The Ultimate ControlNet Tutorial: Understanding UNet, Exploring Pro Max, & Kolor's New Model

---------------------------

Welcome back to the ComfyUI Ultimate Guide series!
Start your ControlNet journey here! This comprehensive guide covers everything from the fundamentals of UNet, the backbone of Stable Diffusion, to mastering the advanced All-In-One Pro Max model. Explore the exciting new Kolor ControlNet model and learn how to harness these tools for precise and creative AI art generation. This video is perfect for beginners looking for a solid foundation and experienced artists seeking to expand their skills.
-------------------------
extensions:
ComfyUI-Advanced-ControlNet

----------------------------------
00:00 ControlNet
00:48 what is ControlNet
02:08 Difference between ControlNet models
02:33 openpose controlnet
04:04 normal map controlnet
04:48 semantic segmentation
06:11 Depth ControlNet
06:18 Tile ControlNet
07:26 QR code ControlNet
10:04 qulity of ControlNet
12:04 Quality evaluation of ControlNet
12:50 Download Controlnet models
16:50 Kolor ControlNet highest quality
18:51 Controlnet preprocessors
20:43 How to work with controlnets
22:20 default nodes for loading ControlNet
26:23 Apply ControlNet setting
29:26 load multi ControlNet
34:55 ComfyUI-Advanced-ControlNet Extention
39:49 Pro setting, replication of the "My prompt is more important"
42:28 UNet deep understanding of different layers and weights
47:49 ProMax Union Controlnet for SDXL, All in One
49:07 ProMax Union Controlnet+Defualt nodes
51:59 ProMax Union Controlnet + Advance ControlNet Extention
54:04 It is not compatible with perturbed attention guidance!
----------------------------------
#controlnet
#unionControlNet
#ComfyUIWorkflow
#ComfyUICommunity
#ComfyUITutorial
#ComfyUITips
#AIArtist
#DigitalArt
#StableDiffusion
#GenerativeArt
#AIArtCommunity
#ProductivityHacks
#CreativeWorkflow
#ArtTools
#AI
#MachineLearning
#DeepLearning
#NeuralNetworks

[АлексейЮжаков-лш]

Thank you for the lessons, you give a deeper understanding of working with neural networks and help solve applied problems

[bgtubber]

Excellent lecture! More of these, please! It's always fascinating to learn how things actually work under the hood with examples. You're approach is great. You don't just tell people "do this and that and you're done". You actually explain things like a good teacher. You and Matteo (Latent Vision) are my favorite Stable Diffusion/ComfyUI youtubers. I never miss a video from you, guys. 👍

[inspiredbylife]

Thank you for this video!!! Very useful

[dominichawgood]

A very clear explanation and some useful tips that other people tend to miss out. Thanks!

[mrasa]

Wow! Comprehensive, very useful, interesting and visual!
What software did you use to present Unet in 3d ?

Also, it would be interesting to know if the same t2i adapters differ in any significant way from standard controlnet.
Plus there is also the coadapter (also for 1.5 models) I think they were trained a bit differently, but there are usable models. For example there is colors, it is much better at rendering colors compared to tile and inpaint models

[bgtubber]

Not sure about the other controlnets, but from my experience, the SD 1.5 tile controlnet is still unbeatable to this day. I've tested pretty much all tile controlnets, including 2-3 tile controlnets for SDXL and the latest one for Flux. SD 1.5 still seems to produce the best creative image with the richest detail. If anyone cares to explain why, I'd be interested to hear. I would expect the Flux or SDXL tile controlnets to be the best, but that's not really the case in my experience.

[TUSHARGOPALKA-njjx]

Is there a difference is speed if multi controlnet node for multi controlent in your workflow compared to using single controlnet and combining them one by them. Doesn't that increase the dimensions of the vector in latent space?

[nkofr]

nice, very interesting! I'm trying to understand why we would keep usind SD1.5 controlnets vs SDXL?
also, isn't the "soft weights" for 'my prompt is more important' the same thing as lowering the strength of the controlnet? what's the difference?
Cheers!

[TUSHARGOPALKA-njjx]

How can we combine controlnet with inpaint - since that usually becomes too constrained and the somehow the model is unable to follow the prompt better

[WhatsThisStickyStuff]

Are the kolors controlnets better than using the xinsir controlnet with the adapter for kolors?