Is This Photo Out of Focus, or Is It Just Small?

You take a photo, it looks fine on the back of the camera, and a week later on a proper screen it is soft. Nothing changed. The photo was always soft — the screen you first judged it on was too small to show you.

This is the single most common way a photograph is lost, and it is entirely preventable with one check that takes a second.

Why a small screen lies

Focus is a question about detail at the pixel level. A phone screen showing a 24-megapixel image is displaying roughly one pixel in twenty. Nineteen out of twenty pixels — including every pixel where the softness lives — are being thrown away before the image reaches your eye.

So the rule is simple and unhelpful: you cannot judge focus at anything other than 100%. Zooming to 100% on a 24MP photo means inspecting about thirty separate screensful, one at a time, for every photo you shoot.

What focus is, as a number

An in-focus edge changes brightness abruptly across two or three pixels. A soft edge spreads that same change over eight or ten. Measure how sharply brightness changes across small patches of the image and you get a number that tracks focus directly, without needing a sharp copy to compare against.

That last part is the important one. Most quality measurements are full reference — they need the original to compare to. Focus does not. The evidence is in the one file you have.

The part that goes wrong

A naive version of this check fails on perfectly good photographs, and it fails in a way that sounds authoritative. Three cases in particular:

Shallow depth of field. A portrait shot at f/1.4 is mostly out of focus by design. Measure the whole frame and you will condemn a photograph whose subject is razor sharp.

A plain sky or a flat wall. There is no detail there to be sharp or soft. A featureless patch measures as soft because there is nothing in it — not because the lens missed.

Compression noise. Save that same flat sky as a JPEG and the compression adds fine structure that reads, to a naive measurement, as detail. The picture has not changed. The number has.

The way past all three is to measure many small patches and look at the spread between them rather than the average. A genuinely soft photo is soft everywhere. A portrait has sharp patches and soft ones. A flat gradient is the same in every patch — which is a different signature again, and the honest answer there is this check cannot read this image, not this image is soft.

A quality check that says nothing when it cannot see is worth more than one that always produces a verdict.

What to do when a photo reads soft

• Check the shutter speed first. Below roughly 1/focal-length, handheld, softness is motion, not focus. The shutter speed is in the file’s EXIF and it is usually the whole answer.
• Check the ISO second. High ISO adds noise, and heavy noise reduction removes fine detail along with it. A soft, clean, ISO-6400 photo is a denoiser’s work, not a focus miss.
• Sharpening does not undo it. Sharpening raises local contrast at edges that already exist. It cannot recover an edge that was never recorded.

Check one photo

Upload a single image and DiffALL reports focus alongside noise, compression and exposure — each with its own verdict and the measurement behind it, and each free to decline when the image gives it nothing to read. There is no single overall score, deliberately: a photo that is perfectly exposed, clean, and completely out of focus is not “75% good”, it is unusable.

If you have both a before and an after — an original and an export, a raw and a delivery file — comparing the two answers a different and often more useful question: not is this sharp, but what did that export cost me.