While scrolling through discord I found a link to the tweet shown above of an artist saying "Guess this shows that my style is hard to reproduce by one (too much variations between each works), there is some attempt with the lines but it’s barely even close overall" I've been learning how to train LoRAs since a few days ago and this was the best way to test my knowledge, by training the Style from that artist at the same time I may learn how to make LoRAs better while lowering his ego 2birds with 1 stone! I have left a Base64 encode inside the model's metadata whoever finds it will have the dataset/config inside it! |