![](https://i.imgur.com/9UQ54uc.gif"){kind=small}
**Model: SoteMix V1**
In this guide, we will create images using Stable Diffusion WebUI. For the installation of Stable Diffusion WebUI, you can refer to other guides as we assume you have already installed it.
**Stable Diffusion Models:**
First, you need to download a model that fits your needs. You can find models on websites like HuggingFace, CivitAI, and TensorART.
When downloading models, choose options like Pruned, FP16, and SafeTensors if available.
- Pruned: Trimmed version of the model required for training.
- FP16: A smaller and faster version of the model.
- If the file format is not .safetensors, there is a possibility of virus integration.
Place the model files in the Stable-diffusion folder within the model folder of the Stable Diffusion WebUI files.
**VAE:**
In most new models, you don't need to download VAE separately as a proper VAE should be included in the model.
However, in older models, there may be issues with the VAE, so check the descriptions.
If you want to use an additional VAE file:
- Place the VAE files in the VAE folder within the model folder of the Stable Diffusion WebUI files.
- Add sd_vae to the Quicksettings list in the User Interface menu of Stable Diffusion WebUI settings.
- It should look like this: sd_model_checkpoint, sd_vae
- Restart Stable Diffusion WebUI after making these changes.
**Text2Img Usage:**
*Positive:*
- Write what you want the model to draw here.
- Input tags similar to those in Shutterstock for realistic pictures or in Booru for anime style, rather than full sentences.
- The AI is trained on these tags; using full sentences may yield nonsensical results.
- The model's description will guide you on what to input; omitting these may result in poor outcomes.
*Negative:*
- Specify what you don't want the model to draw, following a similar format to the positive input.
- Ensure to follow the input guidelines provided by the model description.
For the full content, please accept third-party cookies to view the detailed information provided in your initial request.
In this guide, we will create images using Stable Diffusion WebUI. For the installation of Stable Diffusion WebUI, you can refer to other guides as we assume you have already installed it.
**Stable Diffusion Models:**
First, you need to download a model that fits your needs. You can find models on websites like HuggingFace, CivitAI, and TensorART.
When downloading models, choose options like Pruned, FP16, and SafeTensors if available.
- Pruned: Trimmed version of the model required for training.
- FP16: A smaller and faster version of the model.
- If the file format is not .safetensors, there is a possibility of virus integration.
Place the model files in the Stable-diffusion folder within the model folder of the Stable Diffusion WebUI files.
**VAE:**
In most new models, you don't need to download VAE separately as a proper VAE should be included in the model.
However, in older models, there may be issues with the VAE, so check the descriptions.
If you want to use an additional VAE file:
- Place the VAE files in the VAE folder within the model folder of the Stable Diffusion WebUI files.
- Add sd_vae to the Quicksettings list in the User Interface menu of Stable Diffusion WebUI settings.
- It should look like this: sd_model_checkpoint, sd_vae
- Restart Stable Diffusion WebUI after making these changes.
**Text2Img Usage:**
*Positive:*
- Write what you want the model to draw here.
- Input tags similar to those in Shutterstock for realistic pictures or in Booru for anime style, rather than full sentences.
- The AI is trained on these tags; using full sentences may yield nonsensical results.
- The model's description will guide you on what to input; omitting these may result in poor outcomes.
*Negative:*
- Specify what you don't want the model to draw, following a similar format to the positive input.
- Ensure to follow the input guidelines provided by the model description.
For the full content, please accept third-party cookies to view the detailed information provided in your initial request.