If you have any questions you can post them in Pixaroma Community Group facebook.com/groups/pixaromacrafts/ or Pixaroma Discord Server discord.gg/gggpkVgBf3
Excellent tutorials. You are a natural at this. I've taken courses at Udemy, LinkedIn Learning, HubSpot Academy and Google Skillshop and yours are by far some of the best. For someone who is just doing this solo, you're awesome. Definitely going to donate at the end of this series! Your "baking a cake" analogy and reorg of the nodes to match Auto1111 were simply brilliant.
The mark of how well someone understands a concept lies in how simply they can convey understanding to a layman. You clearly understand this well, thank you for sharing that with all of us
Wow! Amazing tutorial and your ability to explain complex concepts to simple understandable themes is a gift. You are gifted with teaching and thank you so much for this tutorial.
Thank You...! Your workflow shows the process clearly, New to Gomfy and I like to understand, what I'm doing... This video helped, this series is so promising...looking forrward to more...!
You are by far my fav SD teacher! Coming from A111 and forge i'v been dreading the comfy UI plunge! But this series is already having me pulling spaghettis! Thank you for this!
Great tutorial. Most talk about nodes only without explaining what they do and why they are there. To really go from A1111 and others, understanding clip, vae, checkpoints, and how they connect is necessary.
OMG Man, I can not express how amazing your tutorial is! I am a motion designer for 20 years now, but I was scared to use AI for a while. Now, watching your tutorial makes me LOVE ComfyUI already, even I only started it one day ago. :D So, THANK YOU!
i am a grandpa that recently became interested in AI art. tbh i was not even going to attempt to use comfyUI but because of this tutorial im all in. all the sudden it does not seem so overwhelming thank you.
amazing simple and thorough explanation, wonderful tutorial, i recommend everybody to watch it if you want to learn fast about ComfyUI and Diffusers internals top view, loved it!
I've been trying to get into Comfy, but was always overwhelmed. This was really explained well for a forge user myself, thank you for this. The the idea of moving the nodes in place to mimic the forge/A1111 UI, makes it so much easier to transition to Comf, Cheers
Thanks for the info man. I did learn a lot, but being new to the system, I was still confused. So just a little feedback and clarity for anyone in the same boat. When you are using an analogy, make sure you bring it back to what it is the ai is actually doing and why. I took a little break while you were prepping for dessert and had to look it up. This is what made it click with me……. The whole software is based off of 3 main categories with an input (encoded text prompt/image/etc) and an output (decoded end result). A model (Stable diffusion..etc), a control (posing, style, idadapters, etc) , and a view (stage/latent space). Your analogy does make sense though, it just needed the context to the topic of interest. Understanding the “how” really puts the “why” in perspective. The idea that these models are literally generating things out of noise is just insane to me. Anyway… keep it up man, anyone spreading the love, should get love….I’m just giving you my 2c from one viewer. Good luck with the channel!
Great job! Great breakdown! These instructions will help me a lot :-) thank you! I love your bakery breakdown… It really helps me it already has helped me to understand more effectively what my image workflow is accomplishing each step of the way, and most importantly I can now learn how to troubleshoot and tweakmy outputs! Very excited about that prospect, because up until now I've just been dropping workflows like a black box hoping that they have the components that I wanted to accomplish… Thank you so much! Dan Hawkins
I think, learning by you is pure joy. Clear, good speed, well-built. You definitely have talent to do it. - Okay, my question: the Flux is here and we can use it with Comfy and Forge as well. (Probably the Fooocus version will also be done soon.) What is it better with, Comfy or Forge? What is your thought?
I got similar speed on both, is just a mother of preference just sometimes on lastest versions both interferences can have bugs, like it happened to me with both. Like on one version of forge it took 5-6 minutes to generate and after an hour on an update it took 27 seconds. Same with comfyui on an update it started crashing my comfyui and didn't follow the prompt. So i just have both and use both of them each in its own folder. Forge js easier to use, but on comfyui you can do more and updates are way faster, on forge it takes a few days or weeks to get the new stuff, like I was with Flux comfy got it to work from day one
another way to think aboutempty latent image node could be that It makes a blank canvas we can decide there the size of our canvas and then the ksampler paints on the given canvas
Great guide, as always, thanks. Will there be a new version of your episode: Instant Art Upgrade: 260+ Stable Diffusion Styles for A1111 & Forge? Is there such a NODE? Is👍
not sure it is, but if it is I will do a video for it, I plan to do the things I did in forge also for comfyui if is possible, but I am still learning more advance stuff also, seems that you can do really a lot of stuff with it
Maybe in later episodes, i still have more to talk about in the basics so later when we do more complex it will be easier to make things work and fix problems
Great video Decean, as always. Question for you, the speed of Forge seems to have come from Comfy UI (as i've read), does this mean that since Forge works well for me, that ComfyUi will as well? I have 6g VRAM. Keep the awesome content coming!
I just tested today on rtx2060 with 6gb of vram, you can do basic generation, not sure if control net works, the upscale is a little slow, but you use a model that is fast like juggernaut hyper i can do a generation in like 12 seconds because it doesnt need many steps. You can check on my facebook group i posted like a few minutes ago.
It finally clicked for me why all those nodes are needed. Now to go bake some sweet, sweet AI art. Anyone else thinking of trying out that Juggernaut model after this?
Thank you! You are like extremely good at this, like your whole approach is amazing, hopefully my words aren't going to affect your ego, lol, just kidding... But yeah, you have multiple talents combined in what you bring forward. Thanks again, 🙂
I didn't use it yet because i learn it step by step and first i work with the nodes that comes with comfyui and slowly i install new custom nodes. I saw many people using them so I will probably use them also
does comfy ui have the same features about auto1111, with regards to add script text file on th e positive prompt? where you can input many prompts variations on one button?
Found this also, see if it helps, I am not at the computer to test it github.com/comfyanonymous/ComfyUI/discussions/896 in the comments someone posted a custom node
If it doesn't say the value there there are probably the same model on civit ai that has the vales somewhere in description or open a ready made workflow that comes with recommend settings. For example like in episode 5, or do a search on Reddit and probably can find someone asking the same question and got an answer
The "Clipspace" button in ComfyUI's queue sidebar is a tool that allows you to manage temporary data used within the UI, particularly when editing images or masks. Essentially, Clipspace acts as a clipboard for storing and manipulating images or other data. When you perform certain operations, like using the mask editor, the image data is copied to Clipspace, where you can modify it. After editing, the changes are saved back to Clipspace and can then be applied to the original node or used elsewhere in your workflow. check also the comment on this www.reddit.com/r/comfyui/comments/1cql2i1/what_does_the_clipspace_button_do/
Hello, I'm having problems making the workflow in this tutorial works with the following Checkpoint Model: realflux10b_10bTransformerDev.safetensors. The Compact Version, which I'm using, already includes VAE, Clip-L, and T5XXL. Any tips?
is hard to tell since I didnt use that model, but probably needs to be in the right folder, so first you must see which folder comfyui see it, and which node needs to load that type of model. Post some screenshots maybe in the discord server on the comfyui channel discord.com/invite/gggpkVgBf3
salut și bravo pt tutoriale și canal în general! Vin la timp ca tocmai ce am început să învăț comfy. Am o dilemă majoră și poate îmi poți da un sfat. Lucrez pe mac de 15 ani iar acum am un macbook m1 max dar e lent pt Ai în general și vreau sa fac un desktop pc. Am tot făcut research care brand de 4090 este mai reliable și nu prea reușesc să ajung la o concluzie deoarece văd ca sunt probleme cu pcb-ul sau connector-ul acestor placi variând de la brand la brand. Mă poți ajuta cu un sfat? Merci mult și baftă! ps: scuza de romenglish, sunt plecat de ceva vreme din țară și îmi găsesc mai greu cuvintele
Nustiu care is mai bune, eu am placa asta video de un an si pana acum e ok. Eu l-am facut la comanda, si i-am zis eu la ala de probleme cu conectori, a zis sa nu imi fac griji ca imi pune o carcasa mare si are loc sa bage mufa cum trebuie, numa daca nu e bagata bine face scurt, sau ceva de genu, acum ce sa zic, la noroc :) GPU GIGABYTE AORUS GeForce RTX 4090 MASTER 24GB GDDR6X 384-bit
I think ComfyUI user are subconsciously regards it as a game... Because thats how I treat it... I have have no image objective yet I love tweaking and experimenting different combinations... Master Olivio Sarikas said imagine painting on a canvas analogy... or you baking a cake.... for me its Steve Factory Manager 😁🤓... Im gonna be yapping about minecraft again... ComfyUI has those Freedom and Creativity essence which is the reason minecraft are so damn popular back then...
I struggled with creating groups in comfy then I found that the best way to create a group is to select all nodes for that group then right click on an empty area then select create group for selected nodes
If you have any questions you can post them in Pixaroma Community Group facebook.com/groups/pixaromacrafts/
or Pixaroma Discord Server discord.gg/gggpkVgBf3
As a professor, I have to say, you are a natural born teacher, and the baker analogy was excellent.
This is super clear and so well organized! Thank you!
Crisp clear voice. Like it
Excellent tutorials. You are a natural at this. I've taken courses at Udemy, LinkedIn Learning, HubSpot Academy and Google Skillshop and yours are by far some of the best. For someone who is just doing this solo, you're awesome. Definitely going to donate at the end of this series!
Your "baking a cake" analogy and reorg of the nodes to match Auto1111 were simply brilliant.
Thank you ☺️
The mark of how well someone understands a concept lies in how simply they can convey understanding to a layman. You clearly understand this well, thank you for sharing that with all of us
thank you 🙂
Love how simply you explain how the components work. Thanks! 💯
This class so far has been so much fun. Looking forward to the rest of this class. Thank you for your knowledge and incredible teaching skills!
Glad you enjoy it 🙂
This is BY FAR the most clear comfyUI series I've ever watched. THANK YOU SO MUCH!
you are welcome 🙂
Akhirnya ketemu juga. Materi training yang bagus, komprehensif.
Terima kasih telah membagikan ilmunya..
Best tutorial on comfy UI out there.
I'm enjoying this series. You explain everything very well. Great job!
Dude. Thank you. Subbed. Can't explain how awesome this series is. Hope your channel blows up bc it deserves it.
Thank you so much ☺️
This content is insanely valuable! Really hyped to go through all of the videos. Thank you!
Wow! Amazing tutorial and your ability to explain complex concepts to simple understandable themes is a gift. You are gifted with teaching and thank you so much for this tutorial.
Clear instructions that everyone can understand and other channels cannot provide. Thank you.
This is the most systematic tutorial I have ever seen so far! Great work!
This course is amazing, i have played around with Comfyui but never got a full understanding, no wi feel that I do.
Thank you. 🙏🏼🙏🏼
Glad is helpful ☺️
this is the best explanation for the nodes i've ever heard!! Big thanks for this series!
Thank you so much for turning it easy to understand. "This monster is not so scary" when you are teaching. 😉
Glad it helps ☺️
Thank You...! Your workflow shows the process clearly,
New to Gomfy and I like to understand, what I'm doing...
This video helped, this series is so promising...looking forrward to more...!
You are by far my fav SD teacher! Coming from A111 and forge i'v been dreading the comfy UI plunge! But this series is already having me pulling spaghettis! Thank you for this!
Excellent tutorial! The cake baking analogy was spot on! Well done!
Great tutorial. Most talk about nodes only without explaining what they do and why they are there. To really go from A1111 and others, understanding clip, vae, checkpoints, and how they connect is necessary.
I learn the best from analogies, and the cake analogy was spot on! Thank you!
Glad it was helpful :)
love it... easy and clean to understand
your tutorial is amazing. period.
OMG Man, I can not express how amazing your tutorial is! I am a motion designer for 20 years now, but I was scared to use AI for a while. Now, watching your tutorial makes me LOVE ComfyUI already, even I only started it one day ago. :D So, THANK YOU!
Thanks ☺️
excellent, enfin un youtubeur qui explique clairement les choses. vraiment bravo pour ce travail. continuez, je vous suis !
OMG - can't believe how simple that explanation was. WOW 😃
i am a grandpa that recently became interested in AI art. tbh i was not even going to attempt to use comfyUI but because of this tutorial im all in. all the sudden it does not seem so overwhelming thank you.
The cake example makes perfect sense. Thanks for the video!
amazing simple and thorough explanation, wonderful tutorial, i recommend everybody to watch it if you want to learn fast about ComfyUI and Diffusers internals top view, loved it!
Thank you looking forward to this series!
Another fantastic, easy to follow explanation. Thanks!
You are a great teacher! Thank you for your time and patience!
Thank you so much 🙂
I've been trying to get into Comfy, but was always overwhelmed. This was really explained well for a forge user myself, thank you for this. The the idea of moving the nodes in place to mimic the forge/A1111 UI, makes it so much easier to transition to Comf, Cheers
thanks for your informative videos
Cake analogy was fire
Another great tutorial! Thank you!
Glad you liked it 🙂
very helpful - totally new to AI image generation and ComfyUI - this is exactly what I need 🙂
Comfyui rocks! I've been using for a while. This tutorial is good for revisiting some things I didn't know. Thanks!
This suppose to be a paid course, Thanks for sharing it freely
I just downloaded Comfy UI, here goes my weekend...
Have fun 😁
THANKS! very professional easy to follow, great audio and no damn music! : )
im going to watch everything, thank you sir
glad you like them, enjoy 🙂
Thanks!
Thank you again for continuous support ☺️ it helps a lot
Great job explaining such a complex topic in a very simple way 👏 🤩💥
Thanks for the info man. I did learn a lot, but being new to the system, I was still confused. So just a little feedback and clarity for anyone in the same boat. When you are using an analogy, make sure you bring it back to what it is the ai is actually doing and why. I took a little break while you were prepping for dessert and had to look it up. This is what made it click with me……. The whole software is based off of 3 main categories with an input (encoded text prompt/image/etc) and an output (decoded end result). A model (Stable diffusion..etc), a control (posing, style, idadapters, etc) , and a view (stage/latent space). Your analogy does make sense though, it just needed the context to the topic of interest. Understanding the “how” really puts the “why” in perspective. The idea that these models are literally generating things out of noise is just insane to me. Anyway… keep it up man, anyone spreading the love, should get love….I’m just giving you my 2c from one viewer. Good luck with the channel!
Great video. You've explained everything really well.
Nice work!
Link render mode now works without restarting via the manager.
Very detailed and useful tutorial~nice work
This is simply brilliant. Thank you for helping me out
I like the way you make your videos. Easy to follow. Please can you do a video on setting up LivePortrait in Comfyui.Many Thanks
Maybe on a later episode after i finish with basics
@@pixaroma Thanks I will watch out for that one.
Great video and great explanation! Thank you
Another well explained tutorial. Thank you.
Great job! Great breakdown! These instructions will help me a lot :-) thank you! I love your bakery breakdown… It really helps me it already has helped me to understand more effectively what my image workflow is accomplishing each step of the way, and most importantly I can now learn how to troubleshoot and tweakmy outputs! Very excited about that prospect, because up until now I've just been dropping workflows like a black box hoping that they have the components that I wanted to accomplish… Thank you so much! Dan Hawkins
great tutorial! keep up the good work!
Amazing tutorial, thank you!
Agree with the others. Quality content!
thanks 🙂
As a developer and designer for 20 years I’ve done allot of tutorials in my time but this was the best one I’ve ever watched . Subbed
This tutorial made me hungry. Haha jokes aside big ups thanks for the tutorials.
You are doing a great job!
Great clear info. I am just getting familiar with Comfy
Great video. Keep up the good work
👏Another great one. Thanks!
This is fire 🔥
Thank you so much!
very nice sir
Comfyui.... Ullltrasupergreat video bro!
✨🎇🎉🤗🤗🤗🎉🎇✨
I think, learning by you is pure joy. Clear, good speed, well-built. You definitely have talent to do it. - Okay, my question: the Flux is here and we can use it with Comfy and Forge as well. (Probably the Fooocus version will also be done soon.) What is it better with, Comfy or Forge? What is your thought?
I got similar speed on both, is just a mother of preference just sometimes on lastest versions both interferences can have bugs, like it happened to me with both. Like on one version of forge it took 5-6 minutes to generate and after an hour on an update it took 27 seconds. Same with comfyui on an update it started crashing my comfyui and didn't follow the prompt. So i just have both and use both of them each in its own folder. Forge js easier to use, but on comfyui you can do more and updates are way faster, on forge it takes a few days or weeks to get the new stuff, like I was with Flux comfy got it to work from day one
@@pixaroma Thank you.
(great tutorials:10.10), (thank you:∞), (grateful viewer:10)
another way to think aboutempty latent image node could be that It makes a blank canvas we can decide there the size of our canvas and then the ksampler paints on the given canvas
Nice!
Great guide, as always, thanks. Will there be a new version of your episode: Instant Art Upgrade: 260+ Stable Diffusion Styles for A1111 & Forge? Is there such a NODE? Is👍
not sure it is, but if it is I will do a video for it, I plan to do the things I did in forge also for comfyui if is possible, but I am still learning more advance stuff also, seems that you can do really a lot of stuff with it
Hello i love your explanation for ComfyUI. I have a question : do you will try live portrait. I have many difficulties to install it thanks
Maybe in later episodes, i still have more to talk about in the basics so later when we do more complex it will be easier to make things work and fix problems
@@pixaroma cool if i find a way i will tell you thanks buddy
Great video Decean, as always. Question for you, the speed of Forge seems to have come from Comfy UI (as i've read), does this mean that since Forge works well for me, that ComfyUi will as well? I have 6g VRAM. Keep the awesome content coming!
I just tested today on rtx2060 with 6gb of vram, you can do basic generation, not sure if control net works, the upscale is a little slow, but you use a model that is fast like juggernaut hyper i can do a generation in like 12 seconds because it doesnt need many steps. You can check on my facebook group i posted like a few minutes ago.
@@pixaroma what's your facebook group
Thank you
You're the best
It finally clicked for me why all those nodes are needed. Now to go bake some sweet, sweet AI art. Anyone else thinking of trying out that Juggernaut model after this?
Thank you! You are like extremely good at this, like your whole approach is amazing, hopefully my words aren't going to affect your ego, lol, just kidding... But yeah, you have multiple talents combined in what you bring forward. Thanks again, 🙂
would you recommend Efficient nodes? great Vid by the way
I didn't use it yet because i learn it step by step and first i work with the nodes that comes with comfyui and slowly i install new custom nodes. I saw many people using them so I will probably use them also
@@pixaroma that's reason i asked because i've seen everyone and their brother using them
does comfy ui have the same features about auto1111, with regards to add script text file on th e positive prompt? where you can input many prompts variations on one button?
I didn't see one, but probably there is a custom node somehow that can do that, i will have to do more research on that
Found this also, see if it helps, I am not at the computer to test it github.com/comfyanonymous/ComfyUI/discussions/896 in the comments someone posted a custom node
@@pixaroma thanks
This was the best video i came across for stable diffusion hands down. Do you have a discord server where i can join?
Thanks, yes you can join here discord.gg/gggpkVgBf3
Great video! You should do one on AI voice generation :D
I will do it once they fix some bugs they have :)
How do you determine the values needed for the KSampler from HuggingFace? For example, I downloaded stable-diffusion-3-medium.
If it doesn't say the value there there are probably the same model on civit ai that has the vales somewhere in description or open a ready made workflow that comes with recommend settings. For example like in episode 5, or do a search on Reddit and probably can find someone asking the same question and got an answer
perfect
thanks
What's the function of "clipspace" button of the queue sidebar? Thanks
The "Clipspace" button in ComfyUI's queue sidebar is a tool that allows you to manage temporary data used within the UI, particularly when editing images or masks. Essentially, Clipspace acts as a clipboard for storing and manipulating images or other data. When you perform certain operations, like using the mask editor, the image data is copied to Clipspace, where you can modify it. After editing, the changes are saved back to Clipspace and can then be applied to the original node or used elsewhere in your workflow. check also the comment on this www.reddit.com/r/comfyui/comments/1cql2i1/what_does_the_clipspace_button_do/
Love U❤🎉🎉🎉🎉
Hello, I'm having problems making the workflow in this tutorial works with the following Checkpoint Model: realflux10b_10bTransformerDev.safetensors. The Compact Version, which I'm using, already includes VAE, Clip-L, and T5XXL. Any tips?
is hard to tell since I didnt use that model, but probably needs to be in the right folder, so first you must see which folder comfyui see it, and which node needs to load that type of model. Post some screenshots maybe in the discord server on the comfyui channel discord.com/invite/gggpkVgBf3
💚💚
salut și bravo pt tutoriale și canal în general! Vin la timp ca tocmai ce am început să învăț comfy. Am o dilemă majoră și poate îmi poți da un sfat. Lucrez pe mac de 15 ani iar acum am un macbook m1 max dar e lent pt Ai în general și vreau sa fac un desktop pc. Am tot făcut research care brand de 4090 este mai reliable și nu prea reușesc să ajung la o concluzie deoarece văd ca sunt probleme cu pcb-ul sau connector-ul acestor placi variând de la brand la brand. Mă poți ajuta cu un sfat? Merci mult și baftă!
ps: scuza de romenglish, sunt plecat de ceva vreme din țară și îmi găsesc mai greu cuvintele
Nustiu care is mai bune, eu am placa asta video de un an si pana acum e ok. Eu l-am facut la comanda, si i-am zis eu la ala de probleme cu conectori, a zis sa nu imi fac griji ca imi pune o carcasa mare si are loc sa bage mufa cum trebuie, numa daca nu e bagata bine face scurt, sau ceva de genu, acum ce sa zic, la noroc :) GPU GIGABYTE AORUS GeForce RTX 4090 MASTER 24GB GDDR6X 384-bit
tks mate, I’ll flip a coin 😂
how to download vae in comfy ui
Depends for what model, if you go to manager and custom nodes manager you can find a few there. The one i used.juggenaut it had integrated the vae
I think ComfyUI user are subconsciously regards it as a game... Because thats how I treat it... I have have no image objective yet I love tweaking and experimenting different combinations... Master Olivio Sarikas said imagine painting on a canvas analogy... or you baking a cake.... for me its Steve Factory Manager 😁🤓... Im gonna be yapping about minecraft again... ComfyUI has those Freedom and Creativity essence which is the reason minecraft are so damn popular back then...
I struggled with creating groups in comfy then I found that the best way to create a group is to select all nodes for that group then right click on an empty area then select create group for selected nodes
Great tip, thanks :)
Thanks!