Home Creators Posts Import Register Favorites Logout

Downloads

Content

EDIT 3 September:NEW RUNPOD CUDA 13 TEMPLATE!

I’ve spend the last few days, completely rebuilding my RunPod ComfyUI template from scratch.... The previous template was getting very outdated and was starting to cause problems with newer GPUs, ComfyUI versions and recent workflows, so please use this new template from now on.

The new template uses CUDA 13, supports newer GPUs like the RTX 50 series, comes with SageAttention preinstalled, automatically keeps ComfyUI updated and works with my new RunPod installers.

HOW TO USE THE NEW TEMPLATE

I made a quick video showing exactly how to deploy it and use the installers:

👉 https://youtu.be/9wlUXXyy9mE

👉 RUNPOD TEMPLATE LINK

IMPORTANT: When choosing your GPU, use the CUDA 13.x filter on RunPod

The old RunPod template is now completely retired, so use this one for all new installs going forward.To use the installer use this command:bash "QWEN-IMAGE-EDIT-PLUS-AUTO_INSTALL-RUNPOD-NEW.sh"I WILL SPEND THE NEXT FEW DAYS UPDATING ALL PREVIOUS RUNPOD INSTALLERS TO THE NEW TEMPLATE SO IF THERE IS AN INSTALLER THAT I FORGOT LET ME KNOW.Hey everyone! I've created a 1-click installer for THE NEW MULTIMODAL AI KING!", called "QWEN IMAGE EDIT" made by the same team that created

Hey everyone! I've created a 1-click installer for THE NEW MULTIMODAL AI KING!", called "QWEN IMAGE EDIT" made by the same team that created Qwen image. A multimodal model that is able to automatically edits images with incredible precision and that follows your prompt better than Kontext.

You can check out the video right here: https://youtu.be/6EVVV2lUMFU

The workflow: https://www.patreon.com/posts/137036479

The installer of course automates the entire install process for comfyUI and the models downloads!

LOCAL INSTALL!!! RECOMMENDED TO START FROM SCRATCH SINCE YOU NEED THE LATEST COMFYUI TORCH VERSION!!!

Download the QWEN-IMAGE-EDIT-COMFYUI-MANAGER_AUTO_INSTALL.bat

Run the bat file and select either option 1, 2 or 3 depending on your GPU vram.

Wait for the install to finish

The webui will launch, load the QWEN_IMAGE_EDIT_WORKFLOW.json file and there you go😎

DON'T FORGET TO GO UPDATE COMFYUI BY GOING INSIDE THE UPDATE FOLDER AND RUNNING THE COMFYUI UPDATE BAT FILE!!!!

IF YOU ALREADY HAVE COMFYUI INSTALLED USING A PREVIOUS INSTALLER:

Download the QWEN-IMAGE-EDIT-MODELS-NODES_INSTALL.bat and place it in the ComfyUI_windows_portable\ComfyUI folder

Run the bat file and select either option 1, 2 or 3 depending on your GPU vram.

Wait for the install to finish

???

Profit 😎

IF YOU GET AN ERROR WHEN LAUNCHING COMFYUI AFTER USING THIS INSTALLER SAYING "AssertionError: Torch not compiled with CUDA enabled" it's ok don't panic.

Just go inside the python embedded folder, click on the folder path, type cmd press enter, this will bring the command prompt window and there type:

python.exe -m pip uninstall torch

python.exe -s -m pip install --upgrade torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

--->replace cu128 by your cuda version (if you don't know your cuda version just open a new cmd window and type nvcc --version it will then tell you at the end Cuda compilation tools, release 12.8, V12.8.93 or something different like 12.6, in that case just use cu126 instead of cu128)

IF YOU ARE USING RUNPOD:

Create an account if you haven't already: Runpod

Click on Pod (on the left side) then click deploy

When selecting your GPU on RunPod, make sure you click on Additional Filters → CUDA Versions → CUDA 13.0Then choose a GPU, click here for the comfyui template (aitrepreneur/comfyui:3.0.1-cuda13)

Then click deploy, wait for everything to finish and then click on port 8888 and then drag and drop the QWEN-IMAGE-EDIT-PLUS-AUTO_INSTALL-RUNPOD-NEW.sh file on the left side of the UI then click on the "Terminal" icon on the right side on the UI

Copy and paste this line then press enter:

bash "QWEN-IMAGE-EDIT-PLUS-AUTO_INSTALL-RUNPOD-NEW.sh"

Wait for everything to be installed

Then go to the port 8188

Load the QWEN-IMAGE-EDIT-PLUS_ULTRA.json workflow and there you go😎

Also use the command:

tail -f comfyui.log

inside the workspace/logs to get the real time logs

As always, supporting me on Patreon allows me to keep creating helpful resources like this for the community. Thank you for your support - now go have some fun😉!

Files

Comments

Jaina Shop

Working great! Thank you! I was waiting for this one :))

daniel baychev

-------- Downloading ComfyUI -------- % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0 9 1981M 9 180M 0 0 577k 0 0:58:31 0:05:18 0:53:13 713k curl: (18) end of response with 1889257997 bytes missing Download failed. Press any key to continue . . . anyone know why I keep getting this err? it happens on Q4_k_s for under 12 gb. Not sure if that matters

Cristo

Where is the workflow file?

TheRedGuy

I had the same a few times, for some reason it hits a terribly slow download speed occasionally and then it cuts off before its fully downloaded. I just tried it again after and it worked.

MisterCapitalE

After trying multiple times, finally got it installed on RunPod, but then was left with this error after trying to create a picture. VAEEncode CUDA error: no kernel image is available for execution on the device CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1 Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

VFxlen

are you using 4090 or 5090? looks like this workflow is compatible for just 4090

Plaiboy Magazine

Thank you! I also have a question about style transfers. Is there a way to use the second image as a kind of "change image on the left to match the art style of the image on the right?" or something like "match image on the left to the color palette from the image on the right?"a kind of like ipadapter hack? Could it be done with just prompting or is it as simple as just using a different workflow?

Wilhelm Zabłotnikov

We need a LoRa for undressing girls. There are more than 8 billion people in the world, and still no one has put it on the Internet.

Сума Сбродный

Everything is cool, Qwen works as it should, but now my V3_1-ULTIMATE_FLUX_ALL-IN-ONE and WAN2_2-LONGER_VIDEOS_WORKFLOW are broken

thom mick

I have the same problem, already tried 4 times and everytime download failed.

thom mick

everytime shortly after 50% then it fialed...

ANAcoNDAunit

Hi! When I try to install on RunPod, I find this error: • downloading qwen_image_vae.safetensors curl: (22) The requested URL returned error: 504 . Look's like the link for the download is not the correct one , or has been changed in the mean time. I have changed the download link on the script and now is working.

Jaina Shop

you can download a LoRa for that , on Civ pick one nsfw compatible with 4 steps, put in loras folder, press R o refresh the nodes in comfy, add the new LoRa ....

Jaina Shop

Install a new comfy portable folder all the time for test first. Now you have to reinstall the previous comfyui for Ultimate_flux and Wan22, keep them separate, a bigger Drive is the key :)

Neil Rhodes

Installing to my A drive M.2 drive and 90% of the install log is complaining about "The script...... is installed in A:\ComfyUI .....which is not on path" lines and lines of logs in orange will it still work?

Wilhelm Zabłotnikov

Do you mean MCNL LoRa is compatible with QWEN edit? There's not much else out there.

MisterCapitalE

I was using a 5090, but with the above work around, I got it to work.

daniel baychev

what i can suggest to people who experience my issue: Download comfyui manually and extract it (get it from github -> releases and find the archive link - the portable one Then take the Nodes install and run it from inside the comfy ui, just like the explanation in the description for the nodes only :) Hope I helped

Eliot RRG

its working now, quick question for Qwen image you recommended Q4 but for Qwen Image Edit you choose Q8 so i assume for this there is a big difference right ?

RavenWT

Haha tried the transparent dress like you suggested in the video but it's censored huhu

Filament's Folly

First, I want to say I really appreciate you putting these together. Thank you. It'd be great, though, if in the one-click installer, if it would ask where we want our models downloaded. I'm using a YAML file to keep all my models for all my versions of Comfy in one location, so I don't end up with seven versions of the same model in different folders. I know it's a little bit more work, but hopefully it's something you can add to the bat file once and then continue to reuse on future projects.

MisterCapitalE

Is there any possible way of a video where we see how to train a Lora for this on a person or even style?

Jaina Shop

You don't need to install the "nodes.bat" if you have already installed the "auto_install.bat" but if you wanna update a previous (not suggested) previous version of comfyui you need to run the "node.bat" inside the PYTHON FOLDER, not on the main folder of comfyui... to make it short u don;t need the nodes.bat installed :)

Jaina Shop

mine are not censored, but quality of that part is so so and not the best one, thanks civitai you can find a LoRa dedicated to keep good results.

Jaina Shop

i'm using an old 3090ti with the q8 version 19GB model, the result is pretty similar to the q4-q5, I noticed a decent results moving from 4 steps to 8 steps...(use lora 8 step of curse, see aitrepreneur note in WF)

13lackJesus

hey! im getting this error when trying the qwen image edit from my runpod mat1 and mat2 shapes cannot be multiplied (5376x1280 and 3840x1280) any one have any idea how to fix that?

JW

I get this one all the time and when I search up the models i find that they did not download fully. I'm in the process of re-downloading them all one by one. painful.

Mark

replace the comfyUI node "CLIPLoader (GGUF)" with the "Load CLIP" node, and that should remove that error. I had the same error untill i did that fix.

Crux Riajuu

I'm noob to this comfy ui stuff but is it possible to use illustrious models with it for this or the other one you posted before with the img to video?

Koray Birand

Getting Unexpected text model architecture type in GGUF file: 'qwen2vl' for Qwen2.5-VL-7B-Instruct-UD-Q8_0.gguf

theworldofanatnom

sooooo for anyone having the "shape '[1, 16, 51, 2, 80, 2]' is invalid for input of size 4177920" to error, if you look over to the right of the model loader. make sure the qwen vae is selected. even if you see it open it again and select the qwen_image_vae.

Euchale

Going to post here as it is semi related: With Kontext and Qwen Edit, and Res4Lyf, what is currently the best way to do style transfer (so give 2 images and say Image 1 in style of image 2)?

Aitrepreneur

Update comfyui by using the update comfyui bat file inside the update folder and update all the nodes by using the manager by clicking on update all, then you can restart and re drag and drop the workflow in again and make sure that you used the installer and that you have downloaded the mmproj file inside the text encoder

Neker Play

cannot reshape array of size 14392122 into shape (18944,2940) I get this problem , how could i resolve it?

Hamdi Hirzalla h

it's very likely a corrupted model. the download may have interrupted while the Bat file was running. when it happened to me I redownloaded the model and it ran fine after that. just take the model name and look it up and download and replace it. better than a full reinstall.

King Debs

Runpod was picked up as a Trojan Virus by my antivirus . Threat name 'BV:Miner-LO[Trj]'

Blue

Hello, I'm getting an error the installer is not compatible with Windows 64 bit. Anyone else have this issue?

Brandon Waters

getting the matshape error no matter how many times i redownload the models

Eponym

Any plans to release a qwen edit lora trainer tutorial?

Kankatzu

after reinstall it works but the results are... not so great. Would the 3090 run the q8 version well? I thought I'm on the save side using q5 but those are rather meh at combining two images to make a third.

Aitrepreneur

Update comfyui by using the update comfyui bat file inside the update folder and update all the nodes by using the manager by clicking on update all, then you can restart and re drag and drop the workflow in again and make sure that you used the installer and that you have downloaded the mmproj file inside the text encoder

BadgerWi Johnson

I just upgraded to Win 11 and I'm getting an error message "Program or feature git-setup. exe can not run due to incompatibility with 64 bit versions of windows." Any advice?

bakedmuffinman

expired url when trying to download the .bat installer

McDonalds

I copied the json, .sh, and two .bats into a new folder on my desktop and it started working just fine. had the same issue when trying to run everything from my downloads folder which already had some of the files in it

Kankatzu

scroll further down to the blue links at the bottom. These work. The ones within the text for some reason don't work.

Aitrepreneur

oooooh that's why some people have this issue.....they are clicking on the url within the text which for some reason link to the file below...

Bob Loblaw

Does the install of ComfyUI from "QWEN-IMAGE-EDIT-COMFYUI-MANAGER_AUTO_INSTALL.bat" work with WAN2_2-LONGER_VIDEOS_WORKFLOW? I am trying it and getting "Prompt execution failed" "Prompt outputs failed validation: " There's a ton of listings below the error. I could post the whole thing, but here's a sample: "UnetLoaderGGUF: - Value not in list: unet_name: 'Wan2.2-I2V-A14B-HighNoise-Q8_0.gguf' not in ['Qwen_Image_Edit-Q8_0.gguf']"

Sioko

I have a crash juste after 4steps generate finish and "Requested to load WanVAE" without error, a solution ?

Daniel Wright

Hey, this is so good. Thank you very much. Do you have anything as an automatic install on Mac with Apple Silicon?

Darkwan75

What is the Nodes Install bat file for? I thought install missing custom nodes would do that. Is this in case that doesn't work?

Joseph Cheng

Hi anyone getting this with qwen image editTextEncodeQwenImageEdit mat1 and mat2 shapes cannot be multiplied (5476x1280 and 3840x1280)

CodeCrafters Collective

anytime i get that, and it was today is because the clip is set to the wrong model.

Matthew Butcher

I get this error when running the QWEN-IMAGE-EDIT-COMFYUI-MANAGER_AUTO_INSTALL.bat (running Windows 11 RTX 5090 Which QWEN IMAGE Edit quantization? 1) Q4_K_S ΓÇô GPUs < 12 GB 2) Q5_K_S ΓÇô GPUs 12-16 GB 3) Q8_0 ΓÇô Best quality, GPUs 24 GB+ Enter 1, 2 or 3: 3 -------- Checking prerequisites -------- 7-Zip not found ΓÇô downloading... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 162 100 162 0 0 313 0 --:--:-- --:--:-- --:--:-- 314 The system cannot execute the specified program. 7-Zip install failed ΓÇô install it manually then rerun this script. Press any key to continue . . .

asd asd

Expired link for Download the QWEN-IMAGE-EDIT-COMFYUI-MANAGER_AUTO_INSTALL.bat

Alvaro Vilches Diaz

Hi everyone! for me it get stuck in "[ComfyUI-Manager] All startup tasks have been completed. got prompt Using pytorch attention in VAE Using pytorch attention in VAE VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16 gguf qtypes: Q5_K (168), F32 (73), Q6_K (1) Attempting to recreate sentencepiece tokenizer from GGUF file metadata... Created tokenizer with vocab size of 256384 Dequantizing token_embd.weight to prevent runtime OOM. CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16 gguf qtypes: F16 (694), Q4_K (356), Q5_K (44), F32 (1) model weight dtype torch.float16, manual cast: None model_type FLOW Requested to load WanTEModel loaded completely 5680.8 5039.953125 True 0 models unloaded. Requested to load WAN21 0 models unloaded. loaded partially 128.0 123.9649658203125 0 Attempting to release mmap (401) 0%| | 0/2 [00:00

nurtu

Has COmfyUI already jumped to Python 3.13.6?! Feels like 3.12 is much more stable, is there any way to revert to 3.12 ?

EtD

same for me

vectorlandscapes

Where is the QWEN_IMAGE_EDIT_WORKFLOW.json download?

Matthew Butcher

I downloaded 7 zip manually and made some progress

Karim A

how can i make nsfw pictures, for me dosent work

Masaru

Can I install this on ComfyUI Desktop?

TrapHouseMag

Is this working for anyone?!?!?

Foxy

try different browser/ private browser

Pat Johnson

Hi, maybe I'm doing something wrong, but I just can't seem to figure this out. I've followed the tutorial video repeatedly, did the instructions in the post, deleted the folder and reinstalled, and did it all over again. I just can't seem to get this to work. I click Run and I get an error. I'm using a Geforce 1050 ti. Here's the error I get: CUDA error: no kernel image is available for execution on the device CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1 Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions. I even googled the issue and tried to install nightly pytorch. At one point I got a wrong numpy version? In any case, that's the error I get for now. Any help would be appreciated.

Gatharian

What version of nunchaku do I need? I installed 3 different versions of nunchaku trying to get the plugin to work. 0.32 1.0.0 and 1.0.1 install_wheel.json node NUnchakuWheelInstaller says its missing. I think I am screwed.

Eline beesje

the one click installer .bat file link isn't avalable anymore. can you fix it plz?