ComfyUI Tutorial Series: Ep09 - How to Use SDXL ControlNet Union

[@pixaroma]

Just found a way to use it without Art-Venture node (for some caused some errors), you can use instead the AIO Aux Preprocessor - it comes with ComfyUI's ControlNet Auxiliary Preprocessors custom node, so one less node that you need to install. I will update the workflows on discord also discord.gg/gggpkVgBf3 Update 2: Added also one workflow for canny with no extra custom nodes need it. only the comfyui defaults. You can now support the channel and unlock exclusive perks by becoming a member: kzbin.info/door/mMbwA-s3GZDKVzGZ-kPwaQjoin

[@jonrich9675]

holy cow you are so dam good at explaining everything. Ty for the video it was super helpful!

[@imonutiy]

Ye, Open Pose crashes for me...

[@CrunchyKnuts]

Glad I read the comments, I was one of them! 😅 Index out of range

[@kidmeng3773]

Can someone explain what Index out of range issue is?

[@pixaroma]

Install aio aux preprocessor, check the next episode the beginning

[@michaelaaron2415]

Take CARE of you health Pixaroma. We are very grateful to have such a detail instructor. Thanks a lot

[@59Marcel]

Just a little something for your help today. It was very kind of you. Thank's once again.

[@pixaroma]

Thank you again for your continuous support, it definitely helps ☺️

[@septembre1129]

An extremely clear and helpful video with every detail. I'm following along and applying the steps as I watch, and now everything feels much easier. Thank you so much. This information is as valuable as gold.

[@icedzinnia]

it's crazy how the Apply ControlNet node has changed in one single month. It doesn't look like in the video, but the makers are working on intuitive UI so it's still easy to adapt to what you are showing. But in one month! This tech moves so fast

[@pixaroma]

Yes sometimes in a week it looks completely different

[@NitishShiggaon]

So what's the new way to use the newer Apply ControlNet node? I'm having the same issue. What do i connect for the input and output?

[@pixaroma]

Can post on discord screenshot, some recent post on comfy UI had some workflow with apply control net

[@DandelionStrawart]

This is a great explanation. The great thing is you always start with the basic model and start adding nodes to that, and it's easy to follow how to do this. Other guys start with a super complicated workflow to start with and you're supposed to follow what they did, it just makes it very complicated. But yours is simple and direct, in the end, you can use Controlnet with comfyui, and you have the control to expand it or keep it simple. Thanks very much for the video.

[@TheWayneFang]

Good detailed explaination~ Thanks for your sharing.

[@UnrealCpp]

since i watch this video tutorial,i can see in my mind that using two photos to create different action of the same character,which is awesome for the development of unity2D action game! appreciate for saving my college graduation project.

[@ivo_tm]

Short, clear and informative. Thank you!

[@icedzinnia]

Excellent! the way you presented this information, with the source and output and the notes in the middle makes it really easy to understand and come back to for easy, fast reference. Thank you.

[@m3mee2010]

These are great videos. Running you at .7 playback speed and it is perfect!

[@tommie34u29]

These tutorials are absolutely amazing and great quality!! Thanks!

[@ling6701]

Thanks. Another christmas. Makes me think of internet, when internet came it changed communication forever, now it seems to me that those ai tools are changing the way we design/ create image forever

[@danteshume7933]

Sir, thank you very much! I have finally waited for your update.

[@indigooo3d]

That's is an amazing skill to operate ComfyUI show by you btw.

[@GustavNielsen-b8x]

Love this keep up the good work. Like the more detailed explanation!

[@pavelponcar6902]

Thanks man, lot of things to experiment with

[@davidh5700]

Excellent tutorial. Thanks!

[@baheth3elmy16]

Awesome! Love your work...

[@MintFritter]

Amazing tutorial! Every aspect is explained so well, that even a dumb person like me can understand it. Just what I was looking for, thank you!

[@jennifertsang6572]

Great video!! Always learning a lot from them.

[@peppermint594]

you helped me getting better results, thank you.

[@SumoBundle]

Thanks for the tutorial.

[@dankuzmenka261]

Thank you! Well done and very easy to follow

[@thechrisdo]

Thanks for this info, cleared some things up for me!

[@eslamafifi1020]

as usual great tutorial thank you very much

[@Fenkreg]

Thank you for the tutors and ESPECIALLY that you hadn't add background music so I can listen to my own while learning cool stuff :3

[@devon9374]

Amazing video, thanks again :)

[@59Marcel]

Once again a not over wordy or hard to follow tutorial. Concise and to the point. Thank you.

[@Med2402]

19:37 - Opening the image in the browser tab should give height and width ( where there is X Close tab). Opening the image in the standard photos app (win11) should give the dimensions (Bottom Left). ( Load image/image----> VAE Encoder/latent ----> Sampler/latent) the save/preview image should be the same dimensions as load image. Thanks for the tutorial!

[@feraskiki655]

Much appreciated

[@Fayrus_Fuma]

Again a very useful video, but unfortunately not on my models. It will be very useful for others. Thank you for your work. Your work is very detailed and informative indeed.

[@pixaroma]

You can use with any sdxl model :) for other models like v1.5 workflow is similar just you need to download a sd 1.5 control net. For flux the union one didn't work yet how expected so I am waiting for an update on that one

[@Fayrus_Fuma]

@@pixaroma My models are Pony, Autism, and so on. I create characters and I need consistency of style. Roughly speaking I will create 50 different images but with the same character. Sd1.5 doesn't work for me in that respect. Flux, I played around and left it. It takes a very long time to make images by itself. I even such things as a box is easier for me to do in SD 1.5, because I for the same time while Flux will make them, will have time to create a dozen images.

[@pedrogorilla483]

Very good man! And the voice is very natural, how are you doing it?

[@pixaroma]

Is a voice from voiceair and 11Labs but it take in consideration how you write it and the voice adapts depending how you put the text. I am using my own words and only grammar i check with chatgpt so the punctuation is accurate

[@Gatocratas]

Nice tutorial matte. Doing some stuff with AI in my channel. This helps a lot. Hugs from Brazil.

[@benyaminaria283]

best ever seen tut control net

[@mohammedzaidkhan5687]

Thanks a lot bro

[@李云-f1b]

Very good

[@cncpts66]

followed till you mentioned "Art-Venture" I always had an issue with their custom nodes, and getting updated, instead I will subscribe and wait until something similar comes up for flux!

[@pixaroma]

I will see in the future if other nodes appear, so far it worked for me, will see how it acts on long term

[@pixaroma]

You can do it without the art venture node, use the AIO Aux Preprocessor instead.

[@cncpts66]

@@pixaroma OK I will give it a try. Thank you.

[@ralf.starke]

Thank you!

[@Ozstudiosio]

thanks fo your content

[@robrever]

Cant wait to try! But my hopes are low. I have tried to implement similar control net workflows recently on different machines with the same reaults: a bunch of red errors.

[@pixaroma]

Hope it works for you now

[@robrever]

@@pixaroma Yea, once again: when selecting preprocessor, only showing "tile"

[@pixaroma]

@robrever try the workflows i added on discord, i did some changes with different custom nodes without art venture, and have one for canny without custom node that should work. Look on pixaroma workflows channel the last workflow from today

[@YanXiao_AI]

thank you for your tutorial

[@Panacea_archive]

Amazing video ! Could you please make a video on how to use a controlnet to inpaint a face from a reference image ?

[@pixaroma]

I will see what i can do, ip adapter causes some problems on different systems if isn't set right, i need to do more research on it

[@imrabbit9950]

Best one ❤️

[@Andy4Tune]

What was the file that caused an error with "venture", problem with downloading a file or a file was a problem?

[@pixaroma]

I think ar the end was a venture problem so i replaced the venture node with another one AIO aux preprocesor, i updated the workflow on the discord, i will show also in the next video how i did

[@Andy4Tune]

​@@pixaroma much appreciated, as a general rule I prefer to use only absolutely necessary extensions (nodes) as too many of those tend to brake things.

[@Jinjinyajin]

Hi are you running comfyui now experiencing frequent bugs? I've come to the point where I'm experiencing frequent disconects for no reason after generating too large an image in the 3rd image since comfyui update on August 15th yesterday. Do you think this is a bug or something?

[@pixaroma]

Maybe is a bug, i updated this morning and didn't had a problem but it might be a memory problem bug or something but i have a lot of vram so it didn't trigger something for me. Try to update again see if get fixed if not on their github page you can post an issue or see if someone else had that problem and how they fixed github.com/comfyanonymous/ComfyUI/issues

[@giuseppedaizzole7025]

Hi there....one question. Can all this preprocesors work with flux? thanks, great work.

[@pixaroma]

I have an episode for control net with Flux, should work ok except the stacking, using multiple control net didn't work for me, so only one at the time

[@giuseppedaizzole7025]

@@pixaroma I thought that only depth and canny worked and that all the others like openpose or scrible...only worked in sdxl....on the other hand I'd like to take this opportunity to tell you about a doubt I have. In automatic1111 you could easily edit canny, getting very good results to your liking...now in confyui I don't see how to do it. Is it possible? Thank you very much for sharing, your videos are very well organized and explained.

[@pixaroma]

Depends how you edit canny, you can use probably a black and white brush that is thin to edit like add extra stuff, of if you want to edit just the threshold on how much it see you can do that with the comfyui canny node. You can use the Union comtrol net model

[@giuseppedaizzole7025]

@@pixaroma I mean editing with photopea the image created by canny, this is very easy in automatic 1111 but I can't find the way to do it in confuyi, unless it's doing many tasks to get the same result. I don't think you have a chapter dedicated to this, one on this subject would be nice, although I imagine it's a lot to ask. Maybe for Christmas...lol...anyway, thank you very much for everything. Be good

[@mitch150]

for some reason the CR Multi-ControlNet Stack does not detect the "diffusion_psytorch_model_promax.safetensors" when I click the "controlnet_1" option. it just says none. the default Controlnet Loader does detect it though. love your tutorial series btw! been binging it.

[@pixaroma]

it happened to me also once, but after restart comfyui i was able to find it, maybe is a bug

[@mitch150]

oops nevermind, I updated comfyroll and it detects it now. lol fortunately a simple fix!

[@nomoreterban]

Awesome video!. got one question, is it possible to stack flux controlnet models with Comfyroll Studio custom node? comfyroll studio description seems it only supports sdxl models.

[@pixaroma]

It didn't work with flux for me, so only if someone make a node that works with flux

[@nomoreterban]

@@pixaroma thank you sir

[@CsokaErno]

Does ControlNet work with ForgeUI properly or is it better to use with Comfy? Another one. You made me love ForgeUI, I think it is the best for me. But ComfyUI seems more flexible. Do you think it is better to use Comfy or I can stay with ForgeUI? Besides it seems the most improvement go with Comfy. Do you think it is a better choice?

[@pixaroma]

There are many things that don't work yet with forge because they changed the gradio 4 and all extension broken so now they fixing one by one so it might take a while until all things are fixed. Comfyui get the updates faster so if you don't want to wait add in a separate folder cimfyui too

[@spacemike800]

The best!

[@x9v8k]

You still need to download the models when using the AIO or ControNet Preprocessor if you do not have them. I enable Midas Depth and comfyui downloaded model from hugging. Also, could i use setunionControlNettype to use less nodes ?

[@pixaroma]

I don't think using control net type is also preparing the photo with that preprocessor because i didn't see it connecting it to an image. Also the download should happen automatically for all preprocessors just make sure you have activated long path in windows

[@MrRaycaster]

Have to say first your videos are top notch and as others have noted CONCISE! I’m very interested in controlnet as I can’t seem to find a way to implement Canny with xsarchirectual_v11.ckpt checkpoint. The checkpoint is perfect for interior design but I hope to add basic “sketchup” designs to guide it. Of course Im getting errors because the control does not match as you noted. Is there a workable controlnet or another ckpt suitable for architecture images? Thank you.

[@pixaroma]

You can combine depth with canny usually that works ok for architecture

[@SebAnt]

Thank you another superb tutorial. 🙏🏽 Is this compatible with FLUX1-dev 8 model ?

[@pixaroma]

No only work with sdxl, but i saw another union appeared for flux, once it works without problems i will do for that one as well

[@SebAnt]

@@pixaroma Thank you. Appreciate the quality of your videos and joining as a supporter.

[@pixaroma]

@@SebAnt Thank you so much for your support

[@Fret-Reps]

hey, I appreciate what you offer. but I'm frustrated. Besides that missing file, which is never recognized, I keep getting a stop on the apply controlnet because the CLIPTextEncode - 'NoneType' object has no attribute 'device'. And if some reason gets past that it stops with a red line around the "apply Controlnet. and a cirlce around the Neg. plus when trying different controlnets, i offend get a "missing file".

[@pixaroma]

Can you go to discord server and post some screenshot of the workflow and the errors you get, I will take a look. Many of those missing file happens because long path is not enabled on windows so it tries to download the files but it cant because the path is too long.

[@Fret-Reps]

@@pixaroma I'm finally getting back to this. I will post the screen shots asap. As for the long "Long path", I've already dealt with that for a different install. So, my PC will now take long paths. So that's not the issue.

[@Fret-Reps]

@@pixaroma BTW, Discord says I don't have permission to send a message.

[@pixaroma]

@@Fret-Reps use the comfui channel or offtopic are a lot of channels there

[@F99_Digital_Dance_Music_AI]

Following the tutorial, everything work in control net, ( scribble, normal, z, ) everything work but not the segmentations. seg_ufade20k, seg_ofade20k, segucoco, it dont work in artventure node but also loaded in doffrenet preprocessors node. error message: "" no module named "fvcore" what you think about?

[@pixaroma]

I saw something similar on discord, I dont use segmentation so not sure what could be the cause, sorry

[@possiblynotrohit]

Where's inpaint and outpaint in union model

[@pixaroma]

Not sure I didn't get to inpaint part yet, only worked with canny and depth and pose so far, i will do more research when i get to inpaint episode

[@Boiller2012]

How would you use pixel perfect in Comfy? There is a node for it, but I'm not exactly sure where to plug it into in the comfy graph.

[@pixaroma]

I didn't have time to test it, and i am not at pc right now. See if you can find a workflow somewhere to see where it is placed docs.getsalt.ai/md/comfyui_controlnet_aux/Nodes/PixelPerfectResolution/ www.runcomfy.com/comfyui-nodes/comfyui_controlnet_aux/PixelPerfectResolution

[@Boiller2012]

@@pixaroma That just may help. Thanks for the link!

[@pixaroma]

I have tried this morning but I can not figure out how to use it, I asked on the custom node pages in github to see if someone can help, it needs an integer output type node, but not sure what nodes I can put there to take that value and use it. Wish all those people that do custom node to give more info on how to use them.

[@PeymanTaraj]

Hi, I get this error when using perprocessor depth or pose or many others: UserWarning: `local_dir_use_symlinks` parameter is deprecated and will be ignored. The process to download files to a local folder has been updated and do not rely on symlinks anymore. You only need to pass a destination folder as`local_dir`. Could you please help me on that ?

[@pixaroma]

Did you wait a few minutes to download the model? Also make sure you activate long path in Windows so it can download what it needs, first time take longer

[@tabby842]

Is your sidebar called comfui N sidebar? Repo looks different from what you show

[@pixaroma]

depends on the version, that is how it looked like a month ago, right now I am still using the old ui without the bar activated, maybe I will switch in the future to activate that

[@muhamadaliff4445]

why i cannot use open pose it said av_controlnetpreprocessor list index out of range .

[@pixaroma]

Can you try the first minutes from episode 10 to see iif is fixed? I replaced art venture with aio aux node

[@ПавелФамильевич]

Looks like xinsir-union doesn't works for openpose. I see that it tries to draw like a raw pose skeleton on the first iterations. Full xinsir openpose works fine

[@pixaroma]

i think only a few works, not sure if they have the full list somewhere on their page, just try and see what works

[@TUSHARGOPALKA-nj7jx]

Great, how to use controlnet with IP adapter?

[@pixaroma]

I didnt got to that part yet, maybe in future episodes.

[@TUSHARGOPALKA-nj7jx]

@@pixaroma any idea on how to use ip adapter as a controlnet, similar to how it was in a1111

[@TUSHARGOPALKA-nj7jx]

'NoneType' object has no attribute 'copy' - CR Multiconrolnet shows this error. How to fix this error? - It's because I am trying to use an ip adpater sdxl controlnet model with the .safetensors model you have.

[@pixaroma]

@@TUSHARGOPALKA-nj7jx is hard to tell from an error what could cause it, maybe check some recent videos for it kzbin.info?search_query=ip+adapter+control+net+sdxl+comfyui

[@DR34DL0RDD43DRA]

My openpose preporcessor is returning this error: "AV_ControlNetPreprocessor list index out of range"

[@DR34DL0RDD43DRA]

got prompt model_path is C:\ComfyUI\ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts\lllyasviel/Annotators\sk_model.pth model_path is C:\ComfyUI\ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts\lllyasviel/Annotators\sk_model2.pth Prompt executed in 0.43 seconds got prompt !!! Exception during processing !!! list index out of range Traceback (most recent call last): File "C:\ComfyUI\ComfyUI\execution.py", line 323, in execute output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\ComfyUI\ComfyUI\execution.py", line 198, in get_output_data return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\ComfyUI\ComfyUI\execution.py", line 169, in _map_node_over_list process_inputs(input_dict, i) File "C:\ComfyUI\ComfyUI\execution.py", line 158, in process_inputs results.append(getattr(obj, func)(**inputs)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\ComfyUI\ComfyUI\custom_nodes\ComfyUI-Art-Venture\modules\controlnet\__init__.py", line 136, in detect_controlnet image = apply_preprocessor(image, preprocessor, resolution=resolution) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\ComfyUI\ComfyUI\custom_nodes\ComfyUI-Art-Venture\modules\controlnet\__init__.py", line 43, in apply_preprocessor preprocessor_args.update({key: default_args[i + len(required_args)] for i, key in enumerate(optional_args)}) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\ComfyUI\ComfyUI\custom_nodes\ComfyUI-Art-Venture\modules\controlnet\__init__.py", line 43, in preprocessor_args.update({key: default_args[i + len(required_args)] for i, key in enumerate(optional_args)}) ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^ IndexError: list index out of range

[@pixaroma]

Are you using art venture node? Try try to use aio aux instead like i did at the beginning of episode 10

[@travisenglish8033]

I tried getting on your discord and it's not there invalid invite, anywhere I can download your workflows from?

[@pixaroma]

The ome from KZbin header should be always valid discord.com/invite/gggpkVgBf3

[@pixaroma]

@@travisenglish8033 try a different browser, i am on mobile now, i clicked on it and it worked, if you copy and paste the link make sure it doesn't have a space at the end of the link

[@travisenglish8033]

@@pixaroma I got in through the app rather then the link, or is that how I had to do it?... I'm in and thank you for your workflows

[@clay6440]

Can you use it on Forge?

[@pixaroma]

Not sure since they keep updating the forge so eventually will work but didn't test it

[@mojtabapourhady2355]

How can i use a mask to modify a specific part of an image in controlNet?

[@pixaroma]

Maybe use inpainting instead, i used that in episode 19, in episode 29 and used masks in 29

[@RDUBTutorial]

anyone know which version of comfyui-art-venture to install for MAC? - theres like 6 options to chooses from. Theres no mention on their repo page. ugh......OH nevermind I have the AIOAux node.

[@pixaroma]

Use aio from ep10 beginning

[@ronenthc]

I followed everything but I have really low quality output, blurry images, lot of noise, low details, I tried to lower the weights of controlnet but can't fix it

[@pixaroma]

Can you post screenshots in the comfyui chanel on discord, is hard to tell without seeing

[@klarWorks]

❤❤

[@Rachelcenter1]

Trying to find the workflows on your discord. It's hard to navigate this discord to find it.

[@pixaroma]

On the pixaroma-workflows channel, each episode has the workflow with episode in name

[@Rachelcenter1]

@@pixaroma I dont see that. I see welcome, off topic, comfyui, share ai art, share prompts, sd a111, sd forge ui, swarn ui, sd other versions, chatgpt, ai2play, share-ai-tutorials, pixaroma-shops, rules and feedback. I dont see anything that says workflows.

[@pixaroma]

I mentioned there on discord with a link click on it

[@Rachelcenter1]

@@pixaroma I did click the link, thats how i found your discord. and i hit accept on the invite, and I listed all the channels that I saw on the left. Why is "workflows" not one of them?

[@pixaroma]

@@Rachelcenter1 in the welcome channel i mentioned there with a link

[@ImHewg]

When I do OpenPose, I get this error with the preprocessor any idea why? AV_ControlNetPreprocessor list index out of range

[@pixaroma]

for me the problem for me was with art venture node, if you watch the first 4 minutes this is what I did, so you can try that to see if fix it, for me it works ok kzbin.info/www/bejne/j56ToH2mh8edarc

[@indigooo3d]

What's the minimum amount of VRAM to run this? It's possible to do it in a reasonable time in a 1060?

[@pixaroma]

Only if you test it, you can try with a faster sdxl model maybe like the hyper version and maybe a smaller image size to see if your system can handle it. I can not test it on smaller card since i dont have one from that series or similar. I only have from rtx series a 2060 and rtx 4090

[@kgottsman]

@@pixaroma Recently found your site an I love your tutorials. How much faster is the 4090 over the 2060? Ive got the 2060 and it seems to take forever.

[@pixaroma]

@@kgottsman is a huge difference depends on the model, for some like flux can take 6min for something that takes 15 sec on rtx4090, for sdxl lighting is less like 5 sec vs 30 second

[@kgottsman]

@@pixaroma maybe I will look for something in the middle. 3090 looks good and has plenty of memory.

[@GOSFILMFOND]

Смотрю с переводом на русский. Вы самый лучший учитель ComfyUI!!! Спасибо что делаете такие подробные обучающие видео.

[@pixaroma]

Thank you ☺️

[@Antilopax]

Hi there! Thank you for this great tutorial! Would you mind updating the Discord invite link? It seems to no longer be working..

[@pixaroma]

on some older videos is still the old link, the new links is here discord.com/invite/gggpkVgBf3 for some reason it expired, but on the channel under description it should be always active one

[@jasonstetsonofficial]

How to use in Forge?? That is important

[@pixaroma]

I didnt play yet in forge with it because they still are working on the interface and have bugs, I am waiting for the interface o be more stable, I see updates like every hour even minutes sometimes. And they have control net integrated. That why I did for comfyui since is more stable.

[@anusrianusri8946]

I cant join your discord it shows invite invalid ☹

[@pixaroma]

The one from header of KZbin channel always work try that, try different browser or mobile application

[@longvu5341]

I have a problem with Preprocessor list. With canny: everything ok but with others: I can't Queue, it show a error table. [Errno 2] No such file or directory: 'D:\\Long\\set up\\ComfyUI_windows_portable_nvidia_cu121_or_cpu\\ComfyUI_windows_portable\\ComfyUI\\custom_nodes\\comfyui_controlnet_aux\\ckpts\\lllyasviel\\Annotators\\.huggingface\\download\\sk_model.pth.c686ced2a666b4850b4bb6ccf0748031c3eda9f822de73a34b8979970d90f0c6.incomplete Anyone can help? Thank you.

[@pixaroma]

Enable long path in windows, it can not download the model because of 256 characters on path limit, you can search on google how to enable that or I have on discord on comfyui-faq channel some steps you can try to enable long path, then try again and should be able to download it. Also make sure you check the beginning of episode10 because i used a different node instead of art venture

[@willmacneil815]

Hi there. I'm running into an issue with preprocessors. The AV preprocessor node loads but only shows options for Tile (no canny, depth etc). And the AIO Aux prepocessor node fails to load from the Manager. Any ideas? Thanks for your help.

[@pixaroma]

I am using aio aux now, when you say it fails to load like the node itself didn't install or ? Installed but that av don't work? Also control net union don't think it have AV

[@willmacneil815]

@@pixaroma I'm getting an error in the Manager saying IMPORT FAILED. Using TRY FIX doesn't seem to solve it.

[@pixaroma]

@@willmacneil815 try to click on the name to go to their page and install manually with git in custom nodes folder, also update comfyui

[@pixaroma]

@@willmacneil815 can you enter on discord and mention pixaroma there? so i can send you some screenshots and how to do it

[@pixaroma]

@@willmacneil815 Here I tried to make it step by step for you In ComfyUI folder, go to the custom_nodes folder ..ComfyUI\custom_nodes then when you are inside that folder you can see all the nodes installed. In the address bar type cmd and press enter, it should open the command window in that folder then paste this command and press enter git clone github.com/Fannovel16/comfyui_controlnet_aux wait for it to finish then you can close that window. Now restart ComfyUI. You should be able to use it now If you get a git not recognized error install git first git-scm.com/downloads

[@aneebartist7207]

They changed apply controlnet

[@pixaroma]

Yes, i updated the workflows check on discord I updated the nodes

[@JarppaGuru]

3:55 better say need install that too. and wtf is that search

[@pixaroma]

I cut by accident a few frames, drag a link and click search and that appears

[@maniacos9620]

The "Comfy" in "CompfyUI" is sarcasm, right?

[@pixaroma]

Probably 😀 at first is uncomfy, but after that it is quite nice, probably they should have use other name like modular ui, or IKEA ui 😂 diy ui

[@AIChandu77]

for flux

[@pixaroma]

In 15 minutes will be live, control net for flux episode 14 :)

[@pradeeshnagarajan1553]

Hi there. i have some errors loading preprocessors, i tried most of the preprocessor but it shows error except canny & lineart. and some are showing "list index out of range". and i have attached the error which shown in my comfy ui. that's " AV_ControlNetPreprocessor" [Errno 2] No such file or directory: 'D:\\software\\ComfyUI_windows_portable_nvidia\\ComfyUI_windows_portable\\ComfyUI\\custom_nodes\\comfyui_controlnet_aux\\ckpts\\lllyasviel\\Annotators\\.cache\\huggingface\\download\\mlsd_large_512_fp32.pth.5696f168eb2c30d4374bbfd45436f7415bb4d88da29bea97eea0101520fba082.incomplete' can you please help with it. thank you

[@pixaroma]

yes the model didnt download, most of the time that happens because is not activated long path in windows, I posted on discord how to fix that. You can google it. And try again. If still dont work you can download the model manually and put it in that folder, but probably you will get same error when you try to use any model since it can not download them because path is too long. In your error the model is mlsd_large_512_fp32.pth and you download it and put it in annotators folder ..ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts\lllyasviel\Annotators download link huggingface.co/lllyasviel/ControlNet/blob/main/annotator/ckpts/mlsd_large_512_fp32.pth

[@pradeeshnagarajan1553]

@@pixaroma you are great man. im an architect and your videos are so good , you are making the videos that are very easy to understand and learn, I'm really eager to do more after seeing your videos, I can see how much effort you’ve put into this. It’s truly impressive and the preprocessor is working really fine, thank you so much.

[@scarsandlight]

i got an error when i use 'depth' and some other preprocessor also, any solution?

[@pixaroma]

can you post on discord screenshot with error and the workflow you are using, in the comfyui channel