Why your photo is sideways

A photograph taken with a phone held vertically is usually stored horizontally, with a tag in the file saying which way up it should be displayed. Software that reads the tag shows it correctly; software that ignores it shows it on its side.

That is the whole explanation for the picture that looks right in one app and sideways in another, and for the one that was fine until it was uploaded somewhere.

Why store it the wrong way round?

Because the sensor has a fixed orientation and rotating the data costs time and quality. Recording what the sensor saw and noting how the camera was held is faster, and it avoids re-encoding the image at the moment of capture.

The tag has eight possible values covering the four rotations and their mirrored versions. Most software handles the common ones and some handles only the simplest, which is why the failure is intermittent rather than universal.

Stripping metadata for privacy removes the tag along with the location and the camera model, which is the usual reason a photo that was fine becomes sideways after being processed.

What is the permanent fix?

Rotating the pixels rather than the tag. An image whose data is already the right way up needs no tag and cannot be displayed wrongly, which makes it the right form for anything being published or shared.

The cost is a re-encode. Rotating a compressed photograph and saving it again applies another round of compression, so the result is very slightly worse than the original.

There is a lossless rotation for quarter turns that rearranges the compressed blocks without decoding them, and it works only when the dimensions divide evenly into those blocks. Where it applies it is free; where it does not, most tools fall back to a re-encode.

Does cropping lose quality?

Not the pixels that remain, but saving does. Removing part of an image discards those pixels exactly and keeps the rest untouched — and if the result is written back as a compressed format, everything that survives goes through compression again.

The practical rule is to make all the edits in one pass and save once. Crop, rotate, resize and border in a single sitting produces one re-encode; doing them in four sessions produces four.

Why does a rounded image need transparency?

Because the corners have to be nothing rather than a colour. Rounding is a mask, and the area outside the mask must be transparent for the shape to sit on any background.

That rules out the common photographic format, which has no transparency at all — saving a rounded image in it fills the corners with white, which looks correct on a white page and wrong everywhere else.

The same applies to any shaped crop. If the result is not rectangular, the format has to support an alpha channel, and the file will be larger for it.

What is the difference between padding and cropping to square?

Whether anything is lost. Padding adds bands to the short sides and keeps the whole image; cropping removes the long sides and keeps the scale. Both produce a square and only one of them still contains the original picture.

For a profile picture or a product grid, cropping is usually right because the frame is meant to be filled. For a photograph where the composition matters, padding preserves what was shot at the cost of a smaller subject.

A border is a third option and a different intent. It adds a deliberate frame rather than compensating for a shape mismatch, and it changes the dimensions, so it belongs after any size that has to be exact.

What order should the edits go in?

Crop first, then rotate, then resize, then border. Cropping first means everything after it works on less data; bordering last means the border is the size you asked for rather than being scaled with the image.

Resizing before cropping is the sequence that wastes quality, because the crop then discards pixels that were already reduced. Doing it the other way keeps the maximum detail in the part you are actually keeping.

Anything that has to hit an exact dimension belongs at the end of the chain, since every earlier operation changes the numbers underneath it.

Questions people ask

Is a mirrored photo the same as a rotated one? No. Flipping produces an image that cannot be reached by any rotation, which is why text in a flipped photo reads backwards.

Why is my crop slightly off? A displayed image is usually scaled, so a selection made on screen maps back to the original at a different resolution. Working from the reported pixel values avoids it.

Does flipping an image change the file size? Barely. It rearranges pixels rather than adding them.

Can I undo a rotation? Rotating back restores the orientation and not the quality, since each save recompresses.

Do these tools upload my photo? No. Image editing here happens in the browser on the file you chose.

Why did my transparent image get a black background? It was saved to a format without transparency. Some of them fill with black rather than white.

Edit once, then save. The image cropper and square image converter handle the framing, rotate and flip fix the orientation permanently, make image round applies the mask that needs transparency, and add border to image puts the frame on last.