fetch models folder from other drive?

[chnisar515]

Hello Experts,

I am currently facing storage constraints on my primary 1TB SSD, which is nearly full. To free up space, I’ve decided to move my Models folder for ComfyUI to my second hard drive located in the E:\ drive. I do not have any other web UI tools like A1111 installed on this drive.

Here’s what I’ve done so far:
I copied the ComfyUI Models folder from its original location on the SSD to the following path: E:\My AI\comfyui\models. Then, from the ComfyUI portable directory (C:\ComfyUI_windows_portable\ComfyUI\models), I opened Command Prompt and ran the following command:

mklink /D models "E:\My AI\comfyui\models"

This created a symbolic link for the models folder within the ComfyUI directory. Afterward, I started ComfyUI and verified that all the data from my Models folder is correctly displayed.

My question is:
Will this method work consistently in all scenarios, or are there any potential issues or limitations I should be aware of?

[Woisek]

Junctions are the way to go. The only thing I would change, is, to remove spaces in path and file names. Saves a lot of headaches in some cases,

[ltdrdata]

If you do it that way, problems will occur when updating ComfyUI.
Instead, copy ComfyUI/extra_model_paths.yaml.example to ComfyUI/extra_model_paths.yaml and modify it for use.

[chnisar515]

If you do it that way, problems will occur when updating ComfyUI. Instead, copy ComfyUI/extra_model_paths.yaml.example to ComfyUI/extra_model_paths.yaml and modify it for use.

Thank you for your reply, as i already gave you my new path of models folder so can you guide me how i can update it in ComfyUI/extra_model_paths.yaml ? i not a developer so please guide me.

[stoutimon]

thanks for the quick response dude. I'm trying to direct models from comfy pinokio to comfy local, will this work?

[ltdrdata]

comfyui:
     base_path: E:\My AI\comfyui\models
     is_default: true
     checkpoints: models/checkpoints/
     clip: models/clip/
     clip_vision: models/clip_vision/
     configs: models/configs/
     controlnet: models/controlnet/
     diffusion_models: |
                  models/diffusion_models
                  models/unet
     embeddings: models/embeddings/
     loras: models/loras/
     upscale_models: models/upscale_models/
     vae: models/vae/

But I recommend that you remove the blanks in the path as mentioned above.
Blanks in paths often cause trouble.

[chnisar515]

I have removed the blanks from my path and now the path is E:\comfyuimodels\models

my extra_model_paths.yaml now looks like below

comfyui:
base_path: E:\comfyuimodels\models
is_default: true
checkpoints: models/checkpoints/
clip: models/clip/
clip_vision: models/clip_vision/
configs: models/configs/
controlnet: models/controlnet/
diffusion_models: |
models/diffusion_models
models/unet
embeddings: models/embeddings/
loras: models/loras/
upscale_models: models/upscale_models/
vae: models/vae/

but now when i start comfyui i am getting this error

C:\ComfyUI_windows_portable>.\python_embeded\python.exe -s ComfyUI\main.py --windows-standalone-build
[START] Security scan
[DONE] Security scan

ComfyUI-Manager: installing dependencies done.

** ComfyUI startup time: 2024-09-22 05:11:43.100221
** Platform: Windows
** Python version: 3.11.9 (tags/v3.11.9:de54cf5, Apr 2 2024, 10:12:12) [MSC v.1938 64 bit (AMD64)]
** Python executable: C:\ComfyUI_windows_portable\python_embeded\python.exe
** ComfyUI Path: C:\ComfyUI_windows_portable\ComfyUI
** Log path: C:\ComfyUI_windows_portable\comfyui.log

Prestartup times for custom nodes:
0.0 seconds: C:\ComfyUI_windows_portable\ComfyUI\custom_nodes\rgthree-comfy
0.0 seconds: C:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Easy-Use
1.5 seconds: C:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager

Total VRAM 12288 MB, total RAM 32607 MB
pytorch version: 2.4.0+cu121
C:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\xformers\ops\fmha\flash.py:211: FutureWarning: torch.library.impl_abstract was renamed to torch.library.register_fake. Please use that instead; we will remove torch.library.impl_abstract in a future version of PyTorch.
@torch.library.impl_abstract("xformers_flash::flash_fwd")
C:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\xformers\ops\fmha\flash.py:344: FutureWarning: torch.library.impl_abstract was renamed to torch.library.register_fake. Please use that instead; we will remove torch.library.impl_abstract in a future version of PyTorch.
@torch.library.impl_abstract("xformers_flash::flash_bwd")
xformers version: 0.0.27.post2
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 3060 : cudaMallocAsync
Using xformers cross attention
[Prompt Server] web root: C:\ComfyUI_windows_portable\ComfyUI\web
Traceback (most recent call last):
File "C:\ComfyUI_windows_portable\ComfyUI\main.py", line 204, in
utils.extra_config.load_extra_path_config(extra_model_paths_config_path)
File "C:\ComfyUI_windows_portable\ComfyUI\utils\extra_config.py", line 18, in load_extra_path_config
for y in conf[x].split("\n"):
^^^^^^^^^^^^^
AttributeError: 'bool' object has no attribute 'split'

C:\ComfyUI_windows_portable>pause
Press any key to continue . . .

[chnisar515]

Solved with this...

other_ui:
base_path: E:\comfyuimodels\models
checkpoints: models/checkpoints/
clip: models/clip/
clip_interrogator: models/clip_interrogator/
clip_vision: models/clip_vision/
configs: models/configs/
controlnet: models/controlnet/
diffusers: models/diffusers/
diffusion_models: models/diffusion_models/
embeddings: models/embeddings/
gligen: models/gligen/
hypernetworks: models/hypernetworks/
LLM: models/LLM/
llm_gguf: models/llm_gguf/
loras: models/loras/
photomaker: models/photomaker/
style_models: models/style_models/
unet: models/unet/
upscale_models: models/upscale_models/
vae: models/vae/
vae_approx: models/vae_approx/

[ritonV]

I have exactly the same config, and I've copied your script. It gives the same error message than before
File "E:\ComfyUI\ComfyUI\main.py", line 97, in
apply_custom_paths()
File "E:\ComfyUI\ComfyUI\main.py", line 26, in apply_custom_paths
utils.extra_config.load_extra_path_config(extra_model_paths_config_path)
File "E:\ComfyUI\ComfyUI\utils\extra_config.py", line 21, in load_extra_path_config
for y in conf[x].split("\n"):
~~~~^^^
TypeError: string indices must be integers, not 'str'

[flynet]

Why not use extra_model_paths.yaml.example ? It's simple

[DrFexer]

Why not use extra_model_paths.yaml.example ? It's simple

Thats exactly what he's doing and it doesn't work.

[flynet]

It does as supposed
This mine working example
extra_model_paths.yaml.example.txt

`
comfyui:
base_path: E:/
is_default: true
checkpoints: models/checkpoints/
clip: models/clip/
clip_vision: models/clip_vision/
configs: models/configs/
controlnet: models/controlnet/
diffusion_models: |
models/diffusion_models
models/unet
embeddings: models/embeddings/
loras: models/loras/
upscale_models: models/upscale_models/
vae: models/vae/

`

Look this in not valid Yaml formating, this code view deleted all required spaces.
So before just copying validate Yaml format first.

[stoutimon]

I tried this with local comfyui install to comfyui with pinokio, symbolic link didn't work nor did the extra yaml example file, so what I gather from this thread is that the 2 comfyui installs are not compatible with using models across each other, so in order for this 2 work downling models for comfyui local to a different drive will need me to move the models folder and then direct to comfyui via yaml and or hardlink, is this correct?

[ltdrdata]

If the extra_model_paths.yaml file has no effect, it means that it is not properly configured in the ComfyUI environment launched by Pinokio.

For the reference,
https://www.reddit.com/r/comfyui/comments/1iby7t3/fng_needs_help_noob_with_a_technical_question/

[LorisTecnology]

that is working for me, however, let's say that i want to use multiple folder path on different drives since one is already full

[anthwalrt]

Here's the SOLUTION FOLKS --- https://youtu.be/3xbXmREpEGs

(A Wise Man Once Said - Ai solves Human problems, Human solves Ai problem )
Thank you and sayonara waqa waqa aee aee...😎😎

[LorisTecnology]

yeah this is what i'm doing rn on Portable but the issue persist, what if i want to use 2 drives for my models?

[stoutimon]

My aim was to use the model folder from pinokio install of comfyui to a desktop app of comfyui. I learned that the pinokio folder structure does not work this way. Thanks for the solution, I appreciate it.