Alina Volkanova, a fashion model. Retired now.
Training has been done with her early photographs when she was 18-24.
Description
FAQ
Comments (4)
I'd dearly love to know how you managed to train an SDXL embedding. I haven't been able to find a guide anywhere!
There is nothing much diffirent from other SD models. A good dataset and good captions then find find the good training method for the dataset, I rarely use the same one. Tries and tries, tests and tests.
There is no magic, just work.
If you're looking for tools, check out OneTrainer; does a great job of being easy to use, powerful and relatively resource-light. The masking feature makes outputs quite a bit better than unmasked embeddings, in my experience. Check out settings used by embedding creators to get an idea of a decent baseline to start with and tweak to your own satisfaction.
As @hypopo says though: there's no magic, just work.
@sendmefreestuffa8246
One Trainer is a great tool maintained by great developers. Eazy to use - even training is not an eazy thing but it makes thing eazy.
All my previous published embeddings were made with it.