santa hat
deerdeer nosedeer glow
Sign In

Any advice for training single or few images LoRa training?

Someone has some tips for training a LoRa for character that have a single image or few, i tried spliting the image and adding flip versions , i got some ok-ish result , but i will like to know what methods some may use.

1 Answer

Try to flip, zoom in, zoom out, adding some noise, rotate, stretch,etc sometimes the results are pretty good but with some settings that fits your dataset

Your answer