Resizing an image and reframing it are different operations, and only one of them loses anything from the picture. An image scaled from 2000 pixels square to 1080 pixels square contains exactly what it did before. The same image cropped from square to portrait does not.
Almost every problem with images on social platforms comes from confusing the two. The dimensions are a detail; the ratio is the decision.
Which ratios are actually in use?
A short list covers nearly everything.
| Ratio | Shape | Typical use |
|---|---|---|
| 1:1 | square | feed posts, profile crops |
| 4:5 | tall portrait | the tallest most feeds allow |
| 9:16 | full portrait | stories, reels, short video |
| 16:9 | landscape | video, thumbnails, headers |
| 1.91:1 | wide landscape | link previews |
Portrait fills more of a phone screen than landscape, which is why feeds have moved towards 4:5 and why vertical video took over. A landscape image in a vertical feed occupies a third of the screen a portrait one would.
The last row is the odd one, and it is not arbitrary — it is close to the golden ratio, and it became the link preview shape by convention rather than by rule.
What is the difference between scaling and cropping?
Scaling changes the pixel count and keeps the framing. Cropping changes the framing and keeps the scale. A tool asked to fit an image into a different ratio has to do one of three things, and which one it picks changes the result completely.
It can crop, discarding whatever falls outside the new shape. It can letterbox, adding bars to preserve the whole image. Or it can stretch, which distorts everything and is never the right answer despite being the easiest to implement.
The reason this matters is that platforms crop automatically and do not ask. Uploading a landscape photograph to a square slot removes the sides, and nothing warns you that the subject was in the part removed.
Which of the three a tool picks is usually named in its options. Fit keeps the whole image and adds padding, fill crops to the frame, and stretch ignores the ratio entirely — so choosing fill and choosing fit produce two genuinely different pictures from the same source, and neither is a default worth accepting without looking.
Why does my image look soft?
Because it was scaled up, or because it was compressed twice. Enlarging invents pixels that were never captured, and no amount of processing recovers detail that is not in the file.
Double compression is the more common cause and the less obvious one. An image saved as a JPEG, edited, and saved as a JPEG again carries the artefacts of both passes — and platforms compress on upload regardless, so a file that has already been compressed hard arrives worse.
The practical rule is to work from the largest original available and export once, at the size the platform wants or a little above it. Uploading something enormous does not help either, since it will be resized on arrival by an algorithm you do not control.
Does more resolution help?
Up to a point, then it stops. Supplying roughly twice the displayed dimensions covers high-density screens, and beyond that the extra pixels are discarded on upload.
Screen density is why the numbers rose in the first place. A display packing twice the pixels into the same physical area shows a 1x image softly, so assets doubled — and the doubling has largely stopped there, because the returns past it are invisible.
File size is the counterweight. A larger file takes longer to load, and on a feed scrolled at speed a slow image is one that is never seen. The right size is the smallest one that still looks sharp on a dense screen.
What about physical dimensions?
They only exist once something is printed or displayed at a known size. Pixels have no physical dimension of their own — an image is a grid, and how large that grid appears depends entirely on the device showing it.
The number that connects them is pixels per inch, and it matters for print and almost never for screens. A 300 PPI setting inside a file changes nothing about how a browser displays it.
Screen size and resolution are similarly independent. Two displays with the same pixel count can be quite different physical sizes, which is why the diagonal measurement and the resolution are always quoted together.
Questions people ask
Is a bigger upload always safer? No. Past the platform’s own maximum it is resized anyway, and the upload simply takes longer.
Which format should I export? Photographs as JPEG or a modern equivalent, flat graphics and text as PNG. A screenshot of text saved as JPEG is the classic mistake.
Does the platform re-encode my file? Nearly always. Assume one compression pass on top of yours.
Can I upload one image everywhere? Only if it is square and the subject is central. Anything else needs at least a crop per ratio.
Why did my text get cut off? The crop happened at a different ratio than the one you designed for. Keep text away from the edges.
Decide the ratio first, then the pixels. The aspect ratio calculator works out the second dimension from the first, the resolution calculator and screen dimensions calculator handle the display side, screen size and screen resolution cover the physical questions, the image size checker reports what a file actually is, and the image resizer does the scaling.