SeaArt-Furry-XL-1.0

SeaArt Furry XL 1.0

SeaArt-Furry-XL-1.0, built on the SDXL framework, focuses on high-quality furry art images creation. By analyzing millions of furry pictures, it sets new standards in furry imagery understanding and creation. Incorporating vast knowledge of furry characters and extensive species calibration, including mammals and birds, it refines artist styles and quality hints. SeaArt-Furry-XL-1.0 aims to offer furry enthusiasts and artists an accurate and detailed generation tool, encouraging collaboration to enrich the furry ecosystem.

Model Details

• Developed by: SeaArt
• Model type: Diffusion-based text-to-image generative model
• License: Fair AI Public License 1.0-SD
• Summary: This model generates images based on text prompts. It is a Latent Diffusion Model that uses two fixed, pre-trained text encoders (OpenCLIP-ViT/G and CLIP-ViT/L). It follows the same architecture as Stable Diffusion XL.

Diffusers Installation

First install the required libraries:

pip install diffusers transformers accelerate safetensors --upgrade

Then run image generation with the following example code:

import torch
from diffusers import DiffusionPipeline

pipe = DiffusionPipeline.from_pretrained(
    "SeaArtLab/SeaArt-Furry-XL-1.0", 
    torch_dtype=torch.float16, 
    use_safetensors=True, 
)
pipe.to('cuda')

prompt = "canid, canine, fox, mammal, red_fox, true_fox, foxgirl83, photonoko, day, digitigrade, fluffy, fluffy_tail, fur, orange_body, orange_fur, orange_tail, solo, sunlight, tail, mid, 2018, digital_media_(artwork), hi_res, masterpiece"
negative_prompt = "nsfw, lowres, (bad), text, error, fewer, extra, missing, worst quality, jpeg artifacts, low quality, watermark, unfinished, displeasing, oldest, early, chromatic aberration, signature, extra digits, artistic error, username, scan, [abstract]"

image = pipe(
    prompt, 
    negative_prompt=negative_prompt,
    width=832,
    height=1216, 
    guidance_scale=7,
    num_inference_steps=28
).images[0]

image.save("./output/seaart_test.png")

Usage Guidelines

To fully utilize the SeaArt-Furry-XL-1.0 model and generate high-quality furry art images, we recommend following these guidelines:

Prompt Structure:

The model was trained with a specific calibration order: species, artist, image detail, quality hint, image nsfw level. It is recommended to construct prompts following this order for optimal results. For example:

Prompt input: "canid, canine, fox, mammal, red_fox, true_fox, foxgirl83, photonoko, day, digitigrade, fluffy, fluffy_tail, fur, orange_body, orange_fur, orange_tail, solo, sunlight, tail, mid, 2018, digital_media_(artwork), hi_res, masterpiece"

Species and Character Calibration:

We have provided a series of nouns for main species calibration such as mammals, birds, and have repeatedly trained on specific furry characters. This helps in generating more accurate character images.

Quality Hints:

The model supports various levels of quality hints, from "masterpiece" to "worst quality". Be aware that "masterpiece" and "best quality" may lean towards nsfw content.

Artwork Timing:

To get images in the style of specific periods, you can use time calibrations like "newest", "late", "mid", "early", "oldest". For instance, "newest" can be used for generating images with the most current styles.

Recommended Image Sizes:

For best-quality images, it is recommended to generate using one of the following sizes: 1024x1024, 1152x896, 896x1152, etc. These sizes were more frequently used in training, making the model better adapted to them.

User Studies

To gain a deeper understanding of how SeaArt-Furry-XL-1.0 is applied within the furry art community and to assess user satisfaction, we invited artists, designers, and furry enthusiasts from various backgrounds to participate in our user study.

Study Methodology:

Through online surveys and one-on-one interviews, we collected feedback on the furry art pieces generated by SeaArt-Furry-XL-1.0. Participants were asked to create images using the model based on specific prompts and to evaluate the images in terms of quality, style alignment, and inspiration for creation.

Key Findings:

• Highly Personalized Creation: Users generally found that SeaArt-Furry-XL-1.0 offers a highly personalized creation experience, capable of generating images that meet individual preferences based on very specific prompts.
• Enhancement of Artistic Quality: Most users noted that using high-quality prompts like "masterpiece" significantly enhanced the artistic quality of their works.
• Source of Inspiration: Many artists and creators reported that the model not only expedited the creation process but also provided new sources of inspiration for their work.

Showcase of User Creations:

In the study, we collected several outstanding works created by participants to showcase the diverse applications and creative potential of SeaArt-Furry-XL-1.0.

Conclusion:

SeaArt-Furry-XL-1.0 has proven to be a powerful tool, offering endless possibilities for the furry art creation community. We will continue to collect user feedback and optimize the model to better serve artists and creators.

License

SeaArt-Furry-XL-1.0 falls under Fair AI Public License 1.0-SD license, which is compatible with Stable Diffusion models’ license. Key points:

1. Modification Sharing: If you modify SeaArt-Furry-XL-1.0, you must share both your changes and the original license.
2. Source Code Accessibility: If your modified version is network-accessible, provide a way (like a download link) for others to get the source code. This applies to derived models too.
3. Distribution Terms: Any distribution must be under this license or another with similar rules.
4. Compliance: Non-compliance must be fixed within 30 days to avoid license termination, emphasizing transparency and adherence to open-source values. The choice of this license aims to keep SeaArt-Furry-XL-1.0 open and modifiable, aligning with open source community spirit. It protects contributors and users, encouraging a collaborative, ethical open-source community. This ensures the model not only benefits from communal input but also respects open-source development freedoms.

Finally

We welcome and value your feedback, looking forward to your suggestions to help us continuously optimize and improve. Moving forward, we will keep introducing a variety of models, so stay tuned for our latest developments.