How to Install LORA Models for SDXL in Fooocus Under 30 Seconds

[Music] it was Wednesday evening I was sipping my coffee and enjoying my SD journey in focus when someone asked how to install l in Focus my friend you have been heard and I realized I was so engrossed in the beauty and ease of use of fukus I actually forgot to try out any of the loras that I possess so I started my journey to see how we can install L models in fukas is it really easy to use them and do they really perform just like other we let's find out if you have watched some of my previous videos you already know how this is going to turn out we are going to pay a visit to our favorite store where everything is free cv.com let's filter the model specifically for sdxl laa models I think today I'm going to try out this fify sdxl and this simatics sdxl model just download them from this link here the next is a simple game of copy and paste go to focus models ladas and bring our Luna models here now I'm going to restart the focus oi next we click on this Advanced check box here and under the model tab we can see you can select five laa models and apply them simultaneously from this drop down menu I can select any Loa models I want for starters I'm going to start with none here is an option to select the model weight which varies from minus 2 to plus2 and generally is kept at 0.5 that answers our first question can you install laa models in fukas you definitely can is it easy to use this is the easiest way I have ever used any Lura models even in automatic 1111 it is not is much easy to use and don't make me start on comy UI here you can just select the L model and adjust the slider to select the model strength that's fabulous now let us first generate an image without selecting any L models with a fixed SE I'm going to put 8.99 there just random and for the schematics model we have to use this prompt for the L to take effect I'm going to copy this paste it and change the concept with beautiful Omen rainbow hairs green eyes and generate our fast image and this is how it looks without any L model applied now let's go to model and select the schematics Lura model generate the image has certainly changed but not changed as much as we wanted let's try increasing the weight of the Lura model even at 1.05 model weight this hasn't changed the image it show with the laa model let's try increasing it a little bit more now with the model weight amped up to 1.31 this really is looking like the Lura model so that answers our final question L models work in fukas exactly like they work in automatic 1111 or confy UI let us try the second model that we downloaded here the trigger word is anen with double o's replaced by zeros I'm just going to put the trigger word in front of the prompt and again set the Lowa to none this looks a lot like the first image actually the Styles in fukus applies lot of prompts on top of our original prompts into the image therefore the generat image is almost the same as the first one because major part of the prompt is still the same and the seed is same now let us select the uifi Loa model and I'm going to keep it 1.31 from the gate go this time but the image still looks the same let's try setting the model to maximum the image has changed but it still looks nothing like the Luda model maybe in this case the Luda model works best with other types of prompts other types of inanimate objects let's try one prompt from the cvti page I'll just be copy pasting the positive and negative prompts from their page and generate the image and this time this work perfectly maybe it has to do something which kind of prompts are applicable with this Laura let me try the same prompt again without the L model applied this one also looks made of ools just like the previous image and without the L model applied clearly it is in the prompts that this image stand out this way not the other way around and the previous image where I didn't use specific prompts it could not employ the Laura model onto the image well it was really fun today if you want me to keep making these kinds of videos then a thumbs up and a subscribe would be very much helpful for this channel comment below the next topic you want me to make videos about and if you are more of a fan of automatic 1111 or comi compared to fukas then go watch these two videos to learn how you can install Loa models in automatic 1111 and conf see you on our next video

Channel: Debargha Bhattacharjee

Views: 19,088

Keywords: A.I., generative ai, generative a.i., a.i. image, a.i. art, a.i. generated art, Stable diffusion, stable diffusion tutorial, ai, generative art, ai image, FOOOCUS SDXL, lora stable diffusion, artificial intelligence

Id: 7ZMKHmrDr24

Length: 4min 2sec (242 seconds)

Published: Tue Oct 10 2023