This is a LoRA trained on about 8.7k images of Wuthering Waves characters. It's base model is Illustrious WaifuAI (WAI) (whose base model is SDXL). The base model can be found here: <a target="_blank" rel="ugc" href="https://civarchive.com/models/827184/wai-illustrious-sdxl">https://civarchive.com/models/827184/wai-illustrious-sdxl</a>.The model can generate some girls from Wuthering Waves, but not all of them and not every one equally as good. The characters that can be generated pretty accurately (they're recognizable) if they've been tagged, otherwise not as good or I haven't tried (I don't play WuWa so I don't know all chars)The images were downloaded using grabber (<a target="_blank" rel="ugc" href="https://github.com/Bionus/imgbrd-grabber">https://github.com/Bionus/imgbrd-grabber</a>) that grabbed images from sources like konachan, safebooru and deviantart. Because there were over 8.7k images and I could not get the dataset tag editor extension from <a target="_blank" rel="ugc" href="https://github.com/toshiaki1729/stable-diffusion-webui-dataset-tag-editor">https://github.com/toshiaki1729/stable-diffusion-webui-dataset-tag-editor</a> working, I pruned neither the dataset images nor the tags (images probably not that bad, but tags definitely would have helped pruning them a bit). I only added the trigger words "wuthering waves" as the first tag to every image's tags.I trained on an RTX 4070 Ti SUPER (16GB VRAM) and it took about 5.5 hours with gradient accumulation of 3 and bucket size of 4 using these scripts: <a target="_blank" rel="ugc" href="https://github.com/derrian-distro/LoRA_Easy_Training_Scripts">https://github.com/derrian-distro/LoRA_Easy_Training_Scripts</a>.

Files with SHA256: 78a6a24d3ff441d9e51e742b37bb7b741e0ca8d4c899884eed038f5418d7ffd5