Shao Meng shows Ming-Image splitting AI fashion images into editable layers

Shao Meng (@shao__meng) argues that apparel e-commerce needs reusable image layers, because generating every image from scratch fails to keep many similar product photos consistent.

A separate post by li9292 tests the open-source Ming-Image model on an AI outfit image, splitting it into 9 transparent layers so one layer, such as a jacket color, can change while the rest stays unchanged. Shao Meng says each layer must be labeled with its content, for example "layer 1 is the watermark," and that the method is not yet stable. A free online trial page is linked, and Shao Meng notes that it sometimes stalls.