What marketplaces are actually asking for
Every marketplace publishes image requirements, and they are more specific than most sellers realise. Amazon's main image must show the product alone against a pure white background, filling roughly 85 percent of the frame, with no added shadow, no watermark, no logo, no text, no border and no props. Images that fail are suppressed rather than corrected, and a suppressed main image means a listing that does not appear.
Shopify and Etsy are considerably more permissive — a shadow, a lifestyle setting, even a watermark are all allowed — but they still reward consistency. A storefront where every product sits at the same scale on the same background reads as a shop; one where each photograph was framed differently reads as a collection of unrelated items, regardless of how good the individual photos are.
The presets here encode those rules rather than leaving them to you. Choosing Amazon locks the background to white and disables the shadow and watermark controls, because a batch built with them is a batch that will be rejected. Choosing Shopify or Etsy unlocks both. The point is that the constraint is enforced where it exists and absent where it does not, so you do not have to hold three sets of policy documents in your head while preparing photographs.
Why background removal is the whole job
Almost every marketplace image requirement reduces to the same thing: the product, isolated, on a controlled background. Traditionally that meant a lightbox, a seamless backdrop, careful lighting to avoid casting a shadow onto the paper, and a lot of time in an editor cleaning up the edges. It is why professional product photography costs what it does.
Segmentation models changed the economics. Given a photograph of an object, a model can now predict a per-pixel mask of what is product and what is not, accurately enough that the cut-out edge holds up at full resolution. That turns a lighting problem into a software one: shoot the product anywhere reasonably lit, and let the model separate it from whatever happened to be behind it.
It is not unconditional. Cut-outs are excellent where the product has a clear outline against a contrasting background, and progressively worse as the boundary becomes ambiguous — fine hair on a brush, mesh on a trainer, the wire of a whisk, a glass bottle whose background shows through it. Reflective and transparent products are genuinely hard, because the correct answer is not a binary mask at all. Shooting against a plain backdrop in even light remains the cheapest way to improve every cut-out you will ever make.
Consistency is what makes a storefront look professional
The difference between an amateur and a professional listing gallery is rarely the quality of any single photograph. It is that the professional set is consistent: every product occupies the same proportion of its frame, sits at the same height, and appears on the same background. Scroll a well-run store and the products change while the frame stays still.
That is difficult to achieve by hand, because it requires the same margin around objects of wildly different shapes. A tall bottle and a wide box, cropped by eye, will end up at different scales even when both look fine individually. The kit handles this by measuring each cut-out and scaling it to fit a fixed safe area inside the canvas, with the same margin on every side, so shape variation stops translating into scale variation.
The same logic applies to the background. Picking one background and applying it to the entire batch is worth more than getting any individual photograph perfect. If you sell across several channels, run the batch once per preset — the source photographs are the same, and each marketplace gets images at its own dimensions rather than a resized version of somebody else's.
Shadows, watermarks and the line between them
A cut-out product on a pure white canvas can look like it is floating, because there is nothing anchoring it to a surface. The soft shadow option addresses this by placing a blurred silhouette of the product beneath it, which reads as a contact shadow on a seamless studio backdrop. It is a compositing effect rather than a physical simulation, and that is the correct choice: listing photography wants a neutral, evenly lit look, not a directional light that implies a scene.
Watermarks are a different kind of decision. On your own site, a discreet mark is reasonable protection for photography you paid for. On a marketplace listing it is usually a mistake even where it is permitted, because it signals distrust to the buyer and competes with the product for attention. On an Amazon main image it is prohibited outright, which is why the control is disabled when that preset is selected.
If you do use one, apply it consistently and keep it small and low-contrast. A saved watermark preset carries the same mark, position and opacity across every image in the batch, which is what makes it read as branding. A watermark that moves or changes size between photographs reads as an afterthought, and one large enough to obscure the product simply loses you the sale.
What the batch costs, and what runs where
One credit per image, charged for the AI background removal, which is the only step that needs a GPU. Everything downstream of it — scaling the cut-out into the marketplace canvas, painting the background, generating the shadow, applying a saved watermark, and packaging the results into a ZIP — runs in your browser and costs nothing. A hundred-image batch is a hundred credits, and the arithmetic does not have anything else hiding in it.
That split is also the privacy answer. Each source image is uploaded for the cut-out and returned; the composition happens on your machine. The page carries a hybrid-processing badge rather than leaving you to work this out, and it is the honest thing to weigh if you are photographing a product that has not been announced yet.
Practically, the workflow is one pass: add up to a hundred photographs, pick a preset, choose background, shadow and watermark, run, and download a single ZIP with every file named after its original. Re-run the same source photographs against another preset when you need a second marketplace. The source images are never modified, so there is no quality lost to a second pass.