Trained on NoobAI V-Prediction.
Trigger word
h1t0maruRecommended strength is 0.8 on NoobAI; using 1.0 may make the image too blurry.
Due to the selection of regularization images, I found some tags may cause lora be less effective, for example "azur lane". If you notice any other tags, please tell me and I will adjust regularization images for the next train.
V2
Switch to LoKr training.
Added trigger word.
Added Regularization images.
Removed old images; added latest curated samples.
Model size reduced.
V1
Trained on Noob V-Prediction
No triger word
LoRA