ComfyUI EP01: Create your first AI Image [Stable Diffusion]

[Music] foreign YouTube channel my name is Sarah I came from Thailand I created this channel because I want to share my knowledge about generative AI especially generative AI that can create a beautiful image I have experienced both my journey and stable diffusion and this clip I will teach you how to create the beautiful image with stable diffusion and using the tool called config UI please follow me I will teach you how to use the tool first you need to search the Google comfy UI and you go to the first link is GitHub link click and that link and go to the installing config UI you can click the link to download the zip file the file is quite large but it contains many things in that you don't have to download the python of any other stuff you just load this file and extract it if you extract the file you can put it in some folder for me I put the zip file in this folder and create the folder its name and unzip it when you unzip you will got this folder configure I Windows portable and you go into that folder and run this file if you have Nvidia GPU but if you want to run on your CPU you can click add another file but it will be very slow I recommend that you should have the energy at GPU and run this file if you rendify they will try to execute the program and wait a little bit this 2 is quite fast and you pop up the new window called comfy UI this window is run on your machine this is uh the IP of localhost on your machine and you will get some buff with some one note or something like this it's look kind of complex but it's not that difficult you can put the pump the problem is uh the thing that you want the eye to generate in this block this box is uh called positive prompt because it's go to positive and another box is called negative prompt because it's go to negative the positive prompt is the thing you want in the picture the negative prompt is the things you don't want in the picture if you just put it like this this is the default one and you press cute prompt the program will tell you that you still do not have the model the checkpoint model the check price Model is the very important file that you have to download and only the stable division came with the default model in this name you can download this file but this model is is not so good so I recommend that you go to another website because see whichai.com in this website you will find many beautiful models and you go to download the thing that's called checkpoint checkpoint is the thing that you need to generate the image and find the thing that you want for example I go with this model if it realism and you download the file at this blue button press download and put the file in the folder config UI models and checkpoint because you download the checkpoint if and it will download the file the model is around 2 gigabytes per model the style of the image will be different for each model some model can create anime style image some model here a beautiful Asian image some models create beautiful versatile image include European people or many things you can explore the model there and if they have finished the file and come back to configure I then refresh you can reload the page again and now you can choose the model if it realism pure Evolution this is the one I just download and you press Q prompt you will see the green box move and go to the last step this buff is the process of creating the image you can create new box and wiring stats but you don't have to do it you can use the image that created from this tools this config UI tool and drag into this config UI and you will get the workflow of that image but for now I will try to change the prompt so you understand how to create the thing that you want for a sample I will change the prompt to beautiful visualization women in red chat or something like that and you press Q prompt they will create the image based on your prompt right and you can change the width and the height of the image here we can assemble the size that many people love to use is 5 12 and 768 this one is Portrait size and you can put some prompt to make it more beautiful for example Ultra details are realistic high quality and you you from a grain and create the image you see this the result is more beautiful and you can put the negative from like uh that quality illustration because I want the realistic one I don't want the things that look like illustration image I don't want the one look like a tone uh anime and you Q Prime again okay you will get the new image and if you kill prom again or you press Ctrl enter to kill prompt again you will see that you will get different image because you see the image is changed because the seat is got randomized if you want to fix the fifth you you change the control and after generate is fixed and you fix the fifth for example one two three four the speed is the number that is the starting point of generating image if the seed is the same and the problem is the same model in the Stream if everything is the same you will get the same image for example if fish is one two three four and it's fixed and you kill prom again you will get the same image and the configure I will be smart smart enough that they will not generate New Image because it will get the same one but if you change something for example change the sampler sampler is the one that if also affects the result of image you will have to try the one that you like each model works with different sample you have to try it for example the one that I love to use is DPM plus plus sde GPU and change schedule as Karas and Q prompt if you change the sampler even the seed is the same the image will change see the image will change so this is your first step to generate AI image this box is called save image so it means that every image that you generate it will be saved and it will be saved in your config UI folder and output you see every image is saved in your folder and then convey UI is very powerful you can add new buff is called node you can add new node and you can connect and create create very complex workflow that can help you create better image that you want for the last example I will show you that if you have image that's created from comfy UI you can put it drag it in the config UI window and you will get the workflow of that image for example if you go to the confi UI GitHub and you go to config UI example and you drag something for example if you go to the sdfl example sdsl is the newest model that is just came out for example I I save this image into my computer for example I put in this folder and and then I go to at that this is image this is normal image right and but if you drag this image into the config UI you will get the workflow of that image completely this is wonderful and it's very good however if you want this workflow to to work you will have to get all the model uh the note that they want for example they want this model sdsl base and sdsl refiner you can go download sdsm model hacking face and you go to download the file it will be the best one and refiner one the best one is about seven gigabytes and refiner one is also 6 Gigabytes it's quite large however these two model is very powerful so this is the example of uh how to create your first AI generated image using configure ad please enjoy creating new AI image in the next clip I will tell you exactly how to create the floor by yourself how to connect the flow connect the node and what is the meaning of each one if you like this clip please press like And subscribe to my channel I hope you like it thank you [Music]

Channel: AI Angel Gallery

Views: 6,581

Keywords: stable diffusion, ai image, ai art, comfyui, generative ai, sd

Id: MtuG2kHq6UA

Length: 14min 3sec (843 seconds)

Published: Tue Aug 01 2023