Sign In

万象熔炉 | Anything XL

22.3k
353k
1.3m
2.9k
Updated: Oct 5, 2024
base modelanime
Verified:
SafeTensor
Type
Checkpoint Merge
Stats
75,253
574,876
Reviews
Published
Mar 10, 2024
Base Model
SDXL 1.0
Hash
AutoV2
8421598E93
Tsundere Lover
Yuno779's Avatar
Yuno779

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

本系列模型及衍生模型,禁止上传LiblibAI或ShakkerAI

This series of models and their derivatives are prohibited from being uploaded to LiblibAI or ShakkerAI

このシリーズのモデルおよびその派生モデルは、LiblibAIまたはShakkerAIにアップロードすることは禁止されています

—————————————————

AI art should be looked like AI, not like humans.

—————————————————

Anything-XL

AnythingXL β 4 is the AnythingXL here. If you download the previous model, there is no need to download it again.

AnythingXLbeta4 is another product encouraged by friends in the group. The premise is the highly developed SDXL model, with the quality of the trained model getting higher and higher. The negative hint word in the display graph's embedding name is only copied, and it actually has no effect. Beta4 was created because the model with the fourth test version was the only one I could tolerate looking at. Model fusion is a dead end, with a complete mess in terms of artistic style diversity and the accuracy of some prompt words.

Formula:

aingdiffusionXL_V0.6       x 0.144375
animagineXLV3              x 0.144375
cutecore_xl                x 0.12375
kohakuXLDelta_rev1         x 0.1375
BAXLBArtstyleXLv2          x 0.3375
ponyV6                     x 0.1125

Model order only represents fusion order and has nothing to do with model quality

Model merge is implemented using the Webui-Supermerge plugin. According to FairAIPublicLicense1.0-SD, the recipe needs to be publicly disclosed, and any merged models that merged this model must also disclose the merge recipe in accordance with this license.

Parameters+:

Prompt words are different from SD1.5, and for best results, it is recommended to follow a structured prompt template:

<|special|>, 
<|artist|>, 
<|special(optional)|>, 
<|characters name|>, <|copyrights|>, 
<|quality|>, <|meta|>, <|rating|>,……

<|tags|>, 

special(optional):These prompt words only need to be typed once, put in the front, there is no need to put in the back

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.

years:

These words help guide the results towards modern and retro anime art styles, with a specific timeframe of approximately 2005 to 2023

newest	        2021 to 2024
recent	        2018 to 2020
mid	            2015 to 2017
early	        2011 to 2014
old             2005 to 2010

NSFW:

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.

safe	            General
sensitive	        Sensitive
nsfw	            Questionable
explicit, nsfw	    Explicit

quality:

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

masterpiece	            > 95%
best quality	        > ?
great quality	        > ?
good quality	        > ?
normal quality	        > ?
low quality	            > ?
worst quality	        ≤ 10%

Resolution:

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:

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).

Because of models merge, some labels in the original model that have not been fully trained may be lost, and some labels may need to have a weight of over 1.5 in order to be effective.

Resolution:

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
2048 x 2048     not recommended
……
1280 x 2048     
1280 x 1536     
960  x 1536     Recommended
1024 x 1024	    1:1   Square
…… 
960  x 640                
768  x 512      SD1.5
……
2048 x 512      ¿  Unable to guarantee the quality
512  x 2048     ¿  Unable to guarantee the quality

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:

Anything now uses the Fair AI Public License 1.0-SD, compatible with Stable Diffusion models.All versions of the models in the Anything series are open source using this protocol. Key points:

  • Modification Sharing: If you modify Anything, 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.

—————————————————

您所看到的是单独给日本和中国AI玩家的备注:

①除非需要生成2048x2048以上的图或者遇到严重问题,否则请放弃高清修复。

②如果你感觉模型偏向某方面,请先检查提示词。一些提示词在之前使用的模型上可能无效,在这里可能有效。

③请不要用SD1.5的使用习惯来使用SDXL,因为两个模型本质上不同。如有必要,请使用简介中的质量词,而不是8k高清等。

④最好不要和NegativeXL一起使用,也不建议使用高权重的负面提示词,如(ugly:2)。

⑤想要一张好图片,请尽可能详细描述内容,而不要仅标注"1girl, nsfw",这样无法得到好的图片。

あなたが見ているのは、日本と中国のAIプレイヤーへの単独の備考です:

①2048x2048以上の画像を生成する必要があるか、深刻な問題が発生している場合を除き、高画質修復をやめてください。

②モデルが特定の側面に偏っている場合や他の問題がある場合は、ヒントワードを確認してください。一部のヒントワードは以前のモデルでは機能しないかもしれませんが、ここでは有効になる可能性があります。

③SD1.5の使用法ではなく、SDXLの使用法は異なるため、注意してください。必要な場合は、8kの高画質などではなく、以下に示す品質の言葉を使用してください。

④NegativeXLとの組み合わせは避けるべきであり、(ugly:2)のような大きなウェイトのネガティブなヒントワードの使用もおすすめしません。

⑤良い画像が必要な場合は、内容を可能な限り詳細に説明してください。単に「1girl, nsfw」というラベルを付けるだけでは良い画像が得られません。