A crop is a decision about what leaves

Ask the other question

When people crop, they ask what should be in the frame. The more useful question is what is being removed, because that is the decision that changes the picture's meaning.

A wide photograph of a vegetable market is about a market. Crop to two hands and a note passing between them and it is about a transaction. Crop to one vendor's face and it is about a person. The same file, three subjects, and the difference was made entirely by subtraction.

This reframing is not wordplay. It changes what you look at while cropping — from the subject, which you were already looking at, to the context, which is where the decision actually lives.

Three passes

A habit that produces better crops than adjusting one rectangle until it looks right:

1. Crop for the subject. The tightest frame that still contains what the picture is about.
2. Crop for the context. The widest frame that adds information rather than clutter.
3. Choose between them deliberately, and be able to say what the middle option gains or loses against each.

Most weak crops are the accidental middle: too tight to explain anything, too loose to be about anything.

What a crop cannot do

Cropping changes framing. It does not change perspective.

Perspective is fixed at the moment of capture by where the camera was. A portrait taken from twenty centimetres away has a distorted nose because of the distance, and cropping the same file cannot produce the portrait you would have got from two metres away with a longer lens. Likewise a building photographed from below has converging verticals, and cropping does not straighten them — that needs a perspective correction, which is available free in GIMP's Perspective tool and in Krita, and which costs resolution and introduces stretching.

So when a photograph is wrong because the camera was in the wrong place, cropping is not the repair. Reshoot if you can, and if you cannot, choose a crop that makes the distortion look intentional rather than one that half-hides it.

The resolution budget

Cropping throws pixels away permanently, and it is worth being able to do the arithmetic rather than discovering the problem at the printer.

For print, the working figure is 300 pixels per inch. So:

print size in inches = pixels ÷ 300

A 12-megapixel phone photograph is roughly 4000 × 3000 pixels, which prints at 13.3 × 10 inches. Crop to a quarter of the area and you have 2000 × 1500, which prints at 6.7 × 5 inches. Crop to a tenth of the area and you are at roughly 1265 × 950, which is a 4-inch print and nothing larger.

For screens the budget is far looser. A full-width hero image on a phone needs around 1200 pixels wide to look sharp on a high-density display; on a desktop, around 2400. A crop that is useless for a poster is often perfectly adequate for a web page, which is why the same asset library can serve one and not the other.

Upscaling, including with machine-learning upscalers, invents detail that was not recorded. It can look good and it is not the same as having the pixels. For anything where accuracy matters — a product, a document, a face that somebody will recognise — treat invented detail as a risk rather than a solution.

Keep the original

Crop last, for a specific output, and never over the master file. This sounds obvious and it is the most commonly broken rule in practice, usually because somebody cropped in a phone gallery app that overwrites.

The workflow that survives:

• Keep originals untouched in one place.
• Produce crops as exports, named for their purpose — hero-16x9.jpg, card-4x5.jpg, thumb-1x1.jpg.
• In GIMP, Photopea or Krita, do the crop on a layer mask or as a canvas-size change in a saved working file, so the original pixels remain in the document.

You will be asked for a different crop. You will always be asked for a different crop.

The crop as a composition decision

One last point that connects this to the rest of the block. When you crop, you are not only choosing content — you are setting the aspect ratio, deciding where the subject falls relative to the edges, and creating the negative space. All three are composition decisions being made by the same rectangle.

That is why cropping well is a skill rather than a trim, and why a photograph that looked ordinary can become good with one decisive crop. What you removed was doing damage.