⚠️ Version Compatibility Notice:
v1.0 & v2.0: Trained exclusively for krea2.
v3.0: Trained specifically for moss. Please be careful to use the correct version for your base model.
Note on v3.0 (moss): While moss has great native quality, using this LoRA may cause some quality degradation around the facial area. Please keep this in mind and use it only when you specifically need a cartoon-rendered in-game graphic style.
For more details on v3.0, please refer to the version notes.
📜 License: Please strictly follow and comply with the license terms of the respective base model (krea2 or moss) used for each version.
This is an anime-style 3D in-game rendering LoRA (Wuthering Waves)
Base: Trained on Krea 2 RAW / Tested with Krea 2 Turbo fp8 (step:8, cfg:1.0, euler, simple)
Trigger: WUWAstyle
Recommended strength: 0.9
Because the LoRA was trained with many diagonal compositions to help convey the signed distance field shader look, generated character poses may also tend to appear at an angle. In addition, Krea 2 sometimes struggles to frame the entire body. If that happens, try adding prompts such as POV, looking at the camera, full-body shot, and more explicit descriptions of the feet or shoes.
License & Attribution
This LoRA is a derivative model based on Krea 2 and is distributed subject to the Krea 2 Community License Agreement.
Krea 2 is licensed under the Krea 2 Community License Agreement.
For more information, please visit:
https://krea.ai/krea-2-licensing
This model modifies Krea 2 through LoRA fine-tuning for an anime-style 3D in-game rendering aesthetic.
This is an unofficial community-made model and is not affiliated with, endorsed by, or sponsored by Krea.ai, Kuro Games, Wuthering Waves, or any other game developer or publisher.
Use, redistribution, modification, and commercial use of this LoRA must comply with the Krea 2 Community License Agreement and all applicable third-party intellectual property rights.
Please do not use this LoRA to reproduce or distribute copyrighted game assets, characters, logos, or other protected material in ways that infringe the rights of their respective owners.
Users are solely responsible for how they use this LoRA and for ensuring that their use complies with applicable laws, licenses, and third-party rights. Please use it responsibly.
Description
This LoRA is specially designed to capture the high-end 3D cel-shaded anime game aesthetic, heavily trained on datasets from popular cartoon-rendered games such as Wuthering Waves and Genshin Impact.
It builds upon the existing ver2(krea2) dataset by incorporating additional refined facial data, and has been specifically fine-tuned and optimized for moss.
Recommended Settings
1. LoRA Weight: 0.7 – 0.85 (Adjust within this range depending on your prompt and base model for the best balance).
2.To bring out the best cartoon-rendered game quality and shading, it is highly recommended to include the following keywords in your positive prompt:
animation, cartoon-rendered game, volumetric face shading, layered hair shadow
3. Avoid Realistic Keywords: Please DO NOT use realism-related tags (e.g., realistic, photorealistic, real life, raw photo) in your positive prompt. Including realistic tokens may cause the character's face to become distorted/mangled, or shift the style toward an unintended uncanny 3D game look.
moss images already have terrific native quality on their own. Please use this LoRA only when you specifically want a cartoon-rendered in-game graphic style, even if it comes with a slight trade-off in raw quality.
Do not use this model to generate harmful, illegal, non-consensual, or defamatory content.
You are solely responsible for the outputs you generate using this model. Please respect platform guidelines and copyright laws when sharing your creations.