There are four kinds of blur and only two of them are recoverable. Identifying which one you have takes thirty seconds and saves an hour of pointless sharpening.
The short answer: identify which of four blurs you have before doing anything. Lens softness and mild camera shake are genuinely recoverable with deconvolution. Badly missed focus and severe motion blur are not — AI will produce something sharp, but it will be a plausible reconstruction rather than your subject. Thirty seconds of looking saves an hour of wrong sharpening.
"Can you fix this, it is a bit blurry" is one of the most common requests a retoucher gets, and the honest answer depends entirely on which of four different things has happened. They look similar in a thumbnail and are completely different problems.
Identify the blur first
Open the file at 100% and look at a high-contrast edge.
Missed focus. Something in the frame is sharp — it is just not the thing you wanted. The blur is even in all directions and gets worse with distance from the focal plane. Partially recoverable.
Camera shake. The whole frame is blurred in a consistent direction. Point lights become short streaks, and those streaks all point the same way. Partially recoverable, and the streak direction tells you how.
Subject motion. The background is sharp and the subject is not, or one part of the subject is not. Barely recoverable, but often not a problem — motion blur on a moving subject frequently reads as intentional.
Diffraction or a soft lens. Everything is slightly soft everywhere, with no direction. Usually from a very small aperture or a lens wide open. Recoverable to a degree, because there is still real detail — it is just low in contrast.
Thirty seconds of looking saves an hour of applying the wrong fix.
What is genuinely recoverable
Low micro-contrast. Diffraction softness and lens softness leave the detail present but flat. Deconvolution sharpening — which models the blur and reverses it, rather than just increasing edge contrast — genuinely recovers a real improvement here.
Mild, uniform camera shake. If the streak is short and consistent, motion deblur that is told the direction and length can recover a meaningful amount.
Softness from noise reduction. If the file was over-denoised, the detail may still be in the raw. Reprocess from the original with less aggressive settings rather than trying to sharpen the damaged JPEG.
What is not recoverable
Badly missed focus. If the eyes are 30cm behind the focal plane, the information about the eyes was never recorded. No tool can produce it. AI tools will produce something, and that something is a plausible pair of eyes rather than your subject's.
Severe motion blur. Same reason.
Blur in a heavily compressed file. Compression has already thrown away the detail that would have been used to reconstruct.
The order that gets the most out of a soft file
- Start from the raw if one exists. A JPEG has already had sharpening and compression applied, and both fight you.
- Denoise before sharpening. Sharpening noise makes noise. This order is not optional.
- Use deconvolution rather than unsharp mask where your software offers it. It is modelling the blur; unsharp mask is just raising edge contrast.
- Sharpen selectively. Mask to the areas that matter — eyes, edges, texture — and leave smooth areas alone. Global sharpening makes skin worse while making eyes better.
- Stop earlier than feels right. Look away for a minute and come back. Over-sharpening looks fine while you are doing it and obvious the next day.
- Check at 100% and at final output size. Halos that are invisible at 50% are unmissable in print.
What AI adds, honestly
Machine-learning sharpening and deblur tools are genuinely better than the sliders that came before, particularly on mild shake and lens softness. They are worth having.
What they also do is invent. On a badly out-of-focus face, an AI deblur will produce a sharp face — and it will be a face that is confidently, subtly not the person. On a portrait that is worse than delivering the soft version, because the soft version is honest.
Test on a file where you know the answer: take a sharp frame, blur it deliberately, and try to recover it. You will see exactly what the tool reconstructs and what it invents.
When to accept it
Some frames are soft and the picture is still the picture. A slightly soft candid with the right expression beats a technically perfect frame with the wrong one, and clients agree far more often than photographers expect.
Two things help: deliver it smaller — a soft image at 1200px is a fine image at 1200px — and convert to black and white if it suits, because monochrome is more forgiving of softness than colour.
Prevention, since it is cheaper
Faster shutter, better light, single-point focus on the near eye, back-button focus, and burst on anything moving. None of that helps today's file, but it means fewer of these conversations.
What we do with a soft frame
If you send us a set with one soft frame in it, we will tell you which of the four it is and what is realistically achievable, rather than sharpening it until it looks processed and calling it done. Sometimes the honest answer is that the frame is what it is — and knowing that is worth more than a halo.
Everything that IS recoverable is part of Standard Retouching; nothing about sharpening is charged as an extra.
Deciding what to do with the frame
Once you know which blur you have, the decision is usually quick.
Recoverable: denoise, then deconvolve, then sharpen selectively, then stop earlier than feels right. Check the next day.
Not recoverable but the picture is still good: deliver it smaller, and consider black and white. Monochrome is more forgiving of softness than colour, and a soft frame with the right expression regularly beats a sharp one with the wrong one.
Not recoverable and the picture is not carrying it: leave it out. One frame that does not hold up draws attention to itself among frames that do.
The frames worth spending real time on are the ones where softness is the only problem. Everything else the frame needs — even skin, a clean background, matched colour — is ordinary work, and it is what makes the difference between a rescued frame and a finished one.
Common questions
Can AI actually fix an out-of-focus photograph? It can produce a sharp photograph. Whether that is your photograph depends on how far out of focus it was — badly missed focus means the information was never recorded, and what comes back is a plausible reconstruction rather than a recovery.
What is the difference between deconvolution and unsharp mask? Deconvolution models the blur and attempts to reverse it. Unsharp mask simply raises contrast at edges. On genuine softness, deconvolution recovers real detail; unsharp mask makes an impression of it.
Should I sharpen before or after noise reduction? After, always. Sharpening noise makes noise, and this order is not a preference.
Is a soft frame ever worth delivering? Frequently. A slightly soft candid with the right expression beats a technically perfect frame with the wrong one, and clients agree more often than photographers expect. Deliver it smaller, and consider black and white — monochrome is more forgiving of softness.
How do I tell which kind of blur I have? Open at 100% and look at a high-contrast edge. Even in all directions is focus; consistent in one direction is shake; sharp background with a soft subject is motion; slightly soft everywhere with no direction is the lens or diffraction.
Related reading: How to Upscale a Photo Without Making It Look Fake and RAW vs JPEG for Professional Retouching.
BeEdits Editorial Team
Post-production studio
A photography post-production company: high-end retouching, creative editing and scalable production support for photographers, studios and brands.