How To Do Stable Diffusion Lora Training In ComfyUI (Tutorial Guide)

so in this tutorial we are going to walk through how to use this new comy UI custom node to train Laura easily now it is in the GitHub page and they have a lot of descriptions about how it works what kind of models it can train Etc and right here they have shown it's really simple you can just use one custom node from this GitHub project and it is able to train Laura they have also provided another custom node to create image captions for image preparation so these are the essential steps first we are going to install this one the Laura training comfy UI custom notes now as you can see there are a few things in this package that are going to be installed after installation you need to click restart and wait for the comy UI to restart then you'll have a fresh work flow diagram here now right click on the rightclick menu you'll see this label in the rightclick menu and then you select the luras now you will see the luras sub menu they have the load training in comy UI Laura training in comy UI and also the other one that is below those items there's an advanced mode of this Laura training comfy UI custom nodes now it's up to you whether you want to use the simple one or the advanced one of course the advanced one has lots of options that you can explore and the last one is the tensor board actually you don't need that one to be on the workflow diagrams all you need is this one custom nodes are able to do that so before that we have to install this image captioning in comy UI this is also provided by the same author of The Laura training custom nodes and what you need to do is copy this name and we go back to actually there's a very simple basic workflow for image captioning using this one but they don't provide the Json files so what I'm going to do is I will duplicate this workflow in comy UI in these videos and I will will create all these four custom nodes that are used for generating captions for a list of images that I prepared for creating Laura models now here I have the other two components that you have to install so first go to the comfy UI manager paste this name the caption and comfy UI and install this image captioning in comfy UI custom notes now after that I will also install the other one that is the WD 1.4 now this is a model that enables image processing as text prompt as you can see they have a demo page as well on their GitHub page from the GitHub page they're showing the hugging face demo you can play around with that but let's install this straight away after we've got both of those we can restart our comy UI right here so as you can see there are two custom node packages already installed in the command line and right here I will be going to the second setup step which is coming to the WD 1.4 folders and typing the command prompt CMD on the toolbar it will pop up the toolbar that is locating this folder in these custom notes and you copy that command prompt from their GitHub page to install the required dependencies components for the custom nodes now after that we will be able to use this custom node the WD 1.4 and have these model files for generating image to text prom which is workable for this workflow so once you have this right click or click the menu search menu and you can type this word WD 1.4 and you can select this custom node as well the other thing you have to do is the Laura captioning custom notes and the first thing we've got to do is load captions custom notes so Laura captions loader so we have to use this caption loader to load folders of images that we have prepared the second step we have to do is show text and finally we have show text and save those text strength lists in our destinations folder so we connect the pad connections here and then the WD 1.4 connects each other from the image text to the Laura caption save custom notes and we don't need the C text to strength duplicating in there so one is enough this is the same workflow that we can do as what they have mentioned in the official GitHub page of this project so there you go we have four custom nodes only and it allows us to create batch image captioning for these Laura preparation steps so what we have right now is that while we are waiting there will be taking some time to do all the loading on each image caption and then save those caption ions individually for each image file it will generate another text file that is for the image for example if I have 15 images that are going to be prepared for training Laura there will be another 15 text files that are correlated to those images and what happens is that in each of these text files it will have text prompts like this one so it's good that they have also shown this text prompt here so go to the file explorer here and you can see there are also the text prompts save that into the text files for each image and in the command prompt window as you can see it's also showing all those text prompt generation progress as well and save that into individual text files in the Laura training subfolder that I created so let me create this group as well so it's clear for us to know which function that is these four custom nodes are used for image captioning and then the second step we are going to do is use the Laura training in comfy UI custom nodes now I'm going to use this simple one for demo purposes you can use the advanced one for your training first you need to select your desired models for training Laura here I have tried the SD 1.5 realistic Vision 6 and right here I will copy this folder path and paste it exactly the same here so that the Laura training process is accurate for the output name you have to type your own name here for example I'm doing a fashion Laura here so I type the name and it will save the Laura files in the models Laura subfolder in your comy UI after training this is the result you will see the Laura file saved in the models Laura subfolder this custom node is really handy for us to play around with if you want to train Laura in an easy way there you have it this is like a life-saving timesaving method compared to the previous methods this is the advantage of using comfy UI so give it a try guys and I will see you in the next video have a nice day bye

Channel: Future Thinker @Benji

Views: 2,240

Keywords: Lora training, Comfy UI custom nodes, tutorial, installation process, easy Lora training, image captioning, step-by-step, simple mode, advanced mode, image-to-text prompts, WD 1.4 custom nodes, efficient workflow, tutorial video, stable diffusion, stable diffusion tutorial, ai art, install stable diffusion tutorial, stable diffusion ai, stable diffusion controlnet, comfyui, comfyui tutorial, comfyui controlnet, comfyui install

Id: vDitNjFH9To

Length: 8min 25sec (505 seconds)

Published: Wed Mar 20 2024