Home
Creators
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

1-Click INSTALL KONTEXT AI IMAGE EDITING KING ComfyUI WebUI! (Patreon)

Published:
2025-07-02 19:39:53
Edited:
2025-07-18 16:12:35
Imported:
2026-06
Tags:

Downloads

Content

Hey everyone! I've created a 1-click installer for the newest "THE NEW MULTIMODAL AI KING!", called "KONTEXT". Simply THE MOST POWERFUL AND INSANE AI IMAGE EDITING MODEL EVER! A brand new 12-BILLION-PARAMETER model from the makers of Flux called “Kontext”. A multimodal model that is able to automatically edits images with incredible precision. The amount of things you can do with this model is just mind blowing AND by using the brand new Nunchaku model, can generate images IN SECONDS!!!… It's just amazing!

You can check out the video right here: https://youtu.be/Lsz3tlbBaqM

The workflow: https://www.patreon.com/posts/133144010

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!!! AND FOR EASY NUNCHAKU INSTALLATION!

1. Download the KONTEXT-COMFYUI-MANAGER_AUTO_INSTALL.bat

2. Run the bat file and select either option 1, 2 or 3 depending on your GPU vram and then determine if you want the FP8 model (best if you have more than 16gb of vram) THEN select if you want to install the Nunchaku model (HIGHLY RECOMMENDED FOR EVERYONE!). Press 1 if you have a 5000 series GPU or 2 if you have another older GPU.

3. Wait for the install to finish

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

IF YOU ALREADY HAVE COMFYUI INSTALLED USING A PREVIOUS INSTALLER (not recommended):

1. Download the KONTEXT-MODELS-NODES_INSTALL.bat and place it in the ComfyUI_windows_portable\ComfyUI folder

2. Run the bat file and select either option 1, 2 or 3 depending on your GPU vram and then determine if you want the FP8 model (best if you have more than 16gb of vram) THEN select if you want to install the Nunchaku model (HIGHLY RECOMMENDED FOR EVERYONE!). Press 1 if you have a 5000 series GPU or 2 if you have another older GPU.

3. Wait for the install to finish

4. ???

5. 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:

1. Create an account if you haven't already: Runpod

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

3. Choose a GPU with at least 24gb of VRAM (4090 is best), click here for the comfyui template  (aitrepreneur/comfyui), then edit the template and choose 80gb for both the container and volume disk, then deploy on demand

4. Go to my pods, wait for everything to finish and then click "connect", then "Connect to HTTP SERVICE port 8888"

5. Then go inside the comfyui folder and then drag and drop the KONTEXT_AUTO_INSTALL-RUNPOD.sh file on the left side of the UI then click on the "Terminal" icon on the right side on the UI

6. Copy and paste these two lines then press enter:

chmod +x KONTEXT_AUTO_INSTALL-RUNPOD.sh

./KONTEXT_AUTO_INSTALL-RUNPOD.sh

7. Wait for everything to be installed

8. Go back to the port 3000 and click on the manager to update comfyui and restart it

9. Load the FLUX_KONTEXT_IMAGE_EDITING.json file and there you go😎

Also use the command: tail -f comfyui.log

inside the workspace/logs to get the real time logs so that you always know when the install is done, It will then ask you to restart, so click on that restart button, wait for the webui to reboot, click refresh once it's done and there you go.

IF YOU ARE USING RUNPOD AND YOU WANT TO STOP THE POD THEN RESTART IT:

1. Go inside your pod

2. SERVICE port 8888

3. Go inside the Comfyui folder, open a terminal icon and re copy and paste those two lines:

chmod +x KONTEXT_AUTO_INSTALL-RUNPOD.sh

./KONTEXT_AUTO_INSTALL-RUNPOD.sh

4. And then in a few seconds it will reinstall all the requirements for the nodes

5. Go back to comfyui port 3000, click on manager, click update all, wait for everything to be updated then restart.

6. Re drag and drop the FLUX_KONTEXT_IMAGE_EDITING.json file inside, it will then tell you that a node is missing, just click on the manager button, click on install missing nodes, select the nunchaku node then click install, restart the webui and there you go😎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😉!Ps: Sorry for being very quiet again, some of you kinda know the drill by now....it's nothing new...spend a few days in the hospital and was bedridden for a while so thank you all for being patient! 🫶

Files

Comments

cool1

It says "5. Profit" but they say "This model falls under the FLUX.1 [dev] Non-Commercial License.". So I assume we can't profit from it as that would be using it commercially. Unless there's some other page (that I can't see) where they say it can be used/the outputs can be used commercially. BAGEL allows commercial use though and that's open source.

Aditya Sharma

Not sure why the VACE_IMAGE_TO_VIDEO WORKFLOW is mentioned here for the runpod part

Anton Karpuzikov

i've repeated steps exactly as in tht video, with the same prompt: Prompt execution failed Prompt outputs failed validation: LoadImage: - Custom validation failed for node: image - Invalid image file: Volume_12_29.webp UnetLoaderGGUF: - Value not in list: unet_name: 'flux1-kontext-dev-Q8_0.gguf' not in ['flux1-kontext-dev-Q4_K_S.gguf']

Sam

Best one yet Just got to see if someone brings out a nsfw lora

Brumfondl Majikthise

Getting a similar error to Anton: Prompt outputs failed validation: LoadImage: - Custom validation failed for node: image - Invalid image file: doll_1.webp LoadImage: - Custom validation failed for node: image - Invalid image file: assets_task_01jyx74gdaftj8szqkve06bpch_1751179817_img_1.webp

Matthew Salinas

Hello! I am not sure if I am doing something wrong, I downloaded the auto installer and i keep getting an error saying "Unsupported 16-bit application" and to check with the software vendor to ask if a 64-bit windows compatible version is available. Is this on my end? For context, I am running this on a windows 11 box, I have had no other issue running your other auto-installers! Thanks you :)

Aitrepreneur

They clarified the license, outputs are authorized for commercial use. Also in reality, in practice, it's not like they actually enforce the license, I have never heard of anyone actually having an issue because of the license.

FractalD

I'm hoping for better ones. The current nsfw loras aren't very good. The one mentioned in the video is blurry and the other one seems unreliable.

Aitrepreneur

2 issues, looks like the image you are using might be corrupted or an invalid format and you have not selected the correct model in the gguf workflow.

Aitrepreneur

The secret is to actually combine multiple loras, the main kontext lora + an anatomy flux lora, this gives a much better result

uisato

Glad to see you back! I had to install ComfyUI from scratch, but it is now working like a charm, thanks. Brief question: How does the second image input work? Can I add specific objects to a character?

Aitrepreneur

Watch the video If you haven't already I explain everything there and also read the prompting guide, Kontext is kinda difficult to use if you don't know how to prompt for it correctly hence why so many people missed out on it.

Yu Yeh

Hello, this is great. Can I use my custom LoRas I trained on Flux.dev with Flux Kontext ?

Aitrepreneur

Some work but usually no, you need new loras since it's a new model

realexis christofides

Hi, :) Thx for this. I have the latest confyUI instaled. Do i really need to remove mine to get this .bat Installation working. I have so many other models/Custom nodes and workflows in my current set up. I don't wanna louse or do smth that will make my files not work anymore. Thank you

cool1

Thanks. They clarified saying "We did not intend to alter the spirit of the license concerning the issues you raised. To clarify this, we have reverted Sections 2.d and 4.b to be in line with the corresponding parts in the FLUX.1 [dev] Non-Commercial License.". When I questioned perplexity.ai about it after giving that quote from Black Forest, perplexity thinks we can use it locally to generate images we can use commercially but we can't host the model etc. for others to use commercially. Though this page (linked from the discussion about Flux Kontext rights): https://bfl.ai/legal/non-commercial-license-terms says "grants you a non-exclusive, worldwide, non-transferable, non-sublicensable, revocable, royalty free and limited license to access, use, create Derivatives of, and Distribute the FLUX.1 [dev] Models and Derivatives solely for your Non-Commercial Purposes. ". so "solely for your Non-commercial purposes". So I think if we can use the model (Flux Kontext) locally for commercial reasons and use the images it outputs for commercial reasons that Black Forest haven't been clear about that (if they're linking to that non-commercial use page). BAGEL license (Apache I think) is a lot more permissive and allows commercial use as far as I know. Could you please create an install option for that (a GUI)? Whether or not it's as good technically as Flux Kontext, the better license with BAGEL should make things a lot better/simpler if we might want to use it commercially.

Firmy Van Bas

I used the Kontext model install only because I already had Comfyui installed, but how can I download Nunchuk now? It wont update properly through the manager

Kerfelox

For some reason it does not update, apply the nunchaku node and also i try to update comfyui it is not doing it, tried refresh, stopping it and starting it again. I am Runpod user. Any idea?

Wiktor Waszak

Yo, how long does it take to deploy template on runpod? Usually pod deployment took at most couple of minutes but now I selected your template and it's been deploying for 30 minutes with no end in sight. I just see "still fetching image" message...

LW

Sorry to here of your health issues, K. I hope you recover quickly. In your video you asked if people might be interested in a video on how to train loras for Kontext.... Hmmm, Is the Pope Catholic? I think you have your answer right there! Do take your health in to account though! Thanks again for all your great videos and installers.

Aitrepreneur

Bagel license is great and all but....have you tried it???? It's not great and it requires way more vram than the nunchaku Kontext model for a slower speed.

Aitrepreneur

No you don't have to, you can use the KONTEXT-MODELS-NODES_INSTALL.bat instead like explained in the video but you might have to first update comfyui using the updater inside the update folder first to make 100% sure you got the very last version before running the installer.

Aitrepreneur

you might have to first update comfyui using the updater inside the update folder first to make 100% sure you got the very last version before running the installer. You also need to check is the nunchaku wheels were installed correctly with your version of comfyu. If you have any issue you can dm me

Aitrepreneur

Dm me but have you followed the video, I show everything at the end

Aitrepreneur

It can happen, some GPU have slower connection speed, the way to solve that is to simply run a new pod, with the same gpu and wait until this one starts so you can stop the one still running. It usually takes like a minute or 2 for the pod to start

Aitrepreneur

Yeah I'm interested in that as well, I will make it my priority, Kontext is such an incredible model fr..

anooying

tried the auto installer 'git-setup.exe cannot start or run due to incompatibility with 64bit versions of windows'.

VR AnimeTed

Sorry to hear about the health issues, hang in there! Also, thanks for the new 1 click installer.

cool1

I've only tried the online version which is quite slow (I assume it might be slower than the local version if multiple people are using it in a queue). I know it's not perfect and it gets something things wrong and some things aren't very realistic but others are okay. There's no "undo" button on the online demo. It would be helpful for when it's needed if a change didn't work well. Adding text didn't work well in my tests but adding other things could work quite well. Replacing a flag worked well. It removed both people from a photo well I think in the 2nd attempt (though I asked it only to remove one). I assume they could still be improving it or other people might have more accurate versions of it since it's open source. I think they had demo images of it working well in their page about BAGEL. Sometimes on the online demo I had the image/text output option set wrong for what I wanted. I think it would be good if it could have an option where it output the thing that it thought was most relevant (either text or image) based on the prompt. So I think it would be good if we could have a 1-click install for it, even if it's not quite as good technically as flux kontext because the license is much better and clearer about commercial use. I don't know how much vram it uses but I've got 24 GB VRAM and I assume we could use quantized versions if needed.

anooying

Thanks. Anything else I need to install? I assume cuda? Do i need visual studio?

Dawan Star

Hi, i run this on runpod. When trying to install nunchaku nodes i get error message "import failed" and nunchaku is not available. After restart and refresh it again shows a message that nunchaku nodes are missing

Alpha Ghost 47

Thanks and I hope your feeling better soon. Your videos are the best 👌

joel lind

Could you make a video exclusively for patreon, where you go through how to make nsfw generations since you can’t upload it on YouTube?

Anton Karpuzikov

No, I figured it out. The “Invalid image file: Volume_12_29.webp” error occurred because activating that section of the workflow automatically enables the second input image by default. I just needed to turn it off. The UnetLoaderGGUF error happened because, for some reason, the GGUF model didn’t download automatically. I downloaded it manually, and now everything works.

Tyler89537

Same here after following the video. Last night I was able to get nunchaku to load fine after update all, but not today.

HealingPaint

how much HDD space is needed to run V3-MODELS-NODES_INSTALL.bat ?? Will python.exe -s -m pip install --upgrade torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124 work or do I have to have 12.6 or higher CUDA? *edit: To confirm that installation command/CUDA install did indeed work.

Dawan Star

Now how we can fix it?

Aitrepreneur

yup cuda toolkit and visual studio with desktop development with c++

Aitrepreneur

there is no controlnet models for Hidream, the model unfortunately completely flopped with the community to my dismay

Aitrepreneur

what gpu have you rented and have followed the steps I show in the video at the end?

Aitrepreneur

Yeah like 30gb Maybe? I don't remember but something around that

Yu Yeh

Thanks. Have you trained loras using this new model?

Aitrepreneur

I did, it's great. I'm preparing a new video on it

Christian LOUIS

Where do I have to put the nsfw Loras and activate them

Artyom

Any chance to get a One click installer for the new Stable Diffusion 3.5 Models Optimized with TensorRT released in June by StabilityAI and Nvidia ? Thanks, take care !

Dawan Star

rtx 4090, yes I followed your video

Yu Yeh

Can't wait!!!!!!!!!!!!!!!!!!!!!

Aitrepreneur

I'm aware of the issue, I'm trying to find a solution. No worries I got them all saved up locally

Aitrepreneur

Probably not, stability ai is dead as a company and stable diffusion 3.5 doesn't hold a candle to flux so kinda useless

Aitrepreneur

In the models/lora folder then load it from the lora loader node (the pink one)

Padraig Kane

hello when i try to install it says unsupported 16 bit app what can i do to fix it can anyone help? thanks!

fBombX220

I'm getting a SyntaxError early in the installation of the KONTEXT-COMFYUI-MANAGER_AUTO_INSTALL.bat See below: File "dataclasses.py", line 1265, in wrap File "dataclasses.py", line 1063, in _process_class File "dataclasses.py", line 619, in _init_fn File "dataclasses.py", line 473, in _create_fn File "", line 1 def __create_fn__(__dataclass_type_nx-loopback__, __dataclass_HAS_DEFAULT_FACTORY__, __dataclass_builtins_object__, __dataclass_return_type__): ^ SyntaxError: invalid syntax Press any key to continue . . .

halloking123

update phyton your problem is that the file is called nx-loopback phyton does not recognize "-" it should be called nx_loopback. but i am not an expert

fBombX220

That didn't work. I also noticed these "ΓÇô" are showing up so I don't know if that problem is related or not. See below: Which Kontext quantization? 1) Q4_K_S ΓÇô GPUs < 12 GB vRAM 2) Q5_K_S ΓÇô GPUs 12-16 GB 3) Q8_0 ΓÇô Best quality, GPUs 16 GB+

Xyzzx

Me too. When I open the workflow json from you other post I get: Missing node types. NunchakuFluxDiTLoader NunchakuFluxLoraLoader Install all missing nodes doesn't seem to do anything. Opening the manager an updating all doesn't seem to do anything. Not sure what I'm missing here. I guess I should do the -tail thing to get more details...

Ari Verzosa

Finally found a setup that works. Thank you ! Able to produce image but have an error that says until finally completed 0%| | 0/20 [00:00

Luma Rose

Same for me i tested to downgrade it to 0.2.0 aswell but still didint work Missing node types. NunchakuFluxDiTLoader

halloking123

ΓÇô means "–" open the file in the editor, copy everything it contains into chatgpt and ask if it finds any errors and can correct them. this actually works pretty well for me. otherwise, unfortunately, i have no idea either

fBombX220

I finally got it working by uninstalling all versions of python on my PC and then reinstalling the KONTEXT-COMFYUI-MANAGER_AUTO_INSTALL.bat however I still have the ΓÇô character encoding issue.

Joseph Ronie

The lora is missing can someone link me please

Aaron1611

Everything works great, ty. No nsfw lora though, gone from hugging face. Is there a way you could share your download? Ty

Padraig Kane

i can't seem to find the lora needed can someone pls link me it, much appreciated

Padraig Kane

hi ari do you have good lora could send me a link pls

ПЕТЪР ЕВТИМОВ

Hello, Please Help - it started normally, I put your older Json for video, tried to install missing , and after restarting is giving this message: File "C:\AI\COMFI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\aiohttp\http.py", line 5, in from . import __version__ ImportError: cannot import name '__version__' from 'aiohttp' (unknown location) C:\AI\COMFI\ComfyUI_windows_portable>pause Press any key to continue . . . I tried to update , restarted several time, computer too. The same result. What to do?

Christopher Carlson

Update everyone, I found the problem. I was only using one image and did not disable the second image input box.

Wangers n Mash

I deleted the aiohttp directories N:\AI\Programs\cmfyv3a\ComfyUI_windows_portable\python_embeded\Lib\site-packages (aiohttp*.?) then ran this, replacing 'N:\AI\Programs\cmfyv3a\' with your install dir.. N:\AI\Programs\cmfyv3a\ComfyUI_windows_portable\python_embeded\python.exe -m pip install --force-reinstall --no-deps aiohttp==3.12.13 your mileage may vary and it may break or make things worse but it allowed mine to run.

Joshua Whigham

Did the auto install, and followed along with the video. I put the checkpoint in the checkpoint folder and tried to run something. It stopped at 33% and said reconnecting. The terminal says it's paused and to "press any key to continue", I do so and it boots me out completely. I havent been able to generate one image. Would love some assistance

Mailman

hugginface "cleaned house" and removed the... spicy Loras. Any idea where to find them now?

Artyom

Oh ok, thanks for your response ! I installed Flux, I have a 4060 TI 16GB, it takes about 130 seconds to output an image while SD was 10-20 sec for very acceptable results, that's why I was very interested. Take care

Aitrepreneur

I show how to get a new copy in my last video, go to tensort art and search for the "kontext" user

Aitrepreneur

I show how to get a new copy in my last video, go to tensort art and search for the "kontext" user

Aitrepreneur

I show how to get a new copy in my last video, go to tensort art and search for the "kontext" user

Aitrepreneur

I got more things for the ai videos coming yes :)

Aitrepreneur

I show how to get a new copy in my last video, go to tensort art and search for the "kontext" user

Padraig Kane

i am getting this error RuntimeError: The size of tensor a (4032) must match the size of tensor b (8300) at non-singleton dimension 1 when trying to run the new lora any fixes

Per

These never install on RunPod, thus Nunchaku doesn't work. NunchakuFluxDiTLoader NunchakuFluxLoraLoader FP8 works though.

Giant Jack

Edit : the mediafire link is actually up. It's edited just below the demo images, but the complete link is in the side or bottom panel "Description". It does work with the loras downloaded from this source. Although the results are not great to be honest.

Aitrepreneur

Yup a recent update broke it, I've fixed it now. Redownload the installer from the page again. This time no need to reinstall missing nodes, once you have run the installer and updated comfyui, everything will be there

Benni

Does anybody know where to find celeb loras for flux Kontext especially now that civitai removed them

Goldsight Daemon

Any chance to get a 1-click install for the mac version of ComfyUI (M4 Max chip)?

Slap Dash Dolt

I am not able to install NunchakuFluxDiTLoader or NunchakuFluxLoraLoader in my current conda comfy install and I get black output with Nunchaku in the Comfyui portable. I have been going in circles with Gemini 2.5 pro and Qwen trying to trouble shoot this. I have updated everything and even ran the installers twice. Does anyone else have this problem on Windows or is this an operator error on my part?

DravenLock

Any Install options for AMD graphics cards?

Szymon Niewolski

Hey - quick question. When I want o install nunchaku but my gpu is Nvidia L4 -> should i choose the number 1) Nvidia 5000 series GPUs or 2) older gpus? L4 is not really the consumer product with 3000 4000 or 5000 series, so i wasnt sure which one to choose to make it work

Aitrepreneur

Good question, I haven't any model trained on celebrities yet

Aitrepreneur

I don't own any apple products so I can't make sure that they work on my side nor can I provide support for it

Aitrepreneur

send me dm with more info, GPU, screens, error log, etc

Aitrepreneur

L4 apparently uses Ada Lovelace architecture so choose number 2

mak

any update on this matter?

Kappak Amaru

Hello sir,will this work for 8gb Rtx 3050?

Robert Hulsteijn

Did a clear install of ComfuUI and when I drag the workflow.json file in, I'm getting the "Missing Node Types" message, saying it's missing 'NunchakuFluxDiTLoader' and 'NunchakuFluxLoraLoader'. I can't use the 'update all' button. When I open the manager, the 'missing' TAB is empty, in the 'Workflow' TAB the 'ComfyUI-nunchaku' (v 0.3.3) Node is greyed out and when I try to update it, it doesn't work... Please help

Carbureanu Ionut

Hi! Can you try and see it way not work .?? I was made it from scratch and nothing

Ricardo Sanchez

yo pude solucionar el problema por medio de Software "Windsurf" tuve que darle acceso a mi carpeta de instalacion y despues me soluciono el problema. resulta que tenia una incopatibilidad de numpy. ¿Cuál fue la solución? Bajamos la versión de numpy a 1.26.4 Esto resolvió el error de incompatibilidad binaria (ValueError: numpy.dtype size changed...) que impedía cargar correctamente los nodos de ComfyUI-nunchaku y insightface. Reinstalamos insightface Esto recompiló/descargó las extensiones Cython de insightface y sus dependencias, asegurando que fueran compatibles con la versión de numpy instalada. ComfyUI arrancó correctamente Al iniciar ComfyUI después de estos pasos, ya no apareció el error de incompatibilidad binaria, y los nodos personalizados (incluyendo ComfyUI-nunchaku) pudieron cargarse. ¿Por qué funcionó? La incompatibilidad binaria ocurre cuando los módulos Cython (como los de insightface/face3d) fueron compilados para una versión de numpy diferente a la que está instalada. numpy 2.x todavía no es 100% compatible con todos los paquetes de IA en Windows, por lo que la versión 1.26.x es la más segura para estos entornos. Al reinstalar insightface después de bajar numpy, aseguramos que las extensiones se recompilaran (o se descargara un wheel compatible) para esa versión específica de numpy.

QueenBambi

Any update on fixing the matter Missing Node Types error (NunchakuFluxDiTLoader, NunchakuFluxLoraLoader). Just subbed and would love to try this. <3

Padraig Kane

needed that thanks!!! your such a great creator thanks so much

Essence Cartoon

Getting same error but running the special workflow didn't fix it

phail216

Same for me :(

Not_Mad_Man

I did a clean install with option 1. I got this error: Missing Node Types When loading the graph, the following node types were not found NunchakuFluxLoraLoader NunchakuFluxDiTLoader Then I tried option 2. I still get the same errors. The other nodes work fine, I just can't use Nunchaku. Any idea what causes this, and or a fix?

Aitrepreneur

Yup an update just wrecked the nunchaku install, I've updated the installers, redownload them from the page and preferably do an install from scratch

Aitrepreneur

Yup an update just wrecked the nunchaku install, I've updated the installers, redownload them from the page and preferably do an install from scratch

Aitrepreneur

Yup an update just wrecked the nunchaku install, I've updated the installers, redownload them from the page and preferably do an install from scratch

mak

bro I dunno why you deleted all the problematic comments this seems sketchy and only left ones that worked for ya, your new workflow that you uploaded doesn't handle install of comfy leaving you with "comfy ui curl: (18) end of response with 1095629307 bytes missing" comparing to the og ones. As additional feedback it would be great too you considering non destructive 1 click installers after which Comfy Ui crash <3

mak

did all new comfy ui install with ur model 1 click installation but nor of your proposed solutions work same error

Zaiross

Hello , I can load any Loras from civitAi , I put all in lora files but cant load in power lora loader , help please

Zaiross

yes but when I run with the lora in power lora , the lora has no effect - and the output images become blurry with no changes when I put cfg number higher than 1 :0

mak

and won't till proper solution is there, wheel installation is useless... why not to ship proper installer with wheels etc. if that solves the issue?

Aitrepreneur

The installer has been updated and it works for me even right now, I recommend using the KONTEXT-COMFYUI-MANAGER_AUTO_INSTALL.bat and install from scratch

Kevin

yes

mnkjiu

Are you able to share this LoRA: llama-anon/not-flux-kontext-dev-clothes-remover? For research purpose of course

mnkjiu

any help?

AR reality

Running this with a 1080TI. I was under the impression that when I run the workflow for the first time it would auto download all the missing models and loras. Is this not correct or am I doing something incorrect as I get a message that I'm missing clip, model and vae

Aitrepreneur

no you need to use the installer to install all the models and nodes. The workflow is just a file

Tibby

So just trying to get this to work at this point... Apparently 3.44 is no longer available and times out. If you manually get a copy and put it in the same folder the script isn't aware.... So I made a new script... but Either way non of this is working... Was my whole purpose to subscribe. I mean ... even the characters are shown in the wrong encoding. So a ``` :: Set codepage to UTF-8 to minimize display issues chcp 65001 >nul ``` Is required.

james beal

On Runpod... this was frustrating. missing 'NunchakuFluxDiTLoader' and 'NunchakuFluxLoraLoader', and others. I think issue might be runpod will load most recent CUDA and PyTorch, or it loads some that conflict with AI's template. I really have know idea. . I can't speak to local installs, but was able to get Nunchaku working with template 'ComfyUI with Flux.1-dev', dropping in the wheel loader for my CUDA and Torch, and refreshing nodes... probably missed a few steps. Anyways 3 days of fiddle fucking around, and its pretty cool and amazingly fast. Thanks A!!

robin poirieux

Is it possible to have an updated and working batch file for installation (KONTEXT-COMFYUI-MANAGER_AUTO_INSTALL.bat) with Nunchaku fully operational? I had installed it locally on my machine (4060 Ti 16 GB) this summer without any issues, and it worked really well. Unfortunately, I made the mistake of trying to update ComfyUI, and I’ve been regretting it ever since! I’d like to completely reinstall everything, but the batch file you’re currently providing doesn’t seem to be up to date. Although the installation appears to complete without errors, once ComfyUI is launched, the modules NunchakuFluxDiTLoader and NunchakuFluxLoraLoader fail to load properly, showing messages like this: ======================================== ComfyUI-nunchaku Initialization ======================================== [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/extension-node-map.json Nunchaku version: 0.3.1 ComfyUI-nunchaku version: 1.0.2 ComfyUI-nunchaku 1.0.2 requires nunchaku >= v1.0.0, but found nunchaku 0.3.1. Please update nunchaku. Node `NunchakuFluxDiTLoader` import failed: ... Node `NunchakuQwenImageDiTLoader` import failed: ... I also tried installing it manually following the Nunchaku documentation (from GitHub), but I ended up with the same issue. Thanks in advance for your feedback.