Is Fan Art Illegal? Legality and Copyright of Selling Your Art at Conventions

[@KetchumAllCollectibles]

I knew all the answers but interesting to hear the mumbo jumbo behind it. Keep em coming

[@KetchumAllCollectibles]

Glad weird Al satire exclusion is there 🚀

[@EricDrinksWater]

Part of me was hoping to see Dan's "non-descript orange lizard" in the charmander section haha

[@moffdawg]

haven't watched yet, but I'm appreciating you bringing your professional background to the hobby. It's inspired me to do my own videos!

[@milehighpokemon]

That's awesome, good luck with those. I'll take a look. Appreciate your support :)

[@Yquan151]

Thanks for the presentation. I guess i am farily safe to continue making my wooden spoons adorable with pokemon themes.

[@milehighpokemon]

That's for watching! Just don't tell anyone you got the go-ahead from me lmao

[@HorseyWorsey]

AI doesn't "doesn't store images" and "observes art" (what does that even mean lol) . That is just factually wrong. I make AI models. AI needs to be trained on downloaded digital material in order get it to a usable state in many cases, especially fanart. There is software that does this. Unless the base model is already trained on the copyrighted art (which anyone with a brain knows that in order to get it that good, they mixed a ton of private IP into it) it'll produce poopoo. Also every company/open source model is different. So AI trainers do what I mentioned: they go scrape/download the copyrighted material, train a model or a specific network on the images, then use the networks, or re-release the network either for free or a premium, often even literally referring to the specified style of output for the exact IP/copyright holder in their distribution. Its a neural net generator, it doesn't surf the web looking for pictures unless its a specific hypothetical AI observation-generation system... which you didn't specify. So no and yes, with how you currently phrase it, the AI itself is only generating similar images based on training data (no copyright), but guess where it got that original "training data" aka the copyrighted images from (yes copyright)? Its like saying the arsonist didn't create the fire, it was the match that did it. You should brush up on your data laundering law before you start using words like irony.

[@arbitScaleModels]

So how does a public company like Shapeways list copyrighted material?

[@EDJ_3D]

Hi can you provide comment on using Hueforge to 3d images like a lithofane but with color? Would that technically be considered transformative enough? Especially since the market for 3d printed art by the owners doesn't really exist yet?

[@milehighpokemon]

Hi, thanks for watching. But I can't give legal advice and don't know enough to comment on your specific technique (or what the end product would look like). I think the analysis I provide for fan art in other mediums would probably apply equally to your hypothetical, though. I'd recommend securing your own counsel if you're looking for something more concrete.

[@adventurousash]

AI would be unable to produce the images without the initial training data to create the AI. The issue is that training data is (most of the time) not licensed and is being copied directly by AI to create it's "algorithm." It's unfair to say the original images aren't part of the AI, because they are (just translated into a very complex mathmatical model). I think it'll be interesting to see how courts side on IP and AI, because public opinion is so negative towards AI, mostly due to companies abusing it and trying to get rid of jobs.

[@bob450v4]

This is wrong the images are not stored in the model. This can be easily proven by the simple fact the models are usually 1-10 gigs while the training database of images after compression is terabytes if not petabytes. What is in the model is abstract concepts like human,dog,cat and concepts of art style etc. some very popular images like starry night may be an able to be somewhat recreated by a diffusion model if prompted explicitly to do so. But this is simply because starry night for example has appeared In the data set so many times with the tag starry night that the model learns to embed a lot of the aspects of the painting. But even in this most extreme example models cannot accurately reproduce starry night.

[@bob450v4]

To add, the phenomenon of machine learning models producing results that are too similar to its training data is called overfitting and it is not something you want in a model and is avoided as it hurts the model’s understanding. As it learns to “remember”. With all this being said I do sympathize with artists and hope that people still value human art as I do no matter what even if a ai could create the same quality as a human there would still be times in witch I choose to commission or buy from a human as I value human creation.

[@EricDrinksWater]

Boyyyyyy. Lookin' forward to this one.

[@milehighpokemon]

Appreciate you taking the time to watch!

[@IneptCardCollector]

Sonichu medallion good then lol

[@KetchumAllCollectibles]

Whoa whoa take it easy with the generic orange lizards 😆

[@milehighpokemon]

I'm telling Nintendo on you.

[@KetchumAllCollectibles]

@@milehighpokemon ☠️

[@IneptCardCollector]

🚓🚓🚓🚓🚓

[@luxuriousmindset1906]

You said you have your own opinion on it i hope is not you advocating to loosen the grip so you can benefit off of someone else's ip like make fan art and sell it without input of actual creator because basically this the common agenda being pushed

[@milehighpokemon]

My view is that you shouldn't break the law. But I think the law should be changed to loosen up copyright and expand fair use. I don't have an exact proposal for you. However, stronger copyright laws add only marginal incentives for creativity and stop a ton of new art from being created in the first place. For example, I believe that extending the copyright term from 70 years to 90 was harmful, and it almost exclusively helped large corporations.

[@keyynew1]

appreciate work u done for rattle

[@milehighpokemon]

Appreciate you checking the channel out!

[@Dh66548]

🤡🤡🤡

[@Dh66548]

🤡🤡🤡