Learn Face Swap in 3 Mins | Fooocus UI - Stable Diffusion

hey there so I was experimenting with this pretty cool AI workflow and I thought I would share this with you my goal is to get a stock image let's call it a stock image of a man a photo shoot I want to have this face representing from this particular portrait and I want to use this particular pose from this Photograph right so this face with this particular pose so how do I go about it let's get to it right I'll not bore you with any details in here so I'm leveraging Focus which is a stable diffusion UI I'm running it on the cloud with RTX 3090 Nvidia graphics card what I'm doing is I'm leveraging first of all this particular portrait image I'm using it as an image prompt right so here I have image prompt selected on Focus then I have selected the face swap because I want to swap this particular face in the final image and I have set the weight to 1.5 I'm experimenting with one uh below one and 1.7 I'm seeing which one works the best so right now it's at 1.5 then I have also included the image where I want to copy the post from after I upload this image I have selected the panny which is again a control net model which uses the post of the particular image and I have set up the weight to one all right so once that is done I have input my prompt which you see over here is a six a 60-year-old man with a rugged face with a slight tan Gray ha and a beard wearing a white shirt and a black tie is holding papers and looking at the camera right he's standing on a street corner depth of field park car and light in the background so even though I have these reference images I have to describe them in the prompt quite well just in terms of the pose what the subject is plus the background details right so once I do that I have to also select the style which is the focus photography and then the model which is Juggernaut XL by the way Juggernaut XL is one of the latest AI models out there to create really high quality highly realistic portrait photos uh I have tested it multiple times and in fact uh in many of the results it's even better than mid Journey so I love using this and once I have those selected I can set up the aspect ratio 1 is to one and yeah then I'm good to go so I'm going to hit [Music] generate all right so there we have it I was just timing it and it took us around 50 seconds for all of these four generations so yeah it's pretty fast and it's pretty high quality and as you see over here we are very close to what we wanted to achieve I mean some of the results are not that great here I see white gloves but I think most of the images are pretty cool and I can take these images as raw images to then use an a enhancer like Kaa right and then turn it into a much more high quality image from here so imagine all the possibilities that uh this workflow has to offer and yeah give it a try

Channel: AI Lemon Academy

Views: 707

Keywords:

Id: O3c5D645O8w

Length: 3min 11sec (191 seconds)

Published: Sat Mar 09 2024