Every other page here describes a category the engine handles well or handles with known limits. This one is different, and it is more useful to you to say so.
Why it is a different problem
Putting a coat on someone adds an object to a photograph. Putting a wig on someone replaces something that is already there — their existing hair, over an area that includes the face boundary, which is the part of an image a viewer scrutinises hardest.
The pipeline composes garments onto a person. A wig is not a garment in that sense, and the same call that reliably layers a jacket over a shirt is doing something structurally harder when the target is hair.
Where we actually are
Not benchmarked, not tuned for this, and no measured pass rate. It runs — the mechanism does not refuse a wig product — and the results are inconsistent in a way that we have not characterised well enough to describe honestly beyond that.
If wigs are your primary category, this is not the product to build your storefront on today. If they are one line among many, the integration covers them at no extra cost and you can judge the output on your own catalogue before putting it in front of shoppers. Run it in the admin playground on your own products first — that is what it is for.
If you try it anyway
Straight-on, well-lit photographs. Existing hair worn flat and back gives the model the least to undo. Product images shot on a plain background rather than on a mannequin head with a competing hairline.