The workflow is very complete and I already see how I can use it and also improve. The suggestion is to have a choice of up to 4 different types of paints that could replace the coatings, be it floor, walls, and roofs. I still don't know how I would do that, but I will study to help improve
Hey, thank you! I would probably approach this with a segmented mask that either is prepared before from the 3d environment or included in the workflow with some sam2/florence2 functionality. I'd love to see what people would make out of this, please let me know if you succeed in any way with your desired results ♥️
hey, you can drop it where the depth map is and set the depth-lora to canny-lora in mode 0 for example. It's ment to experiment with, depth worked better for my example :)
nice clip! i did see somewhere this VAE anything and everything node, which automatically connects all the vae points and does not give you spagettilines
hey, thank you! i know about these nodes (github.com/chrisgoringe/cg-use-everywhere) but while it further i really like to see what does what, and with the new gui i can just hide them all at once :)
@@dungnguyenba3499 hey, since ComfyUI 0.3.0 i think (when the new UI was introduced) in the lower right corner there is an eye-symbol, click that and all the connections are hidden.
This is groundbreaking, sir. I have one question, what is the maximum size of the input image and how to use the output image to upscale it without losing quality? Thank you again.
Hi, i have tested it with both of the Input-Examples from the zip file: 1440x1920 and 3072x2048 pixels. But it might work with larger inputs, depending on your hardware. In this workflow no upscaling is used so far and you might find better settings for outpainting to kepp the quality. However, you can of course add an upscaler with some lower denoise values behind the image process, this may be also implentend in a future release.
A lot of thanks for share this amazing workflow!
The workflow is very complete and I already see how I can use it and also improve.
The suggestion is to have a choice of up to 4 different types of paints that could replace the coatings, be it floor, walls, and roofs.
I still don't know how I would do that, but I will study to help improve
Hey, thank you! I would probably approach this with a segmented mask that either is prepared before from the 3d environment or included in the workflow with some sam2/florence2 functionality. I'd love to see what people would make out of this, please let me know if you succeed in any way with your desired results ♥️
Hi Paul, where should we use the Curvature pass for Canny?
hey, you can drop it where the depth map is and set the depth-lora to canny-lora in mode 0 for example. It's ment to experiment with, depth worked better for my example :)
nice clip! i did see somewhere this VAE anything and everything node, which automatically connects all the vae points and does not give you spagettilines
hey, thank you! i know about these nodes (github.com/chrisgoringe/cg-use-everywhere) but while it further i really like to see what does what, and with the new gui i can just hide them all at once :)
Can I substitute local models with Flux api?
Afaik you can use bfl api in comfy yes, have a look here github.com/black-forest-labs/bfl-comfy-nodes
looked it up, unfortunately you can't use the api for that kind of control
great work ! how did you hide all the strings between nodes at once ?
@@dungnguyenba3499 hey, since ComfyUI 0.3.0 i think (when the new UI was introduced) in the lower right corner there is an eye-symbol, click that and all the connections are hidden.
This is groundbreaking, sir. I have one question, what is the maximum size of the input image and how to use the output image to upscale it without losing quality? Thank you again.
Hi, i have tested it with both of the Input-Examples from the zip file: 1440x1920 and 3072x2048 pixels. But it might work with larger inputs, depending on your hardware. In this workflow no upscaling is used so far and you might find better settings for outpainting to kepp the quality. However, you can of course add an upscaler with some lower denoise values behind the image process, this may be also implentend in a future release.
@ thank you so much, looking forward to this
love the workflow and the song haha. would you mind letting me know where it´s from?
Thanks, i made that song (with the help of ai, as i would not be able without) :)
@Paul_Hansen that's awesome 😅👍🏼
on.soundcloud.com/HmcMmzydDqi5huqz5
@Paul_Hansen soooo gooood! hab's auf dauerschleife :D
@@yatuex ph01.bandcamp.com/ 😇
OH MY GOD WHAT THE HELL IS IT??
how can learn this complex things
i just have comfyUI
@@Malinarch hey, thanks man.. i just try things out and what works and is of use stays :)
How to download this workflow
Download link in the description ☺
That's a great workflow, thank you for sharing.
You're very welcome!