The one question that settles most cases
Ask this: does the fix require information the camera did not capture?
If the answer is yes, you need a reshoot. No amount of editing invents detail that was never recorded. If the answer is no — the information is present but presented badly — retouching is almost always cheaper and faster.
Zoom to 100% on the area that bothers you. If you can see the texture, edge, or colour you want, it is a presentation problem. If you see mush, clipping, or a flat white patch, it is a capture problem.
This sounds obvious and yet teams routinely violate it in both directions. A blown-out highlight on a satin fabric is a capture failure — the sensor recorded pure white, so there is nothing to recover, and a retoucher can only paint in an invention. Conversely, a cluttered kitchen background behind a ceramic mug is purely presentational: every pixel of the mug is intact, and the background is disposable. That second image gets reshot constantly, and it should not be.
Capture faults that force a reshoot
These are the ones where the file is genuinely short of data. Retouching them produces work that looks manipulated, because it is.
| Fault | Why post can't fix it | Verdict |
|---|---|---|
| Blown highlights on the product | Clipped channels contain no detail to recover | Reshoot |
| Motion blur or missed focus | Sharpening amplifies noise, not detail | Reshoot |
| Wrong product, wrong variant, wrong size | It is the wrong object | Reshoot |
| Missing angle the listing needs | You cannot rotate a 2D image | Reshoot |
| Damaged or misfitted sample | Fixing it misrepresents the goods | Reshoot |
| Severe lens distortion at close range | Correction stretches and softens edges | Reshoot |
Two of those deserve emphasis because they are compliance issues, not just quality ones. Retouching a damaged sample so it looks pristine, or editing a garment so it fits a body it does not fit, produces a product image that misrepresents what ships. That is the fastest route to returns, chargebacks and marketplace enforcement — and no cost saving justifies it.
The missing-angle case is the most common capture fault in practice. A shoot that produced four clean images cannot become the six-image gallery your category needs. If you are consistently short, the problem is the shot list, not the retouching budget — see our breakdown of what belongs in each gallery slot before booking the next studio day.
Presentation faults that should never trigger a reshoot
Everything below is a fix applied to information the file already contains. Reshooting for any of these is pure waste.
Often reshot (wrongly)
- Cluttered or coloured background
- Product too small or off-centre in frame
- Inconsistent scale across a category
- Colour cast from mixed indoor lighting
- Harsh or missing shadow under the product
- Dust, lint, fingerprints, stray threads
- Wrong aspect ratio for the channel
What actually fixes them
- Background removal or replacement
- Reframing against a fixed safe area
- Deterministic scale normalisation
- White balance correction against a known reference
- Shadow reconstruction or preservation
- Spot cleanup
- Recrop from the original resolution
The category that trips teams up most is inconsistency rather than outright badness. A collection page where every item is technically fine but framed at a different scale reads as amateur, and the instinct is to reshoot the whole set to a single standard. It is nearly always cheaper to normalise scale and plate colour in post, because the fix is deterministic and applies to the archive as well as to new work.
If the fault would be identical across 200 images, it is a pipeline problem and belongs in post. If it is unique to this one shot, look harder at whether it is a capture fault.
The cost maths, with the numbers that actually matter
The comparison people make is per-image editing cost against per-image studio cost, and that comparison is misleading because it ignores the fixed costs on the reshoot side.
A reshoot is not one line item. It is sample retrieval and logistics, studio or space booking, a photographer's minimum call-out, styling time, the shoot itself, ingest, selection, and then retouching anyway — because raw capture is not listing-ready. Professional retouching alone runs roughly $25–50 per image at agency rates; a full studio day with a photographer, assistant and stylist commonly lands in the $1,200–2,500 range before any of that retouching happens.
Then there is the cost nobody books: the listing is live while you wait. A product sitting behind a weak hero image for two weeks of a reshoot cycle is losing click-through the entire time, and that loss does not appear on any invoice.
The practical read: roughly two-thirds of reshoot spend goes to things that have nothing to do with the fault you are correcting. That asymmetry is why the default should be retouch-first, with reshoot reserved for genuine capture failures.
How AI editing moved the line
The triage rule is old. What changed in the last two years is where the line falls, because the class of fault that AI tooling handles reliably has widened considerably — and the per-image economics of post-production dropped by a large multiple rather than a few percent.
Faults that used to be borderline are now firmly on the retouch side:
- Complex-edge cutouts. Hair, fur, mesh, chiffon and jewellery chains were once manual-mask work priced accordingly. Modern segmentation handles them well enough that background replacement is no longer a reason to reshoot.
- Colour accuracy. Matching a garment to a known reference swatch used to depend on a colourist's eye. It is now a measurable correction against a target, which means it is repeatable across a catalog rather than negotiated per image.
- Scale and framing consistency. Deterministic reframing against a category rule normalises an entire archive without touching a camera.
- Shadow reconstruction. Losing the contact shadow with the background used to make cutouts look pasted. Preserving or rebuilding a plausible shadow keeps a cutout looking photographed.
Two things have not moved, and it matters that you know them. AI cannot recover clipped detail — a blown highlight is still gone. And AI should not be used to change what the product materially is: fit, condition, colour beyond accurate correction, or included accessories. The first is a technical limit; the second is a line you choose not to cross.
"We can fix it in post" becomes a licence to shoot carelessly. Post-production widening the salvage window is not permission to lower capture standards — cheap fixes on bad inputs still produce mediocre listings, just faster.
Retouchable exists in this middle ground: it takes salvageable capture and normalises background, scale, colour and plate so a mixed-source catalog reads as one shoot. It does not replace the shoot when the shoot is what failed.
A triage workflow you can run on a real catalog
Deciding image by image, in the moment, is how teams end up inconsistent. Run it as a pass instead.
- Batch by fault, not by product. Sort the problem images into the two buckets above before pricing anything. You will usually find the reshoot pile is a third of what you assumed.
- Check the source resolution first. Reframing and recropping spend pixels. If the original cannot survive a crop at your channel's largest required size, it moves to the reshoot pile regardless of the fault type.
- Group the reshoot pile by sample location. Reshoot cost is dominated by fixed overhead, so twelve items in one session cost dramatically less per image than twelve sessions. If the pile is small, it may be worth holding it until the next scheduled shoot rather than calling one.
- Fix the retouch pile as a set, not individually. The same background, scale and colour rules should apply across the batch — that is what produces consistency, and it is also what makes the work cheap.
- Review against the live grid, not in isolation. An image that looks corrected on its own can still be the odd one out on the collection page. Judge it where shoppers will see it.
- Feed faults back to capture. Log which faults recurred. A repeated capture failure is a lighting or process defect that will keep generating reshoots until someone changes the setup.
That last step is the one with compounding value. Teams that track fault categories across a quarter usually find two or three systematic issues generating most of their reshoots — and fixing the setup once removes the cost permanently.
The four ways this decision goes wrong
| Failure mode | What it looks like | Correction |
|---|---|---|
| Perfectionist reshooting | Reshooting for faults nobody but the team notices | Judge at thumbnail size, on the live grid |
| Salvaging the unsalvageable | Heavy edits on soft or clipped files; results look painted | Apply the information test at 100% |
| One-off thinking | Deciding image by image, producing an inconsistent catalog | Triage in batches by fault type |
| Ignoring the waiting cost | Two-week reshoot cycles on live, underperforming listings | Ship the post fix now, reshoot on the next cycle |
The pattern underneath all four is treating the decision as a quality judgment when it is an economic one. The question is not "is this image as good as it could be" — it is "what is the cheapest path to an image good enough to sell, and how fast can it be live."
Most catalogs, run through this triage honestly, discover that the majority of their problem images are presentation faults on adequate captures. Those get fixed the same day. The genuine capture failures are fewer than expected, and grouping them into one properly planned session costs less than the scattered reshoots they would otherwise become. For the operational side of that planning, our guide to product photography turnaround time covers how the cycle length itself shapes what you can afford to reshoot.