020 carries the same 16:9 shape up to 3840×2160 and 7680×4320 for UHD. When a delivery spec says “BT.709 HD”, the aspect ratio is already decided for you.
The formula, and why it is a proportion rather than a subtraction
Hold the ratio R = w ÷ h constant and the new height follows directly from the new width: h₂ = w₂ ÷ R, or equivalently h₂ = w₂ × h₁ ÷ w₁. Going the other way, w₂ = h₂ × R.
People reach for subtraction here far more often than they should. If you take 200 px off a 1600×1200 image you get 1400×1000 only if you also take 150 px off the height, because the ratio is 4:3 and the deltas must be in the same 4:3 proportion. Subtracting the same number from both sides changes the shape. Scaling is multiplicative, and the multiplier — the scale factor — is the single number that describes the whole operation: s = w₂ ÷ w₁, applied identically to both axes.
The greatest common divisor does the other half of the work. gcd(w, h) is the largest integer that divides both dimensions, so dividing through by it produces the smallest whole-number pair with the same proportion. That reduced pair also tells you the step size for exact resizing: at 16:9, every whole-pixel size is a multiple of 16 across and 9 down, so 1280×720 works exactly while 1300×731.25 does not.
For fitting content into a frame, the rule is “contain” rather than “cover”: scale the content until it touches the frame on its tighter axis, which means W_content = min(W_frame, H_frame × R). Whatever frame space is left over becomes two equal bars. If the leftover is vertical you get letterboxing, bars above and below; if it is horizontal you get pillarboxing, bars at the sides. Only when the two ratios match exactly do you get no bars.
Worked example: a 1920×1080 export placed in a 1920×1200 frame
You have a 1920×1080 render. You need a 1280-pixel-wide version for a web page, and you also need to know what happens when the original is dropped into a 1920×1200 display area.
- Reduce the ratio. gcd(1920, 1080) = 120. So 1920÷120 = 16 and 1080÷120 = 9, giving 16:9.
- Convert to a decimal. R = 16 ÷ 9 = 1.777778.
- Solve the missing height. h₂ = 1280 ÷ 1.777778 = 720 px. Equivalently 1280 × 9 ÷ 16 = 720. Both dimensions are whole numbers because 1280 is a multiple of 16.
- Scale factor. 1280 ÷ 1920 = 0.66667, so the new file is 66.667% of the original on each axis, and 0.66667² = 44.4% of the original pixel count.
- Fit the original into the 1920×1200 frame. The largest 16:9 box that fits is min(1920, 1200 × 1.777778) = min(1920, 2133.33) = 1920 px wide.
- Its height in that frame. 1920 ÷ 1.777778 = 1080 px.
- Bar thickness. The frame is 1200 tall and the picture is 1080 tall, so 1200 − 1080 = 120 px of unused height, split equally: 60 px top and 60 px bottom. That is letterboxing, because the content is proportionally wider (1.7778) than the frame (1920÷1200 = 1.6667).
Reverse the case to see the other regime. Put a 4:3 image in the same 1920-wide frame at 1080 tall: the fitted box is min(1920, 1080 × 1.3333) = 1440 px wide, leaving 1920 − 1440 = 480 px of unused width, or 240 px of pillarbox on each side.
How to read the result
Start with the simplified ratio, because it tells you the step size for clean sizes. If the reduced pair is 16:9, any width that is a multiple of 16 gives a whole-pixel height. If the reduced pair is 239:100 — anamorphic scope — then only widths that are multiples of 239 are exact, which is why scope masters are almost always cropped inside a 1920 or 3840 frame rather than sized to the ratio directly.
Check the scale factor next. Below 100% you are downsampling, which is safe and usually improves apparent sharpness after a light sharpen pass. Above 100% you are enlarging, and no resampling filter invents detail that was never captured; treat any number well above 100% as a signal to go back to a higher-resolution source. Because area scales with the square of the linear factor, a 50% scale keeps only 25% of the pixels, and a 200% scale asks the resampler to make up three pixels for every real one.
Then read the bar figure. A bar of zero means the frame and the content share a ratio and the picture fills the frame. A non-zero bar tells you exactly how much of the frame you are paying for and not using — useful when a platform charges by canvas size, when you are designing a title-safe area, or when you need to know how many pixels of real picture survive a 9:16 crop of 16:9 footage. Cropping 1920×1080 to 9:16 leaves a 608×1080 slice (1080 × 9 ÷ 16 = 607.5, rounded up to an even 608), which is under a third of the original width.
Finally, sanity-check the parity. Every mainstream video codec using 4:2:0 chroma subsampling stores one chroma sample for each 2×2 block of luma, so both dimensions must be even. Encoders either pad to the next even size or refuse the input outright. Still images have no such restriction.
Common aspect ratios and the resolutions that match them
| Ratio | Decimal (w÷h) | Typical resolutions | Where you meet it |
|---|---|---|---|
| 16:9 | 1.7778 | 1280×720, 1920×1080, 3840×2160, 7680×4320 | HDTV and UHD per ITU-R BT.709 and BT.2020; YouTube |
| 4:3 | 1.3333 | 640×480, 1024×768, 1600×1200, 2048×1536 | Standard-definition video, older monitors, Micro Four Thirds stills |
| 3:2 | 1.5000 | 720×480, 1620×1080, 6000×4000 | 35 mm film frame and most DSLR/mirrorless sensors |
| 1:1 | 1.0000 | 512×512, 1080×1080, 2048×2048 | Square social posts, texture maps, avatars |
| 4:5 | 0.8000 | 864×1080, 1080×1350 | Portrait social feed images |
| 9:16 | 0.5625 | 720×1280, 1080×1920 | Vertical short-form video |
| 64:27 | 2.3704 | 2560×1080, 5120×2160 | Ultrawide monitors marketed as “21:9” |
| 43:18 | 2.3889 | 3440×1440, 5160×2160 | The other common “21:9” monitor size |
| 239:100 | 2.3900 | 1920×803, 3840×1607 | Anamorphic scope cinema delivery |
Decimal values are rounded to four places. Each listed resolution divides exactly by its reduced ratio except the scope rows, where the height is rounded to the nearest whole pixel.
Mistakes that break a proportional resize
- Rounding both dimensions independently. Round one side, then recompute the other from the ratio. Rounding each separately can shift the ratio by a fraction of a percent, which is enough to show as a visible squeeze on a long panorama.
- Confusing the display ratio with the storage ratio. Anamorphic and DV formats store non-square pixels, so 720×480 can display as 4:3 or 16:9 depending on the pixel aspect ratio flag. This calculator assumes square pixels, which is correct for every modern still image and for BT.709/BT.2020 video.
- Treating “21:9” as a real ratio. 2560×1080 is 64:27 and 3440×1440 is 43:18. Neither reduces to 21:9, and a layout built on 21:9 will be a few pixels out on both.
- Forgetting that area scales as the square. Halving both dimensions does not halve the file size budget, it quarters the pixel count. Pair this calculator with the image file size calculator before you commit to a resolution.
- Cropping when you meant to fit. “Cover” fills the frame and throws pixels away at the edges; “contain” fits the whole picture and adds bars. This calculator computes the contain case, so the bar figures tell you what a cover crop would discard.
- Producing odd pixel dimensions for video. 4:2:0 chroma subsampling requires even width and height; nudge to the nearest even number before encoding.
Where aspect ratio sits among the other display numbers
Aspect ratio is one of four numbers that together describe a picture: shape (the ratio), size in pixels (the resolution), physical size (the diagonal), and colour. Change any one and the others stay put. A 27-inch and a 32-inch monitor can both be 16:9 at 3840×2160 and differ only in how tightly the pixels are packed — that is pixel density, which the screen PPI calculator works out from resolution and diagonal.
For video, the ratio feeds directly into bitrate planning, because bits per pixel is the quality yardstick and pixel count is width × height. Once you have settled a resolution here, the video bitrate and file size calculator converts it into a data rate and a file size, and the data transfer time calculator tells you how long that file takes to upload.
On the design side, the frame you fit into is usually a CSS box, and the bar colour matters as much as the bar size: bars that sit behind text need to clear the accessibility threshold, which the WCAG contrast ratio calculator checks. If you are specifying that bar colour across a design system, the HEX, RGB and HSL converter gives you the same value in every notation your tools expect.
Reach for a different tool when the pixels are not square (consult the format's pixel-aspect-ratio flag rather than this calculator), when you need a crop rectangle positioned within a frame rather than just its size, or when the constraint is a file-size budget rather than a dimension — in that case start from the bitrate and work backwards to the resolution.
Key terms
- Aspect ratio
- The proportion of width to height, independent of size. Written as a reduced whole-number pair (16:9) or as a decimal (1.7778).
- Letterbox
- Horizontal bars above and below the picture, produced when the content is proportionally wider than the frame it is placed in.
- Pillarbox
- Vertical bars to the left and right, produced when the content is proportionally taller than the frame.
- Scale factor
- The single multiplier applied to both dimensions during a proportional resize. Area changes by the square of it.
- Pixel aspect ratio
- The shape of an individual pixel. Square (1:1) in every modern format; legacy DV and anamorphic formats use non-square pixels, so their storage ratio differs from their display ratio.