下载模型之前，请仔细查看模型介绍

Please review the model introduction carefully before downloading the model

モデルをダウンロードする前に、モデル紹介をよく見てください

—————————————————

I warmly welcome you to share your creations made using this model in the discussion section. If you encounter any issues, please feel free to add a comment, and both the community and I will do our best to help and provide solutions.

This is the unofficial version of abyssOrange.Original model link(SD1.5):

AbyssOrangeMix3 (AOM3) - AOM3A1B | Stable Diffusion Checkpoint | Civitai

—————————————————

dataset：

Dataset is from the following two websites, but I can't provide a complete dataset because of something all know reason.

①https://cosaag.cyou/ (base:NeverDream)

②nyanko7/danbooru2023 · Datasets at Hugging Face (base:AnythingXL)

The model is trained separately in two parts: real and animation. Merge after training.The model uses AnythingXL & NeverDream as the base.

Used MBW: Merged model loaded:① x (1-alpha)+ ② x alpha

alpha=(0.8,0.4,0.4,0.4,0.4,1.0,1.0,1.0,0.4,0.4,0.35,0.45,0.45,0.2,0.2,0.6,0.6,0.6,0.6,0.8,0)

Parameters+：

Prompt words are different from SD1.5, and for best results, it is recommended to follow a structured prompt template(Please do not include any artist name as a prompt):

1girl/1boy/1other/……, character name，special tags, another tags

Special tags：

The model can still be used without these special cue words, but incorporating these special tags when necessary can help steer the generated results towards the desired direction.

NSFW：

• nsfw(explicit)/sensitive/sensitive：

  These words help guide the results towards adult content, but generally do not generate adult content if rating words are not included.

  Of course, you can also put it in negative prompts.

quality：

• best/high/medium/normal/low/wprst quality：

  While this model can function without quality words, in practice, these words can still be used to adjust the output.

Resolution： 分辨率：

• You are free to use the vast majority of reasonable resolutions, whether it is the resolution used by SD1.5 at 512*768 or higher resolutions above 2048, each will have a different effect. However, using images that are too large or too small may cause the picture to break down or the character/background structure to become distorted.

Tags： 标签：

• If you want to generate high-quality pictures, you can use negative prompts, such as:

nsfw, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name

Negative tags can include common negative tags, but it is best not to assign too high of a weight to their content, for example (ugly:2.8).

A resolution greater than 1024×1024 is recommended, and hires fix is recommended if you want higher resolution or quality

Most of the generation parameters of the example graph are:

euler_a | 20steps | no hires fix | CFG7

Disclaimer:

All images generated by the model are created by the users themselves, and the model author cannot control the images generated by the users. The model author will not be held responsible for any potential copyright infringement or unsafe images.

License:

AbyssOrange XL now uses the Fair AI Public License 1.0-SD, compatible with Stable Diffusion models. Key points:

• Modification Sharing: If you modify AbyssOrange XL, you must share both your changes and the original license.

• 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.

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

• Compliance: Non-compliance must be fixed within 30 days to avoid license termination, emphasizing transparency and adherence toopen-sourcevalues.