How to Face Swap in Stable Diffusion with Roop Extension

Welcome to this comprehensive guide on using the Roop extension for face swapping in Stable Diffusion. In this tutorial, we will explore how you can create amazingly realistic images. Roop is a powerful tool that allows you to seamlessly swap faces and achieve lifelike results. Whether you're an artist, a content creator, or simply someone looking to have fun with image manipulation, this guide will provide you with steps to download and use the Roop extension in stable diffusion. More information on the roop github repo page.

Here are the requirements to follow:

• On Windows, download and install the Community Visual Studio. During the installation process, make sure to include the Python and C++ packages.
• Open your command prompt in de stable diffusion web ui directory and run the following command: pip install insightface==0.7.3
• Open Stable Diffusion and go to the "Extensions" tab of the web UI.
• In the "install from URL" tab, paste the following URL: https://github.com/s0md3v/sd-webui-roop
• When installed, Close the Stable Diffusion web UI.
• Download the inswapper_128.onnx model.
• Replace the downloaded inswapper_128.onnx model inside the <webui_dir>/models/roop/ directory.
• Start Stable Diffusion by double clicking the webui-user.bat file

Once Stable Diffusion is opened, you should see the roop expansion panel under the "txt2img" and "img2img" tabs.

Roop utilizes a two-step process for face swapping. First, the AI image is rendered completely, and then the face from the roop image is applied to the created AI image. This distinction becomes apparent during the rendering process, where the image appears finished but takes longer as the face is applied.

• Drag an image in the Roop expansion panel
• "Enable" Roop by checking the checkbox.
• Click "CodeFormer" from the Restore Face options.
• Create both a positive and negative prompt.
• Click "Generate"
• The ultimate outcome is the creation of a stunning face resembling Angelina Jolie, generated using your own input prompt.

To achieve the best results with Roop, there are a few considerations to keep in mind:

• Ensure that the face in the photo is not obstructed by objects such as hair or accessorie
• Check the Box "Restore Faces".

The face didn't get swapped? Did you click "Enable"? If you did and your command prompt console doesn't show any errors, it means roop detected that your image is either NSFW or wasn't able to detect a face at all.

In this section, we will showcase a collection of captivating and impressive stable diffusion face swap demonstrations that highlight the remarkable capabilities of this innovative technology. Prepare to be amazed as you witness the seamless transformations made possible by the Stable Diffusion Roop Extension. Get ready for an extraordinary visual experience like no other.

Example 1 (Angelina Jolie):

• Positive Prompt: perfect hand, ultra detailed, 8k, raw photo, beautiful light, perfect shadows, (face portrait:1.5), (face portrait of a young mature women:1.3), (traditional white dress:1.2), (white dress:1.2), (silk white dress:1.3), white dress with golden jewelry, red jewelries, (cleavage:1.3), long dark hair, (green eyes:1.2), big lips, red lips, burning man desert, (wearing designer hat:1.1), (desert background:1.2), burning man festival background, shim light, tight body, dynamic angle, sexy look, natural skin, posing
• Negative Prompt: (worst quality:1.2), (low quality:1.2), (lowres:1.1), (monochrome:1.1), (greyscale), multiple views, comic, sketch, (((bad anatomy))), (((deformed))), (((disfigured))), watermark, multiple_views, mutation hands, mutation fingers, extra fingers, missing fingers, watermark,
• Sampling Steps: 50
• Sampler: DPM++ 2M Karras
• CFG scale: 7
• Size: 704x392
• Model: cyberrealistic
• Clip Skip: 2

Example 2 (Salma Hayek):

• Positive Prompt: perfect hand, ultra detailed, 8k, raw photo, beautiful light, perfect shadows, medium shot of a mature latina women, (mexican traditional yellow dress:1.2), big chest, (yellow dress:1.2), (silk yellow dress:1.3), yellow dress with golden jewelry, golden jewelries, (cleavage:1.3), burning man desert, (desert background:1.2), burning man festival background, shim light, sexy look, natural skin, posing
• Negative Prompt: (worst quality:1.2), (low quality:1.2), (lowres:1.1), (monochrome:1.1), (greyscale), multiple views, comic, sketch, (((bad anatomy))), (((deformed))), (((disfigured))), watermark, multiple_views, mutation hands, mutation fingers, extra fingers, missing fingers, watermark,
• Sampling Steps: 50
• Sampler: DPM++ 2M Karras
• CFG scale: 7
• Size: 704x392
• Model: cyberrealistic
• Clip Skip: 2

Example 3 (Zendaya):

• Positive Prompt: perfect hand, ultra detailed, 8k, raw photo, beautiful light, perfect shadows, (face portrait:1.5), (face portrait of a young latina women:1.3), (silk white dress:1.3), jewelry, (cleavage:1.3), black hair, burning man desert, (desert background:1.2), burning man festival background, shim light, dynamic angle, sexy look, natural skin, posing
• Negative Prompt: (worst quality:1.2), (low quality:1.2), (lowres:1.1), (monochrome:1.1), (greyscale), multiple views, comic, sketch, (((bad anatomy))), (((deformed))), (((disfigured))), watermark, multiple_views, mutation hands, mutation fingers, extra fingers, missing fingers, watermark,
• Sampling Steps: 50
• Sampler: DPM++ 2M Karras
• CFG scale: 7
• Size: 704x392
• Model: cyberrealistic
• Clip Skip: 2

Example 4 (Emma Stone):

• Positive Prompt: perfect hand, ultra detailed, 8k, raw photo, beautiful light, perfect shadows, (face portrait:1.5), (face portrait of a young white women:1.3), (white dress:1.2), (silk white dress:1.3), jewelry, (cleavage:1.3), (red hair:1.3), (green eyes:1.5), (looking straight into the camera:1.2), burning man desert, (desert background:1.2), burning man festival background, shim light, dynamic angle, sexy look, natural skin, posing
• Negative Prompt: (worst quality:1.2), (low quality:1.2), (lowres:1.1), (monochrome:1.1), (greyscale), multiple views, comic, sketch, (((bad anatomy))), (((deformed))), (((disfigured))), watermark, multiple_views, mutation hands, mutation fingers, extra fingers, missing fingers, watermark,
• Sampling Steps: 50
• Sampler: DPM++ 2M Karras
• CFG scale: 7
• Size: 704x392
• Model: cyberrealistic
• Clip Skip: 2

If you're interested in exploring an alternative Face Swapping Extension, be sure to take a look at our comprehensive tutorial on the ReActor Face Swapping Extension. The article outlines numerous advantages when compared to the Roop Extension and demonstrates the capabilities of this alternative.

How to Face Swap in Stable Diffusion with ReActor Extension

Dive into the tutorial to unlock the full potential of the ReActor Extension.

Congratulations on learning how to use the Roop extension for face swapping in stable diffusion. By following the steps and considerations outlined in this guide, you can create amazingly realistic face swaps. The Roop extension in stable diffusion empowers you to explore your creativity and experiment with face swaps, whether for artistic purposes or simply for fun. Enjoy the process of transforming images and unlocking new possibilities with Roop!