ComfyUI Tutorial Series: Ep20 - Sketch to Image Workflow with SDXL or Flux!

[@pixaroma]

Free workflows are available on the Pixaroma Discord server in the pixaroma-workflows channel discord.gg/gggpkVgBf3 You can now support the channel and unlock exclusive perks by becoming a member: pixaroma kzbin.info/door/mMbwA-s3GZDKVzGZ-kPwaQjoin Check my other channels: www.youtube.com/@altflux www.youtube.com/@AI2Play

[@makadi86]

Complex and challenging workflows can become easy and accessible when broken down and explained clearly by the right person. This is the only channel where you’ll reach such levels of depth in ComfyUI without feeling overwhelmed. Thanks to the skilled presenter for this remarkable series.

[@pixaroma]

thank you so much 🙂

[@feraskiki655]

You are the GOAT hands down. You have such high quality it is unbelievable. It is one thing to show someone that this exists, but you are teaching the process with such incredible clarity. You are a gem. I am indebted. (this is my first ever super thanks even if it's little, but cannot go without it)

[@pixaroma]

thank you so much, glad it helped 🙂

[@ashleymoon9379]

This so perfectly demonstrates how it takes an artist to make truly amazing AI generated art. Imagine the things artists will create as this gets easier to learn. Your videos are amazing!

[@markusps3248]

I really like that even at Episode 20 you keep the ComfyUI setups clean and simple, great video!

[@pixaroma]

thank you 🙂

[@labmike3d]

Your videos are always a delight to watch. With your background in graphic design, you have a keen understanding of how to achieve outstanding results with AI tools like ComfyUI. It's inspiring to see you gradually draw the attention of those who are hesitant to embrace AI tools. Your approach demystifies the process, showing that the techniques share common ground with traditional workflows involving brushes, materials, textures, and lighting. In time, the lines between tools like Blender, ComfyUI, and other open-source software will blur, as AI becomes a staple in all of them. What truly matters are the innovative ideas and having the right tools in skilled hands. Keep up the fantastic work.✌

[@pixaroma]

thank you 🙂

[@SebAnt]

You echo my sentiments perfectly, and you articulated it much better than I can.

[@sergeysaulit]

Thanks to you, a complex tool becomes accessible! A rare talent for conveying complex things in simple words. Don't stop under any circumstances, perhaps this is your calling - to teach people. I am ready to subscribe to your paid resources if there are any! Thank you for your work!

[@pixaroma]

thank you so much. I dont have any paid resources, all is free, people who want to support the channel can leave a super thank you or subscribe to membership kzbin.info/door/mMbwA-s3GZDKVzGZ-kPwaQjoin

[@maxam5898]

What a timing! I was working on a workflow to colour my sketches and boom 💥 this video 😅 we are so lucky to have you man!

[@pixaroma]

Thank you so much ☺️

[@marshallodom1388]

I stumbled on some very basic workflow I tied together in a knot late one night and it was turning black and white line drawings into realistic photos. Now I can see how it's SUPPOSED to done with your excellent as always tutorials

[@59Marcel]

Thank you so much for your inspiring tutorials and all your help on your Pixaroma discord channel .

[@pixaroma]

Thank you so much Marcel for all the support 🙂

[@mrbates100]

Great stuff! Your tutorials and workflows are getting better and better. Good to see that you went full ComfyUI ;) Thanks so much for sharing your knowledge!

[@freke80]

Awesome song and great work flow 👍😊

[@WebmasterDeals]

Great tutorial as usual. Keep up the good work.

[@59Marcel]

I think this is of the best tutorial I've seen on this subject and the song at the end is perfect. Thank you for what you do for us who are just starting out on this fantastic comfy journey.

[@pixaroma]

Thank you ☺️

[@AndreyJulpa]

Epic tutorial, thank you so much!

[@dadadies]

Great video on sketch to AI generation.

[@insane3953]

magic as always!

[@ivo_tm]

Another incredible video! Thanks for sharing!

[@pixaroma]

thanks Ivo 🙂

[@UmarandSaqib]

Love this episode!

[@julayen]

Dude!!!! Its awesome

[@zenseiar]

This has been a spectacular journey, my thanks go to you, following all your content I have been able to learn and handle ComfyUI like never before. By the way, I loved the rap at the end of the video! I would love for you to try Reactor with flux in future episodes, no words to thanks you! keep it on legend!

[@pixaroma]

I have been avoiding anything with face switch because it use insightface, that also cause some instalation problems and it can be use only for personal use not commercial, and some youtubers got copyright strike because using that insightface

[@therceman]

Epic! Thank you for sharing!

[@MarkFrank-u4z]

Thanks!

[@pixaroma]

Thank you so much ☺️

[@kallamamran]

Also if you have rgthree installed in comfyui you can rightclick on the preview image and choose "Queue Selected Output Nodes" to just run those without having to run the whole workflow and cancel the generation!

[@konnstantinc]

good!

[@kalabushkin]

💥 , thahk you!!!

[@bm63]

Nice one! Can we have a few more Forge tutorials please :)

[@pixaroma]

sorry but comfyui offers so much freedom and less errors, and I can not count on them that will not stop doing updates like they did a few months back it just stopped. So probably I will not do more unless it gets more stable.

[@AlexTry-k7k]

Hi Pixaroma, I was wondering if it would be possible for you to make a medium/advanced level guide on how to do LoRA Training in ConfyUI, I've been following your channel the last few days learning exhaustively how it works through your great videos, until now I had been using midjurney for the last 4 months, in it you can upload images to the prompts to use as artistic style references, having some pretty good results. I would like to know if it's possible to use all the images I've been generating, to train a LoRA that fits the style I use. Thanks for the awesome content you generate, I haven't seen a level like this in a long time.

[@pixaroma]

My reply somehow disappeared, you can train online lora in different sites, should work if the images have different style

[@thyagofurtado]

Thanks for the video. The Canva module looks cool, but isn't Photopea the standard module for editing images. You even showed Photoshop at the end, probably Photopea could work just as well

[@pixaroma]

yes that is something simple, no comparation with other photo editors, is more like a paint version :) I think krita also have something.

[@PatrickLavender]

How did you make the song at the end, and how long did it take you to make that song? Was that AI generated also? Pretty catchy tune IMHO

[@pixaroma]

you can use udio or suno,I used udio but now suno has v4 thats quite nice, so yes I created the song with ai, is quite easy just needs a few tries to get it right, like with ai images it needs a few generations and prompt adjusting to get what you want. If you want to see more examples you can check my other channels www.youtube.com/@altflux or www.youtube.com/@AI2Play

[@alex.nolasco]

ah! the workflow is not free, but thank you for the tips

[@pixaroma]

it is free, you just download it from discord

[@SebAnt]

Thanks so much.

[@pixaroma]

Thank you so much ☺️

[@topy706]

What is the node called that lets you switch between workflows/groups??

[@pixaroma]

rgthree custom node has a node called Fast Group Bypasser

[@topy706]

@@pixaroma this is very useful. thank you

[@plouismarie]

The discord link is broken, how to access the demonstrated workflow files alternatively ?

[@plouismarie]

for info : it worked after I removed the invisible space at the real end of the link

[@pixaroma]

Glad it worked out, usually the one from header works ok

[@miriyalapraveenkumar9160]

Bro can you let me know How should i learn to create my own dream workflow? I can’t understand the nodes it is confusing

[@pixaroma]

did you start from episode one? start from there and watch an episode, practice, watch another practice, ask questions on discord. It takes time. If you just want to do now your dreamwork without understanding how things works will be hard, you need to understand how nodes connect and why, and what it need to work. So start simple and then go more complex

[@miriyalapraveenkumar9160]

@@pixaroma okay bro thank you

[@ShockWave2048]

How setup this UI? (left menu, cpu/ram/gpu)

[@pixaroma]

go to manager, custom nodes manager, and search and install this node called Crystools

[@tiffanyw3794]

Will this work well with a 4060 8gb vram 32gb ram

[@pixaroma]

sdxl version will work well, for flux migth be too much to run, so try the sdxl vesion workflow

[@victorestomo729]

How about image to sketch?

[@pixaroma]

you will need a model that is doing usually sketches or a lora that do sketches, just like i showed the one with image to cartoon episode, but instead of cartoon to be sketch, is not as easy since sketch need to do be consistent, but if is not with people you probably can make it work ok.

[@victorestomo729]

@@pixaroma tried it with controlnet and sketch lora, the results are not great. I'm hoping that you make a video once the it technology for that is available.

[@Rising_Volt_Tacklers]

Create a video of comfyui ip adapter face I'd

[@pixaroma]

If I can solve the problem with insight face, and find an easy way that work on any configuration

[@kallamamran]

Controlnet union is an ass to use. It eats VRAM!!! We need controlnets that is not unified!!!

[@devnull_]

Well there are some, are they any better or a lot worse, I haven't had time to test. XLabs has separate models, but only for canny, hed and depth.