Create AI Art of Your Friends with Dreambooth Tutorial

Create AI Art of Your Friends with Dreambooth Tutorial

Table of Contents

1. Introduction
2. Collecting Photos
3. Setting Up Photoshop
4. Creating a Canvas
5. Sizing and Layering the Photos
6. Exporting Layers to Files
7. Using Google Collab Sheet
8. Connecting to Google Drive
9. Uploading Images
10. Starting the Training
11. Downloading the Script
12. Converting to CKPT File
13. Using Stable Diffusion UI
14. Copying and Renaming the CKPT File
15. Running Stable Diffusion
16. Result Examples
17. Conclusion

Introduction

Welcome to this tutorial on how to make custom models in Dream Booth for stable diffusion. In this guide, we will walk You through the process of creating your own custom models using Dream Booth.

Collecting Photos

The first step in creating a custom model is to Gather a collection of photos of the subject you want to model. Make sure to capture different angles, lighting conditions, and expressions to ensure accuracy in the final rendering.

Setting Up Photoshop

Once you have the photos, open Photoshop and Create a canvas-sized image that is 512 by 512 pixels. This will be the base for your custom model.

Creating a Canvas

In Photoshop, create a canvas-sized image that is 512 by 512 pixels. This will serve as the base for your custom model.

Sizing and Layering the Photos

Next, import each photo onto a separate layer in Photoshop and arrange them according to your preferences. Include different angles, close-ups of the face, and variations in lighting and expressions.

Exporting Layers to Files

After arranging the photos, export each layer as a separate file. This can be done by going to Edit > File Export > Layers to File. This will generate individual image files for each layer of your custom model.

Using Google Collab Sheet

To Continue the process, you'll need to use the Google Collab Sheet. Create an account and join the Dream Booth project. Copy your access token from the settings and tokens section.

Connecting to Google Drive

In the Google Collab Sheet, run the provided code snippets one by one. You will be prompted to connect your Google Drive and input your access token.

Uploading Images

Once connected, upload the image files you created in Photoshop to the Google Drive. It is recommended to have an empty Google Drive as the files generated during the process can be large.

Starting the Training

After uploading the image files, initiate the training process. Set the prompt using the name you assigned to your custom model, prefixed with the category (e.g., Co001 - Men). Adjust the learning rate to 1E6 and set the training steps to around 2000.

Downloading the Script

Once the training is completed, download the script provided. This script will convert your model to a CKPT file, which can be used in Dream Booth.

Converting to CKPT File

Run the script to convert your trained model to a CKPT file. This file will be saved in your Google Drive. Locate the CKPT file following the given path.

Using Stable Diffusion UI

Move to the Stable Diffusion UI and locate the custom model CKPT file. Replace the existing file with your custom model by copying and renaming it.

Running Stable Diffusion

In the Stable Diffusion UI, select the option to use the custom model. Input the prompt using the name and category of your custom model. Run the process to generate images Based on your custom model.

Result Examples

Here are a few examples of images generated using the custom model. These examples showcase the accuracy and resemblance achieved through the custom modeling process.

Conclusion

Congratulations! You have successfully learned how to create custom models in Dream Booth for stable diffusion. Experiment with different subjects and settings to create unique and personalized models. Enjoy the power of custom modeling in Dream Booth!