Animagine XL 3.1: Open-Source Anime Text-to-Image Model 130

Animagine XL 3.1 is an open-source text-to-image anime model. By integrating new datasets, Animagine XL 3.1 expands its understanding of a wide range of anime works and styles, from classics to the latest releases, covering artistic styles from ancient to modern.

Animagine XL 3.1 Online Experience:https://huggingface.co/spaces/cagliostrolab/animagine-xl-3.1

Animagine XL 3.1 Features:

• By integrating new datasets, Animagine XL 3.1 expands its understanding of a wide range of anime works and styles, from classics to the latest releases, covering artistic styles from ancient to modern.

• Resolves overexposure issues, improving the quality of generated images.

• Adds new aesthetic tags, updates quality and year tags, and supports generating images with specific aesthetics and era styles.

• Adopts tag ordering, optimizing the accuracy of generated results.

• Optimized algorithms and training process, trained for approximately 350 hours using 2x A100 80GB GPUs, specifically enhancing image details and color performance.

• Data Richness: The model was pre-trained using a dataset containing 870,000 ordered and annotated images. They cover a wide range of anime characters, styles, and themes, providing the model with a deep knowledge foundation.

Animagine XL 3.1 Usage

There are multiple ways to get started with this model:

• Animagine XL 3.1 was early-released on SeaArt and Huggingface.

• Animagine XL 3.1 has been launched on Huggingface Spaces, powered by Zero Nvidia A100 GPUs.

• Animagine XL 3.1 will also be released on other platforms later.

Animaine XL 3.1 License

Animagine XL 3.1 is based on Animagine XL 3.0 and falls under the Fair AI Public License 1.0-SD, compatible with the Stable Diffusion model license. Key points:

• Share Modifications: If you modify Animagine XL 3.1, you must share your changes and the original license.

• Source Code Accessibility: If your modified version is accessible via the web, provide a way for others to obtain the source code (e.g., a download link). This also applies to derived models.

• Distribution Terms: Any distribution must follow this license or another license with similar rules.

• Compliance: Non-compliant behavior must be resolved within 30 days to avoid license termination, emphasizing transparency and adherence to open-source values.