<h2 id="a-whole-bunch-of-xl-models-are-coming-soon-just-trying-to-find-the-best-possible-settings-:p-hh4mdo5ra"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/7d8163af-9cd5-4524-92d2-fe1217d016c6/width=525/7d8163af-9cd5-4524-92d2-fe1217d016c6.jpeg" />A whole bunch of XL models are coming soon, just trying to find the best possible settings :P</h2>Works best at:0.9 (Confusion XL) (Replicates the original artists style to an extent)0.9 (Other anime Models such as Animagine, Anything)Generation Parameters: type97, dark brown hair, amber eyes, twintails, long hair, red hair ribbon, black dress, black cloak, white skirt, red necktie, white thighhighs, black boots, ankle pouch<hr />Additional Details / Training Info (For anyone wanting to train SDXL for themselves)Trained using OneTrainer (<a target="_blank" rel="ugc" href="https://github.com/Nerogar/OneTrainer">Link</a>) (<a target="_blank" rel="ugc" href="https://huggingface.co/JVGA/OneTrainerSDXL12GBPreset/blob/main/SDXL%20v2%20Lora%2012GB.json">Preset File</a>)20 Epochs (Final model uses the full 20)10 Dataset Repeats0.0003 Learning rate896x896 Train ResolutionTrained using AnimagineXL V3(NOTE: FOR ANYONE LOOKING TO TRAIN ON THE KOHYA SDXL COLAB, DO NOT. IT IS BEYOND BROKEN AND ANY MODEL IT "TRAINS" DOESN'T WORK)(ANOTHER NOTE: Through some testing, Kohya-SS trained models will only work with the model it was originally trained on, OneTrainer doesn't have this issue. For the best compatibility use OneTrainer).

Files with SHA256: 1e8e99c11d21bb005390481b5b81d4b81cda2f064bb75aa82925cab58f526f90