I’m now creating sd1.5 Lora’s I figured out how to train a Lora with under 15 images and 50 steps it takes only 9 minutes no need to wait for 1000 steps it’s awesome!! All done on iOS 13 pro 6gb ram

Stealth_Echelon

This version took s couple retries but I finnaly got the style I relized I need to add trigger word for style Lora’s as sd1.5 uses the first tokens that u put in your images as the triggers for each image style so to ge the rainy out door style for example put this at beginning of your prompt:A dimly lit, rain-soaked night scene from a sci-fi film features aThis is not a trigger word but an image style trigger from the training data so download the training data and copy the first tokens from what ever images style you want in your images if I had of put a trigger word I can just put the trigger word and it would use all the training data images style because it puts the trigger word as first token in each image caption this is how sd1.5 works next time if I make a version 3 I’ll put a trigger word😅👍🏾

Files with SHA256: 5f5dddc93a5d67903fb946f3d1fc1b41caf04c870906390aa93f15358426d356