RAEMU ANIMA

Raemu Anima is a finetuned model focused on anime visuals with a distinctive 2.5D texture and Anima architecture.

Rather than primarily focusing on expanding character or concept knowledge, this model is trained with a stronger emphasis on style, aesthetics, image quality, and generation stability.

The goal is to produce polished anime artwork with a refined 2.5D appearance while maintaining good detail, composition, and consistency across a wide range of prompts.

Recommended Settings

These settings are recommended as a starting point. You can experiment with other samplers, schedulers, and CFG values depending on your preferred results.

Sampling

• Sampler: Euler / Euler SDE

• Scheduler: SGM Uniform / Beta

Resolution

• Supports resolutions up to 1536px

• Use resolutions divisible by 32

• For easier resolution selection in ComfyUI, I recommend using Resolution Selector nodes

Recommended Prompt

A simple starting prompt:

masterpiece, best quality, absurdres

Negative Prompt

worst quality, low quality, artist name, jpeg artifacts, bad anatomy, extra limbs, deformed, bad hands, missing fingers, watermark, signature,

These prompts are only recommended starting points, so feel free to adjust your prompting style depending on the desired composition, aesthetic, and level of detail.

Support the Project

If you enjoy my model and would like to support its continued development, consider supporting me through Ko-fi

Your support helps me cover the costs of:

• GPU rental

• Model training

• Dataset preparation

• Training experiments

• Future model development

Every bit of support helps me continue experimenting and creating new models.

Thank you for using Raemu Anima and for supporting my work