How to use
This model has been trained on base SD 1.5. It should be compatible with almost any checkpoint although quality may vary. It works with male or female characters. You can invoke it with the keyword “ww1brit”. The LoRA weight depends heavily on what you want to do so it’s hard to recommend something specific, I would say from 0.5 to 1 give the best results in normal scenarios. Playing with the CFG scale and hires fix is recommended.
Historical accuracy
This LoRA is mostly based on photographs of reenactors, specifically on WW1 British uniforms while avoiding WW2 uniforms that are pretty similar. There are some instances in the dataset where an American uniform was used so some insignia can appear.
Dataset
Images of costumes, cosplays, reenactments and original photographs from the war were used. A total of 16 images were used.
I plan to make other LoRA themed about World War 1 so don’t hesitate to give advice and feedback!