Figured I'd write this after I got a feel for how the Ai art that I generated did and to see how many people would either love it or hate it. By far more people love it since I'm filling in a niche of characters that didn't really have alot of art of .i.e Mama bear and Mother Bear and Little Bear and Brother Bear from their respective tv shows.
But most of the negativity I get (at least they PMed me) is mostly from people who down right hate Ai art and think its gonna take over the entire platform or have the fear that its gonna take away from commissions artists get ect ect., idk if they're coming from a point of ignorance, paranoia, jealousy or whatever else there is but I did the math on Ai art.
Inkbunny currently has a total of 2,385,218 total submissions, of those 2.38m submissions, currently 9306 are ai generated, that is a grand total of 0.4% of the total submissions in about the span of 17 months. 17 months ago being when decent quality ai images were starting to pop up. Less than half a percent is hardly "taking over", even checking the popular section of the home page, I counted 16 submissions that were ai generated, out of 252 total submissions, that's still just 15.7% of the popular page.
And generating ai art isn't just as simple as typing what you want into a prompt line and then magically getting exactly what you want every time. To even get started it requires a pretty beefy PC, RTX 2000 - 4000 series GPUs is pretty much mandatory to get generations that don't take 5-10 minutes per picture or even longer, a GTX 1080ti might be able to pull off decent times but its lacking tensor cores which really speed everything up. Then getting a Ai platform installed and running like Stable Diffusion or A1111, finding a main model you want and LoRAs or make your own if you got loads of free time and want to learn how everything actually works. Then learning the way to write prompts for whatever it is you're wanting and tweaking it and re-generating images over and over until you get something decent. But even that will just get you generic boring images. The more advanced stuff like, controlnet, inpainting, LoRA networking, splicing LoRAs together, openpose, ect. is when you get some pretty good looking generation. Then using photoshop to finish up the image and fix any anomalies for images like how I did on my Little Bear x Uncle Rusty story and illustration
And finally as a former digital and traditional artists, I get it if someone straight up steals your art for their Ai models and/or is selling it, use a program like Glaze to mess up Ai models on your art, don't start whining about how Ai art is bad and should be banned.
Viewed: |
71 times |
Added: |
2 months, 4 weeks ago
02 Mar 2024 02:19 CET
|
|