Stable Diffusion 3 in ComfyUI and Workflow

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
hey everyone welcome back to the channel in today's video we are going to talk about stable diffusion 3 we'll talk a little bit about what is a stable diffusion 3 we'll talk about the API I will show you how to set it up inside of confy UI and start generating images all right so a little bit about stable diffusion 3 it is the latest model released by stability AI it's the company behind estable diffusion and they release stable diffusion 1.4 1.5 2.0 2.1 sdxl stable Cascade a stable video diffusion and now we have the stable diffusion three now if we go into this blog post here it tells us that the main characteristic of stable diffusion 3 is that it can understand text prompt better than previous generation it can also do multi subject props this is something that previous generation struggles the image quality is higher it can also do text so we have a couple of example here we can see that it says go big or go home this is pretty much generated inside stable diffusion 3 and we have a stable diffusion 3 here written as paper there's also text here now if you go at the top this picture was generated using stable diffusion 3 so you can see how good it is at generating text now at the top here you will see that we have a stable diffusion 3 is now available on on our developer platform APR and then there's a learn more if you click on it it will open basically a documentation page the developer platform by EST stability Ai and you have to be careful at the top it's mainly for estable image core that is the primary model that they are offering with the API to use stable diffusion 3 you need to click on where it says generate stable diffusion 3 or you can scroll down until you see stable diffusion 3 now where it says try it out there is also a grab your API key click on where it says API key it will open EST stability AI page if you not already logged in it will ask you to create an account and log in once you've logged in it will bring you here to this API Keys section and then at the top here you will get create API key click on it and you will get something like this this is going to be your API key and then you can view it you can copy it from here if if you go into billing you will be able to see how many credits you have remaining and right now you will get 25 credits for free the moment you use those 25 credits you can go into billing and then you can buy for $10 you will get 1,000 credits so if you go back to the stability AI developer platform page we click on pricing will bring you down to the pricing section and for stable image Services you will see that sd3 will cost you 6.5 credits per successful image okay so make sure you have your API key created and then I will have another Link in the description below it's going to be for the confy stable diffusion 3 API just click on the link it will open the GitHub repository in here there will be a green button click on that green button and then there will be like a link https link click on the copy URL to clipboard next go to your confy y installation folder click on confy Y and then look for the custom nde folder go into it click at the top here and type in CMD this will open a terminal or CMD Powershell command prop anything that you have valuable and it will bring you inside the confy UI custom node folder here type in get clone and then paste the link it should be the GitHub link for the confy Y stabil diffusion 3 API press enter is going to copy the repository from GitHub to your local machine now start comy UI in any way you're comfortable I'm going to use the terminal and I will do the run Nvidia GPU now once confi starts make sure that under the import times for custom node you see the stable diffusion 3 API and on the left side there is new import fail notification there now using it is actually quite simple you just need to right click on an empty canvas go into sd3 is stable diffusion 3 and that's pretty much it that you have where it says image just drag out and you can have a save image node or a preview node we just have this aspect ratio if you choose the one by one ratio it's going to give you a 1024x 1024 image and if you choose any of the other resolution it will stretch it and give you the different aspect ratio now the mode you have two options two ways of interacting with the API text to image image to image by changing it here and in the input image drag out and select AUD image now the model sd3 that is the model that we are using but if you want to use the turbo version this one is going to cost you about four credits per image generation we got the seed the strength and that's basically it okay so I'm going to copy this promp here and I will try it inside of stable diffusion 3 I'm going to do a text to image and this isn't going to work because remember we have not entered our API key yet so if you click on the Q prop you will get something that says 401 error basically when it's connecting there was an error and we just need to supply our API key so go back into your file explorer most likely you are under the custom node folder if not go into the custom nodes folder look for the config stable diffusion 3 API folder go into it there will be a config.js file open this using notepad or any text edit up I'm going to use notepad and at the top you will see EST stability key and then it will tell you your key there all right so go back to stability. a where you have your API key copy it and then paste it inside the config.js file and then save it basically it's going to look something like this there will be some characters numbers and you need to make sure that you are pasting your API key inside the double quotation mark make sure there are no spaces inside okay so once you have your API key make sure to save the file and go back into confi click on the Q promp and here we go we got an image now remember before this I had 19.9 credits so we'll check the credit as well and my prompt was basically like the anime all style Mountain stuff and then we had the text stable diffusion three you can see that it got a stable defusion and then we have another diffusion number three there otherwise it's quite interesting let's check the credit so I had 19.9 minus 13.4 which is what I have remaining and that is exactly 6.5 credits for one imag generation and you can see like for example let's make as if this is not good enough for me to use in any of my project this basically like the text here duplication of the text is not good enough despite that I am losing my 6.5 credit of course if I try again it's possible that I'm going to get a good result and you can see that this particular image was made with this exact same prompt now I would like to try an image to image as well so let me change the mode to image to image take the image out use the load image let me try this the strength is probably the denoising factor but right now I'm going to leave everything as it is and let's just give it a try so it can do image to image it can do text to image there's also the turbo version okay so even the image to image took 6.5 credits and I probably have one more generation that I can do before I need to purchase credits so you are going to burn the credit the $10 credit quite fast if you use the normal aspect ratio you're going to get about so 1,000 divided by 6.5 around I would say 150 153 generation I think now one thing I forgot to mention is that stable diffusion 3 is basically like a collection of models right now we have access to two models we have the sd3 base model and then the SD turbo model however in the blog post it also mentioned that stable diffusion 3 is a suite of models ranging from 800 million parameters to 8 billion parameters so probably we are going to to see more models so that's pretty much it for stable diffusion 3 with that I would like to say a big thank you to my patreon supporters thank you for watching until the very end I hope you have a great day happy experimenting and I will see you in the next one
Info
Channel: Code Crafters Corner
Views: 840
Rating: undefined out of 5
Keywords: comfyui, Code Crafters Corner, CodeCraftersCorner, ComfyUI, Stable Diffusion, Stable Diffusion 3, API
Id: g5rLesSLYBY
Channel Id: undefined
Length: 9min 19sec (559 seconds)
Published: Mon Apr 29 2024
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.