AgainTangHuoEr - Tang Huo Er

AgainTangHuoEr Martial Universe Tang Huo Er

Training the base model as AgainCNAnime_v1, this big model is mainly used to generate the antique style effects.

Providing some references for beginners to train the Lora model

Tang Huo Er's materials use screenshots from animated videos. Due to the small pixels in the video, the character images are not clear enough, resulting in a somewhat blurry effect.

Next, start by finding a clearer image of a person's face to first train a facial Lora model, and then use the following methods to correct the image.

You can use AI image generation to enlarge and add details to the image first, or use ComfyUI + Krita to correct the character. In conjunction with the previously trained facial Lora, this can make the image modeling of the corrected face more closer to the person.

After obtaining clearer source material images, proceed with the training, which will result in a better outcome.