Fooocus Face Swap With Ease!

well hello good people today we're going to take a look at face swapping in focus and it's actually pretty easy to do and I've already pregenerated some images here and all you have to do really is let me zoom in a bit here click on input image and you're going to see various options we'll go into image prompt this is where you can drop in your reference images now I'm showing you AI generated images first because I wanted to show you that it works best with AI images but you can use real photos as well with some stipulations we're going to start with one image and at the very bottom you want to select Advanced then you'll have some options here for stop at weight here you'll see face swap so let's select that stop at is basically when it generates think of it as a percentage it's going to stop at 90% and finish the last 10% without the face swap so basically you're telling the AI focus on the face 90% of the time okay in terms of weight start with a default 75 in some cases I've had to go as high as 0.9 even at one so your mileage may vary so what I want to do here is completely change the scene but keep her face so we're going to change the background to let's say nature and mountains and she's wearing street clothes so let's say wearing a a summer dress let's give it you know a nice Sunset atmosphere let's go ahead and generate those images so let's take a look at the reference image and now let's see the new images that were made so it keeps the likeness of the original image obviously the clothing is different and the facial features could be worked on but for the most part it Honors that initial image now as I said earlier you may have to play around with the weight I would say stop at 0.9 is a good place to leave it again you'd have to experiment but for AI generated images the default tends to work really well now the other thing you can do is utilize some of the control Nets let's put in a reference image and I'm going to use canny here we'll leave it at its default settings and I'm going to go ahead and generate four more images and you can see as this generates it's adopting the same pose as the reference image now the way canny works it takes the edges and outlines of the image and uses that as information if we look at the first image generated here the hands look okay this one can use some work but as you see it still looks like the same person Edge is a good way to get decent hands we may have to inpaint this one or we could even adjust the strength here okay let's take a look at the images not too bad we'll definitely have to touch up this hand here the face looks great it honors the original reference reference image this one the teeth is kind of funky but you get my point it works pretty well I think out of all of them this one is the best one we just have to touch up some areas there now you could put multiple reference images to help shape the image even more now how does it work with real photos well I'm going to use some stock photos here and we're going to use this one here again we're just going to leave it at its default settings and I'm just going to put a simple prompt here photo of aand he's just wearing jeans and a t-shirt but here we're going to put him in let's say a purple suit and uh instead of Nature and mountains let's put him inside a cathedral I don't know why that came to me let's just put cinematic and for the model I'm going to leave it with realistic stock photo this is another thing you want to consider when you're using a real photo you might want to start with the base model because custom models tend to have a default look and it may not look like your reference image realistic stock photo I find works pretty well and then in terms of the Styles you might want to experiment with these I find the default ones work pretty good so we're going to go ahead and generate this now and see what we come up with all right let's take one more look at our reference image here and check out our men in a purple suit so as you can see with a face it does resemble him quite a bit not 100% exactly and that's why I was telling you with certain models so you might have better luck with the default sdxl model but all in all not bad using this one now I will tell you for certain people like myself for whatever reason it just doesn't pick up my likeness it tends to make me an older Chinese man which is fine but I'm Filipino and I'm not that old now in this case this is another stock photo that I tried and for the most part it looks like the reference image so I think for some of you unfortunately it's not going to pick up your likeness and others it will but for some reason if you're like me and it's not working don't take it personal there is sort of a bias sometimes when it comes to AI image generation and if you happen to be new to focus make sure to check out this video on how to install it it's super simple but until next one my friends I'll see you when I see you

Channel: Monzon Media

Views: 105,746

Keywords: fooocus, face swapping, image prompt, ai generated images, fooocus tutorial, fooocus for beginners, fooocus tutorial for beginners, ai generated images free, face swap tutorial, free image generator, stable diffusion, sdxl, sdxl 1.0, stable diffusion ai

Id: PaGrBsD2aNM

Length: 5min 28sec (328 seconds)

Published: Sun Dec 10 2023