An experiment at training an item of clothing / body jewellery. Trained on the Civitai training service using 7 photos of the piece draped on a mannequin taken from different angles.

Results are fairly true to the original images, but this also leads to a situation where results are generated with out ??????? (male and female).

Use the activation word ChainHalter and you may sometimes have to use the word mannequin in the negative prompt.

If you can work out how to generate images with ??????? (male or female) with other resources, please post the results - I may use them as training data in a later version. (Also, if you can point me to photos of person wearing the same item, that would help for future versions)