Welcome to Inkbunny...
Allowed ratings
To view member-only content, create an account. ( Hide )
FlameSnow

Update of a sort

Hi friends and non-friends of IB.Been a while since I posted a journal. Or really any art too, you may have noticed. I'm in a art slump right now, if that much isn't obvious. Unfortunate, no?

My artistic drive right now is practically dead right now because of a couple reasons. It's so easy to get discouraged when there are other artists that are much better, much younger than I am. The rise to prominence of AI art has also really hurt my drive; it sort of fills me with a "why bother" mentality.

I've never done art for anyone other than myself, I don't take commissions or requests. I put a bit of my soul into every piece of art I do though, so even as a hobby I still take it seriously enough that I am bothered by these things.

I actually am on a new PC technically, but I got my art stuff installed and ready to go, just haven't used it yet. Hopefully I can get out of this slump sometime though.

I hadn't intended for this to turn into some kind of rant. I hope everyone's doing well in this new year and we're all surviving.
Viewed: 31 times
Added: 3 months, 1 week ago
 
BlitzkriegOmega
3 months, 1 week ago
when it comes to the AI shit, don't let it drag you down. that insincere garbage isn't going to stick around. The only people seriously using it are Elon sycophants, grifters, and Corporations you'd never want to work for anyways.

I know muse is fickle, but I do hope you can manage to find yours again and create something that is imperfect, but beloved by your friends anyways.
FlameSnow
3 months ago
Thanks, I hope I can pick myself up too. I appreciate you being a friend.
TheAllmighty0ne
3 months, 1 week ago
I hope your drive to draw returns strong and full of inspiration, I love your drawings and even have some saved myself!
cheers and keep creating
FlameSnow
3 months ago
Thank you, I do deeply appreciate my friends cheering for me. I'm happy you still have faith in me and enjoy my works, I'm sure I'll be able to get back to it hopefully soon!
There's more pregnant moogles to be doodled.
rainChu
3 months, 1 week ago
Neural networks are really misleading in their scope. They are able to produce things so impressively "human" that it looks like you can never compete. But the reality is actually different. I use AI. You know that; I'm qualified to say how incompetent it is versus a human. Think of a model as a deck of cards, not a brain. If you abstract all its dimensions in your head to just 1D, you have a deck of cards. This works for 2D also: take a window on your computer that has icons, make it narrow, they become one line. Word wrap a paragraph. The idea is the same. For this explanation, it's mathematically unimportant that it has more than 1D. The AI is like a deck of cards, in a big stack, and drawing the top card leads to utter garbage. 90% of the cards has garbage on them. 10% are interesting. Less than 1% are useful. This is to say, it doesn't create images that are new. Calling yourself an AI artist is wrong,  because what comes out isn't a new work. It's finding an existing image inside what's basically a glorified zip file.  Entering a prompt, or a complicated Comfy-UI workflow, or python scripting, doesn't matter: this is how you point to which card to extract/unpack. Now with that explained, you may wonder how so many images can fit in the model, especially because over 99% of the images are completely unusable. And you'd be right to wonder. Unlike a zip file, which compresses specific information, a neural net stores understanding of information; it stores what arms, legs, faces, etc. are, and thus images are "generated" by reversing the storage process. If you had to remember every cell in a person to clone them, you'd never be able. Yet imagining a person is possible because you break the idea down into concepts. So! Here's the final and most wonderful thing to come from this explanation: the model can't create your art. The reason it looks "samey" is because of that very abstraction that allows it to work. Regardless of style, you'll find that the poses, facial expressions, etc. are incredibly similar across all images. AI users would say they can just train a LoRA. Okay, I guess they could spend days training a new model. It's still a model and still suffers from the above explanation. Everything it makes will be like everything else that model makes. To reiterate and clarify: The images you create can be anything you want. The images an AI "artist" extract must already exist in the model, waiting to be discovered. If the AI has no image that they are trying to "generate" it will simply never, ever appear.

So: Why can't an AI draw like you? Because it's "not in the cards."
New Comment:
Move reply box to top
Log in or create an account to comment.