Experiments with Zoned Noise in ComfyUi

[@AnnisNaeemOfficial]

Wow. really great, Rob. Thanks for sharing. this is very good. I'm currently working on some paintings and was frustrated that it wasn't taking my shape and perspective cues without being too literal in what i was drawing.

[@robadams2451]

This something I have struggled with, you lose the mad magic of AI if you use too many constraints, but on the other hand I want the visual ideas to be mine.

[@AnnisNaeemOfficial]

@@robadams2451 For me, the mad magic of AI is wearing off. Like stockphotos, everyone is generating the coolest stuff that it all looks similar again because they're all using the same technique. This right here feels like the perfect blend of guidance instead of being stuck with only word prompting or image to image randomness.

[@robadams2451]

@@AnnisNaeemOfficial It's funny but those first AI images which amazed us are now ordinary. I remember typing in all sorts of nonsense to see what came out. Then I wanted to mimic painting but it seemed pointless. Rather like how early photographers had the ambition to mimic paintings. Current AI enthusiasts want to mimic photos. But AI has to be its own medium and we don't quite know what that might be yet. I think perhaps we will have images which change as we talk to them and reflect our moods and fancies.

[@AnnisNaeemOfficial]

@@robadams2451 i agree. I think photographers panicked when everyone could afford cameras. And iPhones and such made everyone a photographer. But at the end of the day, cinematographers and pro photographers still crush it because they go beyond the “click”. In our case, “beyond the prompt”. Hey, that’s a cool name for a channel.

[@moviecartoonworld4459]

am always grateful for your lectures and am learning well. thank you.

[@AB-wf8ek]

I just learned a new term recently, stochastic resonance. Although it doesn't relate directly, I think this idea of recognizing patterns within noise is fundamental to how we process information.

[@comfyui]

Hey, thanks! Another channel for me to learn about ComfyUI.

[@hexdump8590]

First of All I want to thank you for these videos. They are gold. On the other hand I would like to ask if it is any possibility instead of obtain a final image, get different layers to compose later. I mean, for example on layer for the background and then another for the girl character (With transaparency). Is it possible? Again, Thanks!

[@robadams2451]

Yes, I often work in that way, you can combine the bits in Comfy but I usually composite together in Photoshop.

[@hexdump8590]

@@robadams2451 oh great! Thanks again

[@8561]

Rob with another gem of a video! I have a workflow i'm building out and trying to achieve something similar to RunwayML's infinite image tool... is there anywhere I can contact you to ask some questions?

[@robadams2451]

Thanks! Not as yet, I am plotting a Discord server, but I need to learn the ins and outs first! I try to be responsive to these videos then others can benefit if I have a solution.

[@babbagebrassworks4278]

Looks pretty useful. Been running SD XL on my PI5. Tried to install ComfyAI but it thinks it need GPU card etc.

[@TestMyHomeChannel]

A great video again. Would it be possible to share all Diff_xxxxxx.png, noise, depth and other axillary files so that we can follow the workflows. It is greatly appreciated. The same for other workflows that are posted. Best wishes

[@robadams2451]

Thanks! There are no depth maps in this workflow. The other files for the loaders are noise images trawled from Google. If you make sure they have a consistent size (in this case 1024 sq) then they will work fine. Try all sorts of images not only noises, it's fine to see what comes out. I do occasionally post dependent images but only where they are unusual in some way.

[@TestMyHomeChannel]

@@robadams2451 Thank you so much for quick response. When workflow is loaded, the files are missing. Not sure where to download them from google. I tried to get screenshots from your video, but the size becomes random. I was thinking it would be very helpful if they are all posted with the other noise files that you previously posted for your other videos on your google drive. If you get a chance. Again I very much appreciate your outstanding work both as a painter and as an AI enthusiast, expert, and mentor. Thank you very much.