That's not how image-to-image processing works. A base image gets translated to code so it can understand the layout and colors it's working with, temporarily, for that single generation. If every image submitted affected the training model,
A) Hosts wouldn't need to release big model updates every now and then
B) Someone would have figured out how to poison the training data by now
@exophilt@Leaflit@Isesi_VT If Leaflit's pngmodel generation weren't a big deal, nobody would respond to it to begin with. The story would have never gained traction or attention.
@LuckyPochii I sympathize with that urge, however consider the consequences. Crippling debt, and being under the thumb of a credit card company for potentially years.
@Alex_G_Art@SonicAir1375@Grxit That's not what I'm getting at. If the VTuber does not own the copyright, they generally have no legal standing to file a copyright dispute, issue a DMCA takedown, or sue an infringer. That sort of security would only come with a full buyout/transfer.
@Alex_G_Art@SonicAir1375@Grxit You wouldn't use copyright for that. You would be using trademark law and right of publicity to protect your name/brand/likeness.
@Knightw0lf_@BlondiniVT The practice of using tax incentives and subsidies to lure industrial manufacturing was born in 1936. These are temporary or scaled back over time and generally speaking, the locale benefits more from the jobs brought than taxes lost in the long run.
@SirCorny1234@Leaflit Neuro is not a ground-up AI, the hell are you on about? Her core conversational ability relies on Generative AI. Heavily modified and fine-tuned, sure. But still apples to apples.
@Zekriox@WitnessVoid47 You cannot copyright a purely AI generated content. You CAN however protect your brand with trademark law and right of publicity to stop others from using your likeness.