Contact sheet
One cut-out is easy. Four hundred that match is not.
Removing the background from one photograph is close to a solved problem. Removing it from four hundred and having them look like they belong to the same catalogue is not, and that is a different question with different answers.
Crop driftframe to frame
Shadow kept or droppedinconsistent
Edge hardnessvaries by subject
White that is not whiteoff by a few values
Subject scalefills a different share
Off-centreafter auto-crop
Why a set is harder than a photo
roll a · six framesEvery one of the frames above is a thing that can be individually correct and collectively wrong. A tool judged on a single hero image is being judged on the easy version of the job. Look down a column of two hundred thumbnails and the failures are not bad cut-outs — they are good cut-outs that do not match each other.
Downstream
What the destination demands
Removal is rarely the goal. Something else needs the image in a particular state, and that requirement is what decides whether an output is usable.
Marketplaces want an exact background
Large retail platforms typically specify pure white for a main product image rather than “white-ish”, and often set a minimum share of the frame the product must fill. The numbers differ per platform and change; the platform's own current spec is the only authority worth trusting.
Web layouts usually want transparency
A cut-out that will sit on a coloured page needs a real alpha channel, which means PNG or WebP. Exporting to JPEG discards it and bakes in whatever background was behind the subject at the time.
Print wants resolution you cannot add later
Free tiers often return a smaller image than you gave them. That is invisible on screen and fatal at press size, and it is the sort of limit stated on a pricing page rather than in the interface.
Catalogues want the shadow decided once
Keeping a natural contact shadow and dropping it are both defensible. Doing one for some items and the other for the rest is what makes a grid look broken, and it is the most common inconsistency in a mixed batch.
Questions a batch raises that a single image does not
roll bIs the result deterministic?
Run the same image twice and some tools return the same cut-out, others do not. It matters when you re-process a catalogue and half of it shifts for no visible reason.
Can the settings be fixed across a run?
Automatic per-image decisions about cropping, padding and shadow are convenient for one photo and the direct cause of drift across hundreds. The useful control is the one that turns the automation off.
What happens to the failures?
Some proportion of any batch will come back wrong. Whether the tool flags those or silently returns them is the difference between a review queue and finding out from a customer.
Is there an API, and does it cost the same?
Interface pricing and programmatic pricing are frequently different products. A workflow that is affordable by hand can be expensive automated, which is usually discovered after the workflow is built.
House rules
How this will be written
No invented numbers
No scores out of ten, no star ratings, no tier boards. Where something is better, it gets said in words and the reason is shown.
Only claims we can stand behind
Nothing describes a test that did not happen. Pricing and feature claims are checked against the vendor's own pages.
No paid placements
Nothing here is sponsored, and outbound links carry no affiliate tracking and earn no commission. No company pays to be included or to be written about more kindly.
This site is new and there are no tool write-ups on it yet. The frames on this page are labelled with the things that drift in a batch — they are a checklist, not results, and no batch has been run. When write-ups land they will take the tools one at a time, with the reasoning shown rather than a score attached.