Tasty-Rice/Magic_on_paper

Model Card for Model ID

This modelcard aims to be a base template for new models. It has been generated using this raw template.

禁止融合 禁止融合 禁止融合

Merging models is not allowed。 Merging models is not allowed。 Merging models is not allowed。

Model Details

Users of v3,v2 version can consider the scope of use by themselves but are prohibited from merging models, and are solely responsible for any problems that arise during use. I do not limit the scope of user use.

The base model is about ACGN china guofeng style. It contains a variety of Chinese elements, such as Hanfu,qipao, cheongsam(chinese clothes(qipao)), Chinese ancient architecture, Chinese ancient painting style, etc.

Model Description

• Developed by (English name): Tasty-Rice
• Developed by (Chinese name): 米饭不好吃
• Model type: [Diffusion based text to image generative model]
• Language(s) (NLP): [English]
• License: CC-BY-NC-SA-4.0
• Finetuned from model [optional]: Animagine XL 3.0

Model Sources [optional]

Users of v3,v2 version can consider the scope of use by themselves but are prohibited from merging models, and are solely responsible for any problems that arise during use. I do not limit the scope of user use.

Sample Picture

Magic_on_paper-v3

Magic_on_paper-v2

Multi Aspect Resolution

遵循这样的分辨率设置你将得到最好的体验，基于mj和stable AI的研究测试 并非所有的分辨率都支持的非常好，但遵循这样的分辨率使用你将得到更好的体验。 Following this resolution setting will give you the best experience, based on research and testing using mj and stable AI. Not all resolutions support very well, but following these resolutions will give you a better experience. This model supports generating images at the following dimensions:

Dimensions	Aspect Ratio	Dimensions	Aspect Ratio
1024 x 1024	1:1 Square
704×1408	1:2	1408×704	2:1
1472×704	23:11	704×1344	11:21
960×1024	15:16	1024×960	16:15
960×1088	15:17	1088×960	17:15
896×1088	14:17	1088×896	17:14
896×1152	7:9	1152×896	9:7
832×1152	13:18	1152×832	18:13
832×1216	13:19	1216×832	19:13
768×1280	3:5	1280×768	5:3
768×1344	4:7Vertical	1344×768	7:4Horizontal
1536×640	12:5	1600×640	5:2

Style

class	descriptors
	Monochrome
color	Colorful
	Illustration
	Cyberpunk
	Ink wash painting style
	Traditional Chinese painting style
	Fantasy Art
style	Anime

If your prompt does not emphasize the background, the background will be generated through style or solid color background, Directly generate large images with high precision, and normally do not require high-definition repair or refinement, unless a full body is required, such prompts may not necessarily cause confusion. The normal basic XL model comes with many styles. You can try to summon but the anime subject will not be affected by the concept you call

Based on Animagine XL 3.0, Magic_on_paper falls under CC-BY-NC-SA-4.0 license, which is compatible with Stable Diffusion models’ license. Key points:

1. Modification Sharing: If you modify Magic_on_paper any version, 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.

Gradio & Colab Integration

Try the demo powered by Gradio in Huggingface Spaces:

Or open the demo in Google Colab:

🧨 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(
    "Tasty-Rice/Magic_on_paper", 
    torch_dtype=torch.float16, 
    use_safetensors=True, 
)
pipe.to('cuda')

prompt = "1girl,Elegant,Portrait Photogram,detailed gorgeous face "
negative_prompt = "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"

image = pipe(
    prompt, 
    negative_prompt=negative_prompt,
    width=768,
    height=1344, 
    guidance_scale=7,
    num_inference_steps=28
).images[0]

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

Suggested parameters —— The following is a horizontal introduction, not a vertical correspondence.

Sampler	DPM++2M SDE Karras	Euler	Euler a	Others u can try
CFG Scale	7(recommend)	7.5(recommend)	8(recommend)	2~15(limit)

Notice/The content of the additional terms

Number	Terms and Conditions
1	Do not host this model or its derivatives on websites/applications that generate income or donations. For commercial use, please contact the owner through the platform.
2	Do not directly sell generated images unless they have undergone sufficient manual modification and are deemed personal works. Violators will be subject to legal consequences.
3	Do not intentionally create or share illegal or harmful content. Please abide by public order and morality, and use it for positive purposes.
4	By using this model, you agree to all terms and conditions.
5	The model is limited to personal use and cannot be used for commercial purposes.
6	Do not use the model for illegal or harmful purposes, such as infringing rights, spreading false information, etc.
7	Users shall abide by all applicable laws and regulations, including safety, data protection, etc.
8	Respect the rights of others and do not infringe the rights of third parties.
9	Disputes or lawsuits shall be borne by the user's own legal responsibility.
10	Take necessary technical and management measures to ensure data security and confidentiality.
11	Comply with intellectual property laws and do not infringe the intellectual property rights of third parties.
12	Do not exploit or harm minors, and do not involve content, activities, or services related to minors.
13	Do not generate or disseminate verifiable false information or content that harms others.
14	Do not defame, discredit, or harass others in any other way.
15	Do not make fully automated decisions that adversely affect the legitimate rights of individuals or create legal obligations.
16	Do not discriminate or harm individuals or groups, including based on social behavior or known characteristics.
17	Do not exploit the vulnerabilities of specific groups, distort related behaviors, and cause or potentially cause physical or psychological harm.
18	Do not discriminate or harm based on protected characteristics or categories.
19	Do not use for medical consultation or interpretation of medical results.
20	Do not generate or disseminate information for judicial, law enforcement, immigration, or asylum proceedings, such as predicting criminal behavior.