I just tried training with only 4 images, but from the same country, which have more similarity. I w
I just tried training with only 4 images, but from the same country, which have more similarity. I was thinking of training a LoRA for each country and using it dynamically.
Do you think it could work? But I only have 4 images per country, and I still don’t know if I should remove the backgrounds from my .png files before training? In how many steps?
Do you think it could work? But I only have 4 images per country, and I still don’t know if I should remove the backgrounds from my .png files before training? In how many steps?
