Photoroom picked one job and did it properly, which in this category counts as a strategy. Take a photo of a product on a kitchen table, get a clean listing image. The background removal is the core and it is genuinely better than the general purpose alternatives on exactly the cases that matter for commerce, which is a well lit object with defined edges. The help material is written for the actual audience, meaning someone running a small shop who takes photos on a phone between other tasks, and it is refreshingly free of the language most AI products use.
No transformation, no revolution, just here is how you get a hundred listing images done this afternoon. The batch processing guidance is the most valuable part and gets less prominence than it deserves. Removing one background is a novelty. Removing four hundred consistently, with the same crop, the same margins and the same shadow treatment so a catalogue page looks coherent, is a business problem, and the material handles it well.
The marketplace templates are similarly practical, since every platform has its own image requirements and getting those wrong costs sales in ways that are hard to diagnose. Generated backgrounds are where I would push back. They are convincing individually and they have a look, that slightly too clean studio scene with soft even light and a surface that does not quite exist, and buyers are increasingly able to spot it. Placing every product against the same generated marble surface produces a catalogue that reads as synthetic, and the material never raises this.
A section on when a real background beats a generated one would improve outcomes more than another feature walkthrough. The honesty question is the bigger omission. There is a line between cleaning up a photo and misrepresenting a product, and it sits somewhere around colour accuracy, texture, scale and condition. A generated shadow is fine.
Subtly saturating a fabric so it reads richer than it is will produce returns and bad reviews, and it is easy to do without intending to. The material covers the capability and says nothing about the judgement, and for an audience of small sellers who mostly have not thought about it, that is a real gap with commercial consequences for them. Difficult subjects get glossed over. Glass, transparent packaging, fine hair, jewellery, anything reflective.
These are where the model struggles and where a seller with that kind of inventory will form their opinion of the tool, and the material offers little on how to shoot them so the tool succeeds. Better input photography guidance would lift results more than any editing tip, and there is surprisingly little of it. The free tier is restrictive enough that judging the tool properly means paying, which is understandable and mildly irritating. Three point five.
An honest, focused tool with documentation pitched correctly at its audience, marked down for saying nothing at all about the two things that will actually cause its users problems, which are visual sameness and the line between enhancement and misrepresentation.