How to use anytest with Stable Diffusion AUTOMATIC1111 and Forge

The purpose

This will use Anytest on Stable Diffusion’s AUTOMATIC1111 and Forge.

I’ve also verified that this works on an AMD GPU (DirectML) setup, which was configured in the article below.

ローカルでAMDのGPU（direct-ml）を使ってStable Diffusion Forgeを動作させる

やりたいこと通常Stable Diffusion ForgeはNvidiaのGPU（GeForceなど）が必要です。下のページではCPUで動作する環境を作成しましたが、この記事ではAMDのGPUで動作する環境を作成します。オンボードのGPU

blog.marunokan.com

What is Anytest

Anytest is a versatile ControlNet model that lets you generate images from line art or silhouettes, and even redraw specific parts of an image.

How to install anytest

Please be aware that the installation process differs between AUTOMATIC1111 and Forge.

Install ControlNet

AUTOMATIC1111

In the ‘Extensions’ tab, open the ‘Install from URL’ tab.

Enter the following URL in the ‘URL for extension’s git repository’ field and click ‘Install’.

GitHub - Mikubill/sd-webui-controlnet: WebUI extension for ControlNet

WebUI extension for ControlNet. Contribute to Mikubill/sd-webui-controlnet development by creating an account on GitHub.

github.com

If you encounter any issues installing the extension, please refer to the page below.

URL からExtensionのインストールに失敗する(Stable Diffusion)

やりたいことStable DiffusionでURL からEctensionのインストールをする際に以下のエラーが出て失敗した際の対策です。GitCommandError: Cmd('git') failed due t

blog.marunokan.com

After the installation is finished, restart Stable Diffusion.

If the installation is successful, you’ll see the ControlNet GUI appear below the Seed field after you restart.

Forge

Since ControlNet is already integrated into Forge, you don’t need to install it.

Download model

You can download the model from the following page.

The CN-anytest_v4-marged.safetensors file is recommended.

2vXpSwA7/iroiro-lora at main

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co

Install model

AUTOMATIC1111

Put the downloaded model into the following folder.

ルート\extensions\sd-webui-controlnet\models

Forge

Put the downloaded model into the following folder.

ルート\\models\ControlNet

The installation is now complete.

Use ControlNet

I’ll be using the Forge GUI for this explanation.

First, select the Stable Diffusion XL model (checkpoint).

With Forge, you should also set the UI to ‘XL’. (If you encounter a generation failure, try switching it to ‘SD’.)

Stable Diffusion XL models are usually named with ‘XL’ in them.

For example, I used bluePencilXL_v700.safetensors.

bluePencilXL_v700.safetensors · misri/bluePencilXL_v700 at 2b56ffad7e83fe68218ac6c9b0ae27624c571680

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co

Open the ControlNet menu by clicking it. (In AUTOMATIC1111, the title is ControlNet vXXXXX.)

Check the boxes for ControlNet Unit 0 and Enable.

In the Model field, select the downloaded anytest model.

(If the model isn’t showing up in the dropdown, click the refresh button on the right.)

Set the images for the following GUI by dragging and dropping them.

From there, it’s the same as a normal image generation process: enter your Prompt (and Negative Prompt) and click ‘Generate’.

Result

I successfully used AnyTest.

As the example below shows, it interprets the input quite favorably when generating images.

It also appears that the art style of the generated images is dependent on the model.

Generation Example

Input Image：

model：bluePencilXL_v700.safetensors(Intended for anime-style images)

prompt：a girl , blue skirt, white shirt, yellow hair, from front

Output：

Troubleshooting

AttributeError: ‘NoneType’ object has no attribute ‘shape’

You might get an AttributeError: 'NoneType' object has no attribute 'shape' when you run it, which causes it to fail.

This error happens when the image generation model (the checkpoint you set at the top, not ControlNet) is not an SDXL model.

Please confirm that your image generation model is compatible with SDXL.