Stable Diffusion ReActor Face Swap In ComfyUI (Tutorial Guide)

hey everyone in today's tutorial we're going to walk through the stable diffusion face swap custom note in comfy UI with the easy implementation of comfy ui's custom note you'll be able to create face swap images and animations in comfy UI quickly as we showed in our previous video we demonstrated reactor face swap in automatic 1111 for stable diffusion and fortunately there's also a reactor face swap version in comfi let's start with a clean slate and a default text to image workflow this one is incredibly easy to use for this demo we're using dream shaper checkpoint models We'll add the face swap custom node after the vae decode now let's try writing a text prompt showing people in the image for this demo we won't be using additional settings or negative prompts as you can see the comfy we text to image works just fine now we're going to add the face swap custom node to do this doubleclick and search for reactor face swap this is the custom note we need to work with face swap in comy UI I'll rearrange the placement connecting VA code as the input image and the image output to the save image node one more thing we'll create a load image node for the source image in reactor face Swap this will be the source of our face image such as our AI models face image one important point I forgot to mention is that this custom node can be easily installed using the KFI UI manager you don't need any additional steps or skill set to install it just search for reactor and you'll find it its custom node ID is 115 and you can search this ID number as well simply click install wait for the download to finish and then restart your comfy UI simple and easy right that's why I like comy UI now let's get back to the workflow in the reactor node there are settings similar to the version in automatic 1111 you can choose the face detection method and in this video I'll try out each method to give you an idea of which detection method is suitable for you everything seems ready to go so I'll hit generate and see if we can get a smooth process you can observe the command prompt window with the reactor node analyzing The Source image and Target image here's the first example using face detection method rest net 50 let's generate one more time H it looks a bit cartoonish with this method now let's try Yola 5i looks different I'm going to experiment with other settings as well we'll use DPM 2m SD e with Caris Schuler in the sampler node and try realistic Vision 5.1 for more realistic style face swap images after queuing the prompt we'll wait for the result much better she's back now let's try one more time with a batch size of four you can also set the gender detection in the face swap node which works similarly to the automatic 1111 extension generate one more time with gender detection enabled nice the third one isn't great it got a bad hand but the other three images look pretty good even without much optimization it can still generate good natural faces so that's it for this tutorial face swapping using comfy UI is easy to set up and Implement into existing workflows in our upcoming video we'll use comfi to create cool animation videos along with face swap AI models stay tuned for our next tutorial don't forget to subscribe to our Channel if you haven't already and follow us for more tutorials and fun AI projects until then I'll see you in the next video have a fantastic day

Channel: Future Thinker @Benji

Views: 13,928

Keywords: stable diffusion, stable diffusion tutorial, comfyui, comfyui tutorial, ai art, stable diffusion prompt guide, stable diffusion img2img, install stable diffusion tutorial, face swap, deepfake, stable diffusion comfyui, comfyui custom nodes

Id: q848vXgJGJY

Length: 7min 20sec (440 seconds)

Published: Fri Oct 13 2023