2023/5/24 update

Train the model using new craft. Reduce learning rate to mitigate overfitting and fix hands.

2023/5/7 update

Update dataset. Thanks huangdoudou2019854 for providing images. Remove the watermarks in most of the images. Due to the huge difference in the number of images of the two styles, these images were not classified before training. Therefore, the two styles have a chance to appear during the generating process.

2023/3/29 update

Fix the compatibility problem of non-NAI-based checkpoints. Now arbitrary anime model with NAI's VAE or kl-f8-anime2 VAE can also generate good results using this LoRA, theoretically.

Introduction

A conventional LoRA model trained with ~110 images by Oda Non. The trigger tag is ‘oda non’. The 784mb VAEs (NovelAI, Anythingv3, Orangemix, Counterfeit) are recommended. 0.6 ~ 0.8 weights are good. The preview images are generated using 1.6 version (original) YesMix.

Cautions (Must Read)

• Some preview images are generated using hires fix. Thus, if you wanna generate the preview images, pls download the original preview images and analyze them using 'PNG INFO' in webui to get all generating information.

• The maximum of generating resolution should not exceed 1024 theoretically, otherwise extra/deformed head/limbs will be waiting for you :). Try to use 'hires fix' to prevent this problem and help to generate higher-resolution images.

• For 784mb VAEs (NovelAI, Anythingv3, Orangemix, Counterfeit) users, make sure add '--no-half-vae' in command line of webui to prevent generating black images. See voldy's log for the details of edit method. For high-performance GPU, make sure 'medvram' and 'lowvram' mode are disabled.