<h2 id="umamusume-all-in-one-lora-k6u8az97q">Umamusume All in One LoRA</h2>This is an All-in-One LoRA designed to generate a wide variety of characters from the popular game Umamusume: Pretty Derby with a single model. Without the hassle of managing individual character LoRAs, you can easily bring numerous Umamusume to life, including Stay Gold, Almond Eye, Loves Only You, and many more, with just this one file.<h3 id="notice-g7xb14i02">Notice</h3>This model is a re-upload. The previous model page was accidentally deleted during a version update process. My apologies for the inconvenience!<h3 id="features-ubarm79mh">✨ Features</h3><ul><li>Extensive Character Library: Capable of generating most of the main characters from the game with high fidelity.</li><li>Versatile Costume Support: Faithfully reproduces each character's unique racing outfits, as well as school uniforms, casual wear, and other styles.</li><li>Easy to Use: Achieve your desired images simply by inputting character names or feature tags, without the need for complex prompts.</li></ul><h3 id="character-and-prompt-information-64t29swvm">🌟 Character &amp; Prompt Information</h3>For detailed prompt information regarding the characters, costumes, and features included in this LoRA, please refer to the Hugging Face <a target="_blank" rel="ugc" href="https://huggingface.co/datasets/UmaDiffusion/ULTIMA-prompts/blob/main/README.md">link</a>. Find the precise tags for the character you wish to create and utilize them in your prompts.<h2 id="-61q7ml2no"></h2><h2 id="license-mhq8xfg6n">License</h2><ul><li>This model is released under "<a target="_blank" rel="ugc" href="https://umamusume.jp/derivativework_guidelines/">Derivative work guidelines for umamusume</a>".</li><li>And "<a target="_blank" rel="ugc" href="https://freedevproject.org/faipl-1.0-sd/">Fair-AI public license 1.0-SD</a>" License</li></ul>

UmaMusume All In One LoRA

This is alpha version of SDXL AIO LoRA of Uma musumeRecommended options<ul><li>LoRA weight 1.0</li><li>Trigger words<ul><li>[character name] \(umamusume\) or [character name]</li><li>For example,<ul><li>manhattan cafe \(umamusume\)</li><li>hayakawa tazuna</li></ul></li></ul></li></ul>SDXL version<ul><li>trained on <a target="_blank" rel="ugc" href="https://github.com/kohya-ss/sd-scripts">sd-scripts by kohya_ss</a> and <a target="_blank" rel="ugc" href="https://github.com/KohakuBlueleaf/LyCORIS">LyCORIS by KohakuBlueleaf</a><ul><li>Thank you all a lot!</li></ul></li><li>Base model : <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0">SDXL 1.0</a></li><li>Hardware : 4x RTX TITAN</li><li>Training time : 37 hours</li><li>Dataset : <a target="_blank" rel="ugc" href="https://huggingface.co/datasets/UmaDiffusion/ULTIMA">ULTIMA Dataset</a></li><li>Resolution : 768x1024 with <a target="_blank" rel="ugc" href="https://github.com/NovelAI/novelai-aspect-ratio-bucketing">aspect ratio bucketing</a></li><li>Dim, Alpha : 32<ul><li>with <a target="_blank" rel="ugc" href="https://github.com/kohya-ss/sd-scripts/blob/main/networks/resize_lora.py">dynamic pruning</a> rank=24</li><li>Network dropout = 0.1</li></ul></li><li>Optimizer : AdamW, weight decay=0.1</li><li>Steps : 12,024</li><li>Total Batch size : 48<ul><li>4 GPU x 3 batch size x 4 Gradient Accumulation steps</li></ul></li><li>Learning rate : warmup to 1e-4 for 340 steps and then kept constantLR scheduler<ul><li>LR scheduler is constant_with_warmup.</li></ul></li></ul>Settings<ul><li>For Stable Diffusion XL<ul><li>Use a model derived from Stable Diffusion XL <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-diffusion-xl-base-0.9">0.9</a> or <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0">1.0</a></li><li>DPM++ 2M SDE Karras</li><li>about 30 steps, CFG scale ~6.5</li><li><a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/sdxl-vae">SDXL vae</a></li></ul></li><li>CLIP skip = 1 or 2. Use whatever you want.</li><li>Use Hires.fix to get higher quality image<ul><li>Upscaler<ul><li>Latent, Latent (nearest-exact), Latent (bicubic antialiased) or other Latent series.</li></ul></li><li>Denoising strength 0.50~0.65</li></ul></li></ul>If you want to use SDXL-alpha LoRA with <a target="_blank" rel="ugc" href="https://civitai.com/models/260267/animagine-xl-v3">AnimagineXL</a>, use weight around 0.6

Files with SHA256: 92c3096ca721fc5108156454c6351fa882c823745036a75d95e57f5ee7012118