How to Change the Models Path in ComfyUI: Step-by-Step Guide

[@pixaroma]

Join the conversation on Discord discord.gg/gggpkVgBf3 or in our Facebook group facebook.com/groups/pixaromacommunity. You can now support the channel and unlock exclusive perks by becoming a member: kzbin.info/door/mMbwA-s3GZDKVzGZ-kPwaQjoin

[@Kreative3DMedia]

Your content is one step above others, every single time

[@2008spoonman]

Dare I say, two steps? Yes, two steps 😊👍🏼

[@ZergRadio]

@@2008spoonman wrote "Dare I say, two steps? Yes, two steps " This is absolutely not TRUE! It is three Dimensions above the rest

[@GenoG]

How do you always know what I need??!! I have tried to do this 2 or 3 times for Comfy and I just couldn't quite get it right, so thank you for the demo yaml file to work from. Worked perfectly first time!! I've been doing AI art for about a year and I have HUNDREDS of custom checkpoint merges, so being able to point all of the UI platforms: A1111, Forge, ComfyUI, etc to one repository is crutial. Here is a tech tip from me: Anyone who wants to use an external drive to store their models/checkpoints, do yourself a favor and get a SolidStateDrive (SSD) as opposed to a slower rotating HDD... From the SSD, model load times aren't too bad at all, a little slower than C: Drive, but acceptable. From an HDD, we're talking minutes to just load the model!! Again, thank you Pix!!

[@pixaroma]

Glad it helped ☺️ and I use SSD also for a few years now since everything works faster, windows start faster, Photoshop scratch disks,, etc

[@SebAnt]

I didn’t think I needed this, but ive reinstalled comfy using different paths, so have duplicate models

[@ausbildungbew2189]

Super helpful. I am always ctrl x - ctrl v ing my models from app to app

[@nomorejustice]

great as always man!

[@priroad]

Great! Thank you for your technic . working speed up!

[@advaitchougule242]

I can tell 100% sure that only you and MDMZ (mostly pixaroma has helped) has made me understand the comfyui. One request, can you make video series on animatediff or basically animated ai videos with multiple loras included please And can you check the videos made Mickmumpitz, specifically the "consistent character workflow video" he gave viewers the workflow file but don't explain the workflow in detail. So I feel you can explain it much better.

[@pixaroma]

I plan to get to video once i finish with image basics, i still need to cover upscaling, inpainting and so on

[@Fayrus_Fuma]

Arhi useful stuff, thanks! What do you do your animations with?

[@pixaroma]

With klingai.com i use image to video

[@Fayrus_Fuma]

@@pixaroma That's weird. I try to try, I register, enter my password and nothing happens. It won't load any further.

[@pixaroma]

Maybe they had a problem when you tried. I use it for a while now and it works ok for me

[@Fayrus_Fuma]

@@pixaroma Something's wrong. I tried just logging in, same result. I disabled all ad blockers - still no. After Sign In, simply nothing happens.

[@pixaroma]

@@Fayrus_Fuma i tried just now, logged out, and logged in again with email and pass, it asked to drag a puzzle piece. And it logged in. I have also adblock on so didnt affect that. I am using chrome. Try different browsers maybe

[@KimikimiYKim]

I really like your video a lot of things I learned from your video and it's easy to understand and very simple and good. And I'm waiting for you to upload a video tutorial impainting how ya get a good inpaint in comfyui like the one you get in automatic 1111 like only masked in the img2img menu. Because in comfyui I use inpaint right really is very bad and doesn't fit the expetation

[@pixaroma]

Thanks, yeah I will look at ways of inpainting and see what i can find

[@Fayrus_Fuma]

I have another suggestion for you to make a guide on how to install and work Regional Promter for Comfy.

[@pixaroma]

Thanks, i added it to the list, I made a list of what people recommend and I will see how i can include them in videos

[@Fayrus_Fuma]

@@pixaroma Can you tell me how on Comfy to see how many tokens are used in Promt? I can't find it anywhere. I discovered Apply ControlNet (Advanced). in Comfy. There is a trouble - always need to process faces separately, always disgusting quality turns out (if you do not use separate Lore on the characters). Some things it makes more detailed and some give more details in 1111 and Regional Promter. This is not yet to change and accept?

[@pixaroma]

You have a node called token counter in this custom node github.com/JaredTherriault/ComfyUI-JNodes as for faces not understand exactly what the problem is, what preprocessor ks using how strong is the control net, there are many settings in place

[@PatrickLavender]

Excellent! It works for models just fine. But I'm having some issues with the Florence LLMs that the Florence node downloads when it searches and can't find them. So I can move all those LLM sudirectories and their contents to my central directory/LLM subdirectory just fine... It's just that when I do that, and then they are gone from the Comfy/LLM subdirectory, and then I run that Florence node in Comfy... Now I could be wrong, but it seems like that node is programmed to look in the local Comfy LLM subdirectory first... and as soon as it can't find the LLM model you have selected, its next step is to download it. So I wind up with duplicate LLMs. Have you experienced this also? I could be wrong about that -- it could just be I messed up a step in configuring it...

[@pixaroma]

I am not sure, you can try to post in pixaroma discord server there is the comfyui channel maybe someone can help, we have some members that can understand it better on how codding, nodes works

[@b22091]

You are creating the best tutorial videos on Comfy I've found on youTube, but..... SLOW DOWN!!!. You sometimes begin to rush yourself and the viewer through the material like you have a 20 slide PowerPoint deck to get through in 90 seconds. You made such great stufff, take your time in describing and showing it, and let the video run a few minutes longer.

[@zaselimgamingvideos]

What if i only want to change checkpoint and lora path? I use both A1111 and Comfyui. And my main checkpoint and lora folder is in A1111. So i want comfyui use/load all the stuff from its defualt path but checkpoints and lora from my A1111's checkpoints and lora folder. In other UI paths I only changed the bas path to my A1111 models and the only changed the checkpoints folder and ran the bat file it i did't load other stuff but only the checkpoints. Edit: I think i fixed that, by adding the #comfyui path. And it seems to be fine and no error in the dos window.

[@pixaroma]

for me it worked even if I added all folders three and it still saw them both in original folder and on the path folder, so i can put in either folder and can see it.

[@AnnieDalgaux-s6m]

Hi, I moved my comfyUI folder to an SD card, but now my infightface and FaceID modules won't work anymore... Anyone had the same problem? I'm trying to move the IP adapters folders alongside the /comfyUI folder like it was back on my main SSD at the moment.

[@pixaroma]

I didn't get to that part with ip adapter, i know can cause some problems on some systems. Maybe try to run that dependencies bat from update folder, sorry I cant help

[@ludi5660]

Hi, thank you very much for this video. I managed to create a model folder in my disk and comfy ui can acces it. I have 3 AI that need access to this folder : Automatic1111/Stable Diffusion, StreamDiffusion and Comfy UI. Can I use the same texte file and drop it into their folders ? Or do I need to use another technique ? Thank you in advance for your help :)

[@pixaroma]

I didnt try for other, i think each interface has its own way to set up things. Or you could use symbolic links, there are many videos about that and can set the path to folders for any UI

[@ludi5660]

@@pixaroma Thank you for your answer :) do you have some keywords/KZbin channel so I can find what I'm looking for pls ?

[@pixaroma]

@@ludi5660 change model paths a1111, custom path a1111, things like that, or symlink creator, symlink a1111, things like that for the UI you want

[@swhite9792]

This confused me a bit because it looks like you are redirecting to a path with no models inside, correct? How does that then save space on your hard drive if you leave the files on the hard drive? I've watched the video several times and that part is not making sense to me. seems like it would make more sense to just actually MOVE the files to another place and redirect. That could be what you are doing, but you seemed to note each time that the folder is still empty....scratching my head here.

[@pixaroma]

The folder is empty because I just showed in the video where you can place the folders. Obviously, you'll add your models there. The point was, when you're using multiple UIs, like Forge and ComfyUI (I have a few ComfyUI folders), it can take up a lot of space if each UI has its own model folder. Normally, you would have the same models attached to each UI, meaning I already have the same models in 3-4 different locations. The easier way is to put them all in one place and refer to that folder.

[@swhite9792]

@@pixaroma Ok, thank you. That makes sense. I only have comfyUI...but thinking I need to upgrade the RAM on my laptop now! Digging my RTX 4090 GPU, but made the mistake of getting a laptop with soldered RAM...and it is....not working for me. I need more room. Thanks again for taking the time to reply!!!

[@Jerry-oh7ru]

mklink /D

[@pixaroma]

Symlink can give problems sometimes with certain custom nodes and in certain conditions so just to be safe i preferred this method

[@Jerry-oh7ru]

@@pixaroma learned

[@chromo48]

I use symlinks; ran into a problem when trying to update Comfy because of that. Will be switching over to this method.

[@z1mt0n1x2]

E:\Automatic1111\stable-diffusion-webui\models\stable-diffusion\ < This is where the actual models are but do not use this as the directory. E:\Automatic1111\stable-diffusion-webui\

[@pixaroma]

Sorry, i wanted to make sure people understand it, i will try to make it shorter in the future

[@z1mt0n1x2]

@@pixaroma if you want to be comprehensive and cover more ground with these near-10-minute videos, pls at least add timestamps so we can get to the relevant part :P in my case for instance, i just needed to know which file to edit, could've found it instantly if i knew where to look in the video.

[@pixaroma]

For those who are in the hurry you do know in description of this video are all the steps you need that you can read in a few seconds, the video is just getting a visual look and understanding better

[@z1mt0n1x2]

@@pixaroma Yeah it just made it more complicated. You don't "download" the extra_model_paths.yaml, it comes with the Comfy installation already. Don't know if you just throw everything into GPT and thing it's gonna solve everything. Why are you linking to your google drive when the files are available on github and the official page of ComfyUI? it's sus as hell!