• Sign up or login, and you'll have full access to opportunities of forum.

Discussion about A.I.

Go to CruxDreams.com
Nyghtfall3D said:
@fallenmystic

YT just recommended this to me. It went up 30 minutes ago and is from one of SD's developers. I thought you might be especially interested:
Click to expand...
I only skimmed through it since I’m still in my bed (dang mosquitoes wake me up multiple times at nights :confused:) but I think I get the idea. I feel intrigued by the possibility of having a simpler UI for ComfyUI without sacrificing the ability to build my own workflow.

The other videos you posted from the same source were also very informative for me. This will be my first AI channel I subscribe on YT.

Thanks much for the recommendation! :)
 
Last edited:
I haven't made any render for a while since I've been busy working on a prototype text game driven by LLM.

Considering the current status of the 7B models I used for the experiment, I don't think it'll ever be fully playable. But I've been enjoying it much already, so I'll keep working on it and hope that either a better model or technology will allow me a breakthrough to finish the project someday.

Screenshot_20231113_130011.jpeg Screenshot_20231113_123934.jpeg Screenshot_20231113_123847.jpeg

(Both the story and the choices are generated by LLM)
 
Last edited:
It's invariably invading many threads.
So as to avoid offering moderators, contributors, ARTISTS, and administrators, where can we discuss?
Thanks!
 
It's...a sticky subject. A lot of the controversy is due to the way most generating models are trained, which is basically stealing a lot of artists' IP. You can make an effort to avoid that by training your own, but that takes a LOT more resources and work than throwing a prompt or two at DALL-E.
 
For ethical and technical discussions of AI, I suggest using this thread.

MSTGMA said:
It's...a sticky subject. A lot of the controversy is due to the way most generating models are trained, which is basically stealing a lot of artists' IP. You can make an effort to avoid that by training your own, but that takes a LOT more resources and work than throwing a prompt or two at DALL-E.
Click to expand...
It's controversial whether to call it an act of "stealing", to say the least. But again, I believe things like that can be better talked over in the thread I mentioned above because a lot of points for and against AI (including those about this particular point) have already been made there.
 
Last edited:
I just tested Generative AI for Krita, which has drawn much attention lately, and it was quite amazing. Although live generation (with LCM) was too slow to run locally on my RTX3080 machine, it was sufficient to create sketches for future renders quickly:

Krita.jpg sketch.jpg

For those who haven't heard of LCM(Latent Consistency Model), it's a method to generate decent images using Stable Diffusion with just a few samples. The implication of it is that now it's become possible to create AI renders in near real-time:


I'll probably develop one of those sketches into my next work this weekend.
 
fallenmystic said:
I just tested Generative AI for Krita, which has drawn much attention lately, and it was quite amazing. Although live generation (with LCM) was too slow to run locally on my RTX3080 machine, it was sufficient to create sketches for future renders quickly:

View attachment 1396496 View attachment 1396497

For those who haven't heard of LCM(Latent Consistency Model), it's a method to generate decent images using Stable Diffusion with just a few samples. The implication of it is that now it's become possible to create AI renders in near real-time:


I'll probably develop one of those sketches into my next work this weekend.
Click to expand...
can you create war themes from warhammer in this program?? or my beloved throne of glass book series??
 
Aluafaen et Rimmon Ra said:
can you create war themes from warhammer in this program?? or my beloved throne of glass book series??
Click to expand...
It can be easily done with Stable Diffusion, and the client program is irrelevant as long as it supports Lora.

As for popular themes like Warhammer, you can usually find relevant Loras on Civitai. If you can't find one, you can create one from 20 or so images with some effort.
 
I initially intended to use the sketch shown in my last post only as a base image for ComfyUI, but ended up finishing the whole render using only the Krita plugin because I really liked the workflow.

I didn’t realise how much I missed layers when working with AI tools before. The Krita plugin seamlessly combines the best of the both worlds, such useful tools as layers from traditional image editors and the creative freedom from Stable Diffusion.

And it’s also based on ComfyUI so I can always use it for more advanced workflows.

Krita and the plugin are also available on the Windows platform so I highly recommend them, especially if you’re familiar with working with traditional image manipulation tools like Photoshop.
 
I'm a bit surprised that nobody has said anything about or even reacted to my previous post.

Of course, I don't crave attention that much - I don't even care if my works are shared without credits.

I meant that the Krita plugin I mentioned above could be a very accessible yet quite powerful tool for those familiar with traditional image manipulation tools like Photoshop. It can be a game-changer, especially to those who make "photo manips".

The only hurdle could be the system requirements (provided you don't want to spend money to rent a cloud instance) and the efforts required for the initial setup.

It's not just for making photo manips - I made my last render using this plugin only. But if you've been making photo manips using traditional tools, I strongly suggest trying this combination out, since it will definitely make your life much easier and your output indistinguishable from actual photographs.

screenshot-1.png

(Image linked from an external host)
 
fallenmystic said:
I'm a bit surprised that nobody has said anything about or even reacted to my previous post.
Click to expand...
You may have simply gone so far beyond what anyone here would deem a comfortable technical level of skill to achieve that no one quite knows what to comment on or ask.

As for lack of reactions, perhaps their eyes get too glazed over to bother mousing over to the Like button? ;)
 
Nyghtfall3D said:
You may have simply gone so far beyond what anyone here would deem a comfortable technical level of skill to achieve that no one quite knows what to comment on or ask.

As for lack of reactions, perhaps their eyes get too glazed over to bother mousing over to the Like button? ;)
Click to expand...
Yeah, maybe you're right. But it's probably because I've been writing with too much technical jargon or my English proficiency failed me to present the information in a readable form.

At least, in this case, it's just a matter of downloading and installing Krita with the plugin and start using it like any other image editor, so I had to wonder.

Anyway, I don't really care if my posts get likes or not. But I definitely want to see more people sharing their fantasies in better quality artworks, which a tool like this can help. :)
 
fallenmystic said:
I'm a bit surprised that nobody has said anything about or even reacted to my previous post.

Of course, I don't crave attention that much - I don't even care if my works are shared without credits.

I meant that the Krita plugin I mentioned above could be a very accessible yet quite powerful tool for those familiar with traditional image manipulation tools like Photoshop. It can be a game-changer, especially to those who make "photo manips".

The only hurdle could be the system requirements (provided you don't want to spend money to rent a cloud instance) and the efforts required for the initial setup.
Click to expand...
anyway I have been following many of your reports, to get a better idea of how working creatively with generative AI looks in practice, beyond just typing prompts into a web Interface.

I'm running the krita plus SD Plugin installation right now, hoping that it works automagically on Windows, and that my computer will fulfill the absolute minimum hardware requirement to get a result
 
malins said:
anyway I have been following many of your reports, to get a better idea of how working creatively with generative AI looks in practice, beyond just typing prompts into a web Interface.

I'm running the krita plus SD Plugin installation right now, hoping that it works automagically on Windows, and that my computer will fulfill the absolute minimum hardware requirement to get a result
Click to expand...
Great! Can’t wait to see what you will create with it. If you have any trouble, feel free to let me know. :)
 
You must log in or register to reply here.
Back
Top Bottom