hey why is it only showing 1024 MB or VRAM. I have RX 6650XT 8gb of VRAM **Using directml with device: Total VRAM 1024 MB, total RAM 16307 MB Set vram state to: NORMAL_VRAM**
this tutorial assumes you're already running Python. if you don't, you need to install it and add pip to the environment variables. took me half an hour to realize that.
In case you're having issues with the CMD part at 2:56, make sure that you have installed Python 3.11.5, NOT 3.12. And when installing it, on the first step, make sure to add it to PATH. Then it will work.
Finally!! Someone who can translate tech-instructions to those of us who want to use the technology but cannot/ and do not understand those who speak "tech-splanation" - ("...the act of explaining tech using tech language to those of us who have little or no technical background!") Thank you for "putting the hay down so the horse can reach it!" Great Job! you now have a new Subscriber!
Just a note, another REQUIRED software is Python. I didn't take the time to figure out how to start the embedded (with CloudUI) Python version or add it to the Windows PATH. So I was failing on the PIP commands. Great video, I've used it as a reference multiple times now. This time, I have uninstalled everything and am starting fresh. So if you get the messages below, install Python and click the "Add to Path" on the initial install screen 🙂 Error Messages: 'pip' is not recognized as an internal or external command, operable program or batch file.
With the above being said, all my problems with loading ControlNets and IPAdapter-plus went away, after I installed ComfyUI without Python being installed and I skipped the above PIP command to install the required text listing. Still a great video highly recommended.
I went to Python website and downloaded + installed 3.11. But I still get the "pip' is not recognized as an internal or external command" - I am LOSING it!! What did you do?
I have a new installation, without custom nodes, just the manager installed and the workflow by default and I have a delay of more than 10 seconds before the ksampler works and another before the Vae decoder works... and I didn't find help anywhere... RTX3060Ti 32GbRam.
@MonzonMedia I was able to get everything perfectly going. Until it was time to Install Custom Nodes through the Manager. Seems I can't install any of the custom nodes "fail / impact / and tested others" Errors comes up asking to Restart and Refresh... No avail. Files not showing in Custom Nodes folder. Any work around or solves for this? Any insight would be truly helpful as a little out of my element getting into cmd prmpt window / python or trying to solve this. Thank you!
@@hasanozcan6659 I deleted from C Drive Location. Re Installed per instructions. Everything worked up until it came to typing Fail and Impact. Attempting to use the Manager to Install I was now given this : error: Bad Request So I just went to URL and installed manually to local folder. Which does defeat the purpose of the Manager I suppose. So same errors no response from this tutorial creator. However this method works and I picked right back up on tutorial with Force Snap to Grid showing up in settings. Hope that helps
I wanted to install Comfy UI but it failed because I don't have a NVidia or AMD graphic card on my HP laptop but an Intel iRis XE. Is there a workaround to use Comfy with this graphic card or is it impossible to use Comfy with my configuration?
I don't believe intel is supported. Most AI platforms require NVidia GPU's since they have cuda cores, even AMD isn't optimized really for most of these platforms.
Hi. You are explaining very clear but i have a problem. I can't install nvidia. When i paste and run the commend on cmd, It doesn't run. I receive ( 'pip' isn't recognised ). What can i do?
Very nice tutorial. Thank you! I'll be using ComfyUI to run AnimateDiff. Currently, I'm using Automatic1111, but it seems that ComfyUI is a better option for AnimateDiff. Also, it's good to know how to use more than one Stable Diffusion UI.
@@MonzonMedia Currently, I'm creating a brief RU-vid installation tutorial for AnimateDiff. After that, I'll dive deeper into ComfyUI and AnimateDiff. I'm enjoying it so far :)
Hi, thanks for the tutorial! Which Python version are you using? I remember that for automatic1111 it was up to 3.10, and I see that for comfyUI 3.11 is recommended. I myself am using python 3.10 and I'm not sure if that's having an impact on the generation speed or not. The thing is I still want to be able to use automatic1111.
Assuming you followed the directions exactly, check the trouble shooting guide near the bottom here github.com/ltdrdata/ComfyUI-Manager. I did recall a situation for myself where the manager was just not showing up even though I was using it already. The second point helped to solve my issue. Worth a shot.
You’re welcome! Yes that’s the purpose of the manager however there may be occasions where you would have to do some manual installs or updates. I’ve noticed in the last 2 weeks there have been several changes to the manager.
What hardware setup do you use? On my Macbook pro 2018 / Moterey / intel / 16GB ram, 4GB vram it is really slow.. takes a minute to generate an image base don the default setup. and it also doens’t generate aany image.. it’s all blank.
Do I need Stable Diffusion to run this? Or can I just use it on its own as a standalone? I'm asking because I know that stable diffusion requires an NVIDIA GPU which I don't have.
Just to be clear this is a stable diffusion platform. Any platform would benefit more if you have an Nvidia GPU however ComfyUI can run on AMD cards as well as Mac M1 and M2. I can't speak for the performance though since I don't have either.
Thanks bud! It's good to have as some nodes may not be part of the ComfyUI manager and git is an easy way to install most stable diffusion resources in general. It doesn't take up a lot of space and once you have it, you are pretty much set.
@@MonzonMedia you're welcome! I did wind up installing Git (Oh my all the options it asks you to configure lol) and then updated ComfyUI and now have the Manager installed, as well as the Customs in the video. Also thanks for showing how to put the launcher bat file on the desktop I had copied it and for some reason it didn't work properly but the send to did. Thanks again! Have a great night!
Hey G'day, love your tutorials, calm no rush lots of detail. Have you done any tutorials on how to achieve depth maps using comfyUI? I followed your tutorials for installing that and some models from Civitai, didnt find anything much to do with depth maps there. (using depth maps for 2.5d carving/bas reliefs. Cheers and avagreatday, greetings from the colonies downunder.
I have tried to reply 2/3 times and each time my reply is scrubbed so not sure why. Looking at creating depth maps from generated AI images grey scale. The depth maps I have made so far with things like zoedepth are missing a lot of detail. The depth maps need to have much more detail, so they would be suitable for making Bass Reliefs for use with CnC machines and fiber laser engravers@@MonzonMedia
Best to check the install page on github, link is in the description. Unfortunately I don't have AMD so I can't really say for sure. Try their discussion board.
I tried various solutions (adding paths, running from different folders, with and without admin rights), what finally worked for me was opening a command window at the python_embedded folder, and running the command from the website prefixed by 'python -m'. For the next step, I copied requirements.txt into the python_embeded folder and ran the install command with the same 'python -m' prefix.
Hello my friend, yes you can run it on as little as 2-3GB Nvidia GPU. I'm not sure if there are limitations to the older 4GB cards but from what I hear from others that it runs fine on 4GB cards, obviously won't be as fast as other cards but ComfyUI is the fastest interface I've used so far. Hope that helps!
@@MonzonMediaHi, thank you for your answer and help, as always, ok I will try mine is an old NVIDIA 970 4gb, I will change it for a 3060 12 gb asap (kkkkkk, need the money) since here in Brazil all NVIDIA cards are very expensive.
Not sure about the 970 but you can always try right? I think the 3060 12GB will be enough for most scenarios. My son has one and it can run any SD platform fine, just can't do dreambooth training. GPU cards are very expensive globally, I know what you mean! When are you getting the new card?
I've not had issues with the method I show, you can try the first method manually github.com/ltdrdata/ComfyUI-Manager. Just look for the installation instructions and if you need support you can ask in their discussion page. github.com/ltdrdata/ComfyUI-Manager/discussions
@@MonzonMedia came back from vacay, all my nodes need updating and i find myself going thru them one by one and restarting comfy in between every single one...
Fuck, it was running fine before, but then I followed this tutorial and it refuses to even start. Even if I delete everything and re-extract the original file, still gave me error and can't even open the GUI.
On AMD Radeon Card - Windows 10 (FYI-Report): The Git installer ask you a LOT you can choose, dont be surprised (Open SSH, PATH environment, HTTPS backend, Git line-ending, MinTTY Terminal Emulator, Credential-Helper,) and NEW! Git Bash Profil for Windows Terminal AND what Editor you want to use...if you have VS Code you can choose VSCode as Emulator, Nano uses console Window by default. i skip all and use NANO. Comfy is 1.4 GB large. AMD should work via "cuda:0" (seen in another install tutorial) when u start compfyUI. i got an older Card - AMD Radeon 390 -, so i dont install that NVIDIA stuff. BUT you need Pytorch?? so you need to install or Skip to requirements.txt. You will run later the CPU Version until u get DirectML (its for Direct12 Cards) installed... Maybe you will GET PATH errors then. But you can maybe ignore. I do so. Installing DirectML for AMD not work for me because it not support python 3.11?? But that is from 2020 there is also directml-tensorflow-plugin, but that dont work for me also. -> i run updates bat from the updates folder -> run_cpu.bat - works for the first test. (run_invidia will not work) -> on second run after installation of the manager i got socket server issues (because CMD was open ^^,). god its a mess that is not shown here. I hope for a proper AMD Support. cheers*
Wow that's a lot to go through! Unfortunately because stable diffusion benefits from cuda cores AMD won't be a good choice for GPU at the moment. Who knows if things will change?
This reminds me of why I hate Linux. PIP doesn't work. Oh, you have to install Python. Ugh. Ok. Still doesn't work. WTF? HA! You installed the wrong revision! I appreciate the video, but I ran out of patience after a half hour of trying to get this working.
It can be a pain especially if you use multiple platforms. Have your tried Pinokio? It's a great launcher for various stable diffusion platforms and other AI things as well. Very easy to install pinokio.computer/
Tried to install Git, and there was no "install" button. Just a absolute shit load of questions about options that I didn't understand. The more I dig into Stable Diffusion, the more frustrated I get at the absolutely egregious amount of shit I have to install/learn to use it. And every video tutorial I find for it is convoluted and different. F*** this backwords bullshit art generator.
I know it can be a deep rabbit hole and frustrating but I'm here to help if I can. Are you new to stable diffusion? If you are, I encourage you to watch these 2 videos to understand some of the jargon. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-OeqpwYLMaN0.htmlsi=DOr8CLCjxVO6N_QS and ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-aI-r0C5ubGc.html&ab_channel=MonzonMedia (although I would no longer recommend Easy Diffusion as the developer is no longer updating it but still usable for now). In case you want to install ComfyUI easily, try using the pinokio installer, it does everything for you. Once in stalled you can install any stable diffusion platform and manage it through that as long as you meet the system requirements. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-FWrP8scitBI.htmlsi=toqr18ZQraZH5rfn If I'm totally in left field, give me some context and your system specs I'll try to help. 👍
What a "F"ing mess. This wonderful video is already obsolete only 2 months later. I want to build a video tutorial system that is able to update its self with updates when anything in the change changes.
What are you referring to? The installation process hasn’t changed? But yeah things change really quick in the AI space and a lot of my videos don’t age well unfortunately.
such a great uninformative tutorial video , it was sooooo easy I let my pet gecco do the installation itt only took him he she it them they 1 second to complete the install.
Hi, thank you for this tutorials! Here at the very end you edit the .yaml folder for your models! How can I find my Auto1111 models directory path? I’m on google drive with a collab notebook version if ComfyUi Thanks in advance, have a great weekend!