Aitrepreneur patreon

ULTIMATE ALL-IN-ONE FLUX WORKFLOW V1

🕑 Added 2024-09-21 19:06:55 +0000 UTC

Comments

Roybot

same here

Joelle LB

getting a missing node error for workflowFlux Controlnet & workflowXlabs Sampler but neither can be found when I search missing nodes or manually search (I did update ComfyUI)

Joelle LB

I did that but nothing appears in missing custom nodes and those nodes also dont show up when I manually search. Where do I get them?

Aitrepreneur

This is only for Flux, you'll need different workflows for different model architectures

Soda Sullivan

Once we have this UI installed (and thank you Overlord for making it easy!), we can use various workflows by following the same instructions, yes? Flux is okay, But i am more interested in a Flux.d and Pony checkpoints.

Matheus Andrade Viana

Thanks! That was exactly the problem. Just formatted my PC and forgot to reinstall git.

Aitrepreneur

probably because you don't have git for windows installed already so it couldn't have clone the manager github, it's an easy fix, just dl git for windows and install it: https://github.com/git-for-windows/git/releases/download/v2.47.0.windows.1/Git-2.47.0-64-bit.exe then go to the ComfyUI_windows_portable\ComfyUI\custom_nodes folder click on the folder path then type cmd press enter and in the command prompt window that just appeared you're gonna type: git clone https://github.com/ltdrdata/ComfyUI-Manager.git then press enter now if you reopen comfyui the manager will be here

Matheus Andrade Viana

There is no "Manager" nor "Share" buttons in my ComfyUI. Did it get removed, or did I do something wrong? How do I update ComfyUI and Install Missing Custom Nodes without getting to the manager?

Ramon Rodgers

I fixed it. I just needed to finish the updates and for some reason your workflow was changed "unsaved" after. So I removed and re-added after everything was installed and updated. The problem went away.

Aitrepreneur

Always the same issue whenever you load a brand new workflow, read the patreon post. 1. Update comfyui 2. Click on "Install Missing Custom Nodes" and click on the first small checkbox to select all the nodes and then click install. 3. It will then ask you to restart, so click on that restart button, wait for the webui to reboot 4. Click the refresh button (F5) 5. If the message is still there but all nodes are installed, just clear the workflow and reload the json file inside

Ramon Rodgers

I think I missed a step maybe? Some workflow boxes are still red. "workflowFlorence", "workflowFlux Controlnet", "workflowNodes 3", and "workflowXLabs Sampler"

Dario van Houwelingen

Amazing work, thanks for sharing! Would love to see a proper IPadapter workflow if that's possible. Thanks again!

TomiTom1234

Is it possible to update the workflow to add PuLID model?

Miguel Bzh

hello ! after installing, I don't have the Manager button on the menu, any solutions ?

Soda Sullivan

works great! would love to find as simple aworkflow for Pony, these workflows with 30-50 nodes are INSANE

Michael O'Neill

Perhaps you should take " girl " out of your prompts and replace it with " woman"

Sabsabinx

Same here too.

Archer

Same. Any fix?

Miguel Arce

the red node is for the upscale workflow btw

Miguel Arce

I'm getting the error of missing workflowflorence, how can i install it? manager it's not working for that, and the github link not working, thank you!

Miguel Arce

same, did you fix it?

Miguel Arce

same, did you fix it?

Tim Eilers

missing these nodes, and the manager isn't installing them- workflowFlorence workflowFlux Controlnet workflowXLabs Sampler

Otniel Ocampo

Yep for testing I installed The Workflow on a Fresh windows install and something changed with the permissions of ComfyUI that does not allow the automatic install of Custom nodes, is not only those if you go to the Custom Node Manager you'll see that there is atleast 8 custom nodes that show as "Try Fix" "Disable" "Uninstall" Maybe because of ComfyUI-GGUF and x-flux-comfyui not installing

Aitrepreneur

make sure that you updated comfyui first, then uninstall the node, restart the webui, then click install again, finally restart the webui and refresh otherwise if it still fails, you can try this: First uninstall the ComfyUI-GGUF node then. Go inside the ComfyUI_windows_portable folder then click on the folder path, type cmd then press enter, this will bring the command window and inside type: git clone https://github.com/city96/ComfyUI-GGUF ComfyUI/custom_nodes/ComfyUI-GGUF .\python_embeded\python.exe -s -m pip install -r .\ComfyUI\custom_nodes\ComfyUI-GGUF\requirements.txt then press enter. Then relaunch the webui.

Raevn Eldaer

Its not working. After updating missing nodes I still get this; When loading the graph, the following node types were not found: FluxGuidance (In group node 'workflow/positive prompt guidance') workflow/positive prompt guidance workflow/Negative Prompt ModelSamplingFlux (In group node 'workflow/Other Nodes') workflow/Other Nodes DualCLIPLoaderGGUF UnetLoaderGGUF Nodes that have failed to load will show as red on the graph.

James F

this all in one is failing for me, I am getting Import Failed ComfuUI-GGUF and import failed x-flux-comfyui ( set config.ini security level to weak to install the nodes. yet still failing please help

Panda

Hi, I think there is a "prompt improver" inside thisWorkflow but I don't remember where exactly and it's messing up my prompts. I'm putting "IMG_1028.JPEG "Beautiful 20 year old girl portrait" and "Adult_girl.JPEG "Beautiful 20 year old girl portrait" and it's giving me a child, which is very creepy

Aaron1611

I thought I tried every model. Are the correct models in a folder I'm unaware of or do I need to fetch them? I selected the different models which were available.

Salvador Robles

Hello, I wonder if it would be possible to add negative prompts to the workflow. Thank you.

teste de ia teste

Hi everyone! I’m trying to figure out if it’s possible to modify my workflow in ComfyUI to use multiple different prompts and generate an image for each prompt automatically. I’ve managed to get it working with a single prompt, but I would like to feed a list of prompts and have it generate individual images for each. I tried using the "Inspire Pack", but it only generates one image, even when I attempt to use multiple prompts. I also tried an older method using KSampler, which does generate multiple images, but the quality is not as good as I’d like. Has anyone managed to do this, or does anyone have any ideas on how to set up the workflow to automatically generate multiple images from a list of prompts? I’d really appreciate any help or suggestions you can offer! Thanks a lot! :)

THEO-Ed animation

Never mind; I tore down the version of ComfyUI I already had installed, and re-ran your program from scratch. It's all good. Thanks so much for a brilliant layout!

THEO-Ed animation

I keep getting a whole lot of error messages on startup, and most of the individual sections have bright red nodes, indicating their unusability. I think it boils down to the fact that I can't seem to import the ComfyUI-GGUF custom node, but I don't know what to do to get it to download. Any recommendations?

Josh inmn

Wow not sure if I should try it sound like a lot of errors and no support

Nabil Boulezaz

When V2 is out ?

Dumitru Ploscar

Ok/ Now Ive dwonloaded them and it gives "Prompt outputs failed validation VAELoader: - Value not in list: vae_name: 'ae.safetensors' not in ['taesd', 'taesdxl', 'taesd3', 'taef1']

Dumitru Ploscar

Installed the first option (q4 one) manually and Im getting error: "Prompt outputs failed validation DualCLIPLoaderGGUF: - Required input is missing: clip_name2 - Required input is missing: clip_name1" Also discord link is dead

Dumitru Ploscar

From where do we download the clips? Ive used the q4 option and Ive got same error. Now Ive tried to add them and it says that they are missing/null

LtCzr

I am using a rtx 4090, and when i tried using the comfy ui workflow after installing it it looks like it always uses a large about of vram and goes into shared memory, I am running gguf and with 1 lora and i looks like it takes more than 24gb of vram. Do I need to wait for it to complete load for it to be manageable and run-able? it looks like after the first generation if i set it to 20 steps it takes 14 minutes. but if i set it to 1-4 steps it finishes then reduces to aroun 17-19gb of vram and runs smoothly from then on regardless of steps.

Ken L. Black

I FINALLY got around to trying to run this, using your sample prompt for <12VRAM within the GGUF Image Generation node that you created. I get the following message when I click on QUEUE: Prompt outputs failed validation DualCLIPLoaderGGUF: - Value not in list: clip_name2: 't5-v1_1-xxl-encoder-Q8_0.gguf' not in ['ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors', 't5-v1_1-xxl-encoder-Q3_K_L.gguf', 't5xxl_fp16.safetensors', 't5xxl_fp8_e4m3fn.safetensors'] I even ran the separate MODELS bat file that you had just to make sure I was downloading everything required, and that still did not work. Any ideas what this is referring to? Thanks for any information.

Gio

Thanks a lot for this! But your workflow is returning an alert saying "Loop (510,512) - not submitting workflow" when clicking on "Queue", it's odd that no one else is talking about it.

Otaku_Hedgehog

Ok thank you. I'm using Aitrepreneur's workflow though and I didn't really understand where to connect my Negative Prompt Node to. I'll take another look at it with your suggestions when I get a chance.

Aitrepreneur

Yeah because you haven't selected the right models, if you chose option 2, you need to change the q8 model selected by default in the workflow witht the ones that were downloaded

Aaron1611

Again, same thing. Uninstalled, reinstalled, my screen looks exactly like your screen, and I get this- Prompt outputs failed validation UnetLoaderGGUF: - Value not in list: unet_name: 'flux1-dev-Q8_0.gguf' not in ['flux1-dev-Q5_K_S.gguf', 'flux1-schnell-Q5_K_S.gguf'] DualCLIPLoaderGGUF: - Value not in list: clip_name2: 't5-v1_1-xxl-encoder-Q8_0.gguf' not in ['ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors', 't5-v1_1-xxl-encoder-Q5_K_M.gguf', 't5xxl_fp16.safetensors']

ERock

D'oh! Well, adding a negative prompt in ComfyUI is like adding a "no thanks" to the stuff you don’t want in your picture! *Mmm... filters...* Here’s how you do it: 1. **Start with Your Positive Prompt Node:** You’ve got your positive prompt set up, right? That’s where you tell the UI what you want in the picture. *Mmm... donuts in every picture...* 2. **Add a Negative Prompt Node:** Now, you add a negative prompt node. It’s like telling the UI, "D'oh! Don’t add that!" *No broccoli in my pictures, please!* 3. **Connect It to the Main Graph:** You gotta connect that negative prompt node to your main graph, just like you’d connect the donuts to my mouth. Make sure it’s linked properly, so the UI knows what to leave out. 4. **Run It:** Finally, hit that “run” button and watch the magic happen—minus the stuff you don’t want! *Mmm... perfect pictures...* If you run into any trouble, just let me know. I’m here to help... or at least to try not to mess things up too much. D'oh!

ERock

How do I add this workflow to Open WebUI? Specifically what do I put for the ComfyUI Workflow Nodes in the Settings -> Images tab?

Horst

Hey there, thanks a lot for this lot of work you did. Im currently facing the issue, that I want to use my own loras. Here I put the files inside the model/lora folder, but I cant refresh the comfy ui or I do not find them when Im inside the lora module. How can I use own loras?

Mathilde Leroux

Could you please update the workflow to support region based prompting such as regional prompter? It would be a great help!!!

Lenny Gordon

Phenomenal work on this. If you can add in a section for animation that would be amazing.

Ali fi

hi can you help me for this TypeError: Cannot read properties of undefined (reading 'calcNodeInputs')

Joe Engelmann

I have this working flawlessly on arch linux and dual amd 6900xt's thank you so much for your hard work! I do have one addition request for the workflow though, I would love to see a section for either svd or other t2v/i2v, there are plenty of platforms out there for this that offer great results but none that offer the control comfy can offer. As well, if others like myself don't care to much about the speed of the render it's perfectly possible to render much longer videos AND stay within the confines of 16-32gig of video memory

Otaku_Hedgehog

How can we use negative prompts for the GGUF IMAGE GENERATION section?

Gael Pirate Life

I have tried with this guide but I can't make it: https://www.andreszsogon.com/comfyui-load-prompts-from-text-file-sample-workflow/

Gael Pirate Life

Thank you for the workflow(s)! Can you make a variation for the text 2image where we can load prompts from a text file? So it automatically generate the all the images?

Aaron1611

I deleted everything and re-installed using option 1. I have 12 gb of VRAM so I had started with option 2. I tried the installed models, flux1-dev-Q4_K_S.gguf , and got this message when I clicked Queue - Prompt outputs failed validation DualCLIPLoaderGGUF: - Value not in list: clip_name2: 't5-v1_1-xxl-encoder-Q8_0.gguf' not in ['ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors', 't5-v1_1-xxl-encoder-Q3_K_L.gguf', 't5xxl_fp16.safetensors'] Then flux1-schnell-Q4_K_S.gguf and got - Prompt outputs failed validation DualCLIPLoaderGGUF: - Value not in list: clip_name2: 't5-v1_1-xxl-encoder-Q8_0.gguf' not in ['ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors', 't5-v1_1-xxl-encoder-Q3_K_L.gguf', 't5xxl_fp16.safetensors'] Could it be my AMD processor? Processor AMD Ryzen 7 5800X 8-Core Processor, 3801 Mhz, 8 Core(s), 16 Logical Processor(s) Or my board? BaseBoard Manufacturer ASUSTeK COMPUTER INC. BaseBoard Product TUF GAMING X570-PLUS (WI-FI) My GPU is NVidia

Andres Ospina

PLEASE NOTE: For those that are not seeing the Manager you have to have "git for windows" installed before running the One-Click Installer

Lagartija

Awesome! would love a face swap addition

Baubleheadz

Same. I also tried manually deleting ComfyUI-GGUF folder and reinstalling it using the git commands from the ComfyUI-GGUF github. But it keeps coming up missing and try to fix this wont work.

SUTY THE REAL

you need to download the ComfyUI-Manager from github Then you will get the Manager button

gne gni

why on my confy i dont have the botton mannger

Otaku_Hedgehog

Thank you so much for making this workflow! I've been studying ComfyUI all week and your video and workflow makes using the program a whole lot easier! ^_^

Aaron K

Tried the Python dependencies but its not fixed it for me sadly.

Jannes Pelser

would be great if the concept could be expanded to a larger scale, things like ultimate upscale (a la magnific / supir) for example. also Segment Anything. Using these kinds of workflows are on openart.ai yes, but when you try to run them they're most often broken or missing nodes. The ultimate tool to see from this workflow is if you could copy what playground.com has started doing, seemingly a more advanced version of InstructPix2Pix (or img2img alternative test)

Ricardo Sousa

just rewatched the video end re-read the instructions. where did you find how to install that plugin?

MyBrotherInChrist

Thank you Angel. That worked for me!

James Stewart

Cool, maybe a Basic 101 intro to Comfy settings that suit the 80/20 rule then? But you've probably got so much on the books at the moment. If you get to it sometime, sweet.

Dreddoom

Hi. I love the ultimate workflow. I just added the Fast Groups Bypasser node by rgthree. It automatically adds switches to turn the groups on and off. I'm a noob but this seems easier 😉

Gary

Awesome Workflow! Thanks. Got a question. The GGUF text to image workflow doesn't seem to have a CFG parameter, which is quite crucial in getting accurate text. Am I missing something?

Dallas

I installed with the 1 click installer and All-In-One. Im having an issue where the Unet Loader isn't showing any models. It's undefined. When I look in the models/unet it has stuff their but they aren't appearing in app. Any help?

D'angelo Rodriguez

add ipadapter please

Angel Barba

Update: Update python dependencies in the update folder and run again

Nomsworthy

Same

Sasho Andrijeski

Thanks a lot for your work! It is amazing :) After importing the flow, I'm getting this error: Cannot execute because a node is missing the class_type property.: Node ID '#364' Can't figure out where the issue is. Any help?

Aaron K

Am having the same issue too :(

Angel Barba

Hello everyone! I have an issue, I used the automatic installer and after installing the custom nodes I get an error on 2: ComfyUI-GGUF and x-flux-comfyui (IMPORT FAILED) And cant seem to fix it by "try to fix this" Can anyone help me? thanks!

Aitrepreneur

Might be a good idea indeed

Aitrepreneur

press F5 on your keyboard to refresh and/or use an different browser

Aitrepreneur

Thanks. 1) It's coming 2) I'll put it on the list 3) This is done with the "noise seed" and the "control_after_generate" option

Aitrepreneur

which install option did you choose? If you chose anything other than "3" then just click on the model name and chose whatever was installed, same thing for the rest of the text encoders

Hans

Just refresh the browser window

Dreddoom

Works great, Thanks

Lance Winder

Thanks for the incredibly detailed explanations. Love your work.

Ludovic Carceles

This is great stuff, as usual. The only thing I added, and you can add to the future version is the Fast Groups Muter from rgthree

Basilio Soto

After installing all the necessary files for the workflow Comfyui opens to a white window? any help

Aaron1611

The duelcliploader says there are clips, but I don't really know what that means.

Top6

Face swap yes!

Firmy Van Bas

Amazing! I would definitely like to see an IP adapter, a swap face and a lineart control net option. Thank you so much!

Jim Gale

this is quite excellent. I'm only missing a face replacer flow (unless I'm missing it there already)

Terr0rClown

you didnt install the clips

James Stewart

Mate, you are the man! Thanks for thinking about your community and coming up with a Flux solution for those of us watching other Youtuber vid's about comfy and going "That is such a mess". Thank you, thank you, thank you for an elegant solution! My suggestions for the future: 1. How to create Lora's for Flux (there's a bit already about FluxGym online but your take is always loved by your folks!) 2. Would love to see you add ReActor into your Patreon workflow as I believe it works in Comfy 3. Weird question, but how do you randomize a seed in Comfy. Doesn't seem clear I have one more about the Flux Prompt Generator but will post there. Again, love your work!

Joe Engelmann

thank you, the manager plugin fixed my issues.

Semon Xue

Very cool workflow and video ,thank you !

Joe Engelmann

this is awsome, i do have one request though, i don't suppose there is a way to get an sh file for local installs to grab the requirements for the workflow? i have been trying to get all the requirements installed so i don't have so many red sections but following the manual install, i can't seem to load much using the new workflow, the flux-fast-lowvram or others work though.

RB

you need to install the comfy manager plugin

Ricardo Sousa

Was anyone able to install it from zero with the manual install guide? I get red boxes and no Manager Button. :( such a shame the video only instructs how to use the payed file :'(

Aaron1611

I gots - Prompt outputs failed validation DualCLIPLoaderGGUF: - Value not in list: clip_name2: 't5-v1_1-xxl-encoder-Q8_0.gguf' not in ['ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors', 't5-v1_1-xxl-encoder-Q5_K_M.gguf', 't5xxl_fp16.safetensors']

CCookiess

This is awesome. Would love to see additional fields for SD1.5 and SDXL/Pony so it could be the ultimate All-In-One prompt interface. Many workflows combine some or all of these models due to specific strengths/weaknesses, and availability of loras and specific prompts.

Daniel Cuche

can you include PulID? actually I wonder what do you think is the best models todo either 1. Generate an image with the likeness of another person from a photo reference.. and 2. Modify the face on an input iamge.. basically apply deepfake XD..