Stable Diffusion and better AI art - Textual Inversion, Embeddings, and Hasan

Perfectly explained

Hello, I have question about the embendings. They only work in stable 2.1?? or they work in stable 1.5 too????

WTF is Hassan?

Very helpful. Thank you

Do you think this would be possible with an 8gb 3060ti? I'm able to run automatic1111's webui locally with no issues but have yet to attempt this, assuming I didn't have enough vram. You mentioned training the same embedding multiple times, so would it actually not matter as much if it craps out 1/4th of the way through if you can just train the same embedding with the same images again?

you can make the training data set for mauli better by choosing better images that highlight her actual features (no flying face syndrome, but to include her paws and legs), in multiple lighting setup (most of the images are too dark and also she's dark coloured, so model will obviously have a hard time training)