AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |
Back to Blog
Plugin for affinity photo3/30/2024 But I do find it funny that after I tried to clarify things for you, you decided to accuse me of being a bot and then try to say I'm the one that needs to calm down.īut no worries, like I said I don't really care what you think about me, which is why I don't really need to "calm down." I'm not worked up. Honestly I didn't expect you to jump to its defense at all, much less so forcefully. I was looking for any information about Stable Diffusion plugins when I ran across yours which mistakenly claimed that Stable Diffusion was in violation of EU laws and felt I should correct it for anyone coming by later who didn't know that. Except Google actually distributes those images, unlike Stable Diffusion. If your complaint is that Stable Diffusion is stealing peoples data to train on so that they can copy those artists and replicate their work, then you don't understand what a model and what an algorithm are, and you're getting mad at the wrong target in your confusion.Īnd if your only complaint is that you BET that the model they made internally for testing, that isn't released, must have used data from the internet without people's permission, then so does Google for their search engine. Or alternatively you should also be mad at Limewire for letting people share music, bittorrent for letting people share movies, google for letting people copy-paste text, and VCR's for letting people record movies off television. If your complaint is that Stable Diffusion is allowing people to use data scraped from search engines and publicly available art sites to recreate work, then complain about the people doing the scraping. Its like getting mad at Russia that nukes exist, not their fault. The genie is out of the bottle and won't go back in, and they're not even the ones that opened that bottle. If your complaint is that Stable Diffusion exists then I don't know what to tell you. Much like an artist already had to have the talent and skills, but likely looked at a picture to make sure they were getting details right. Testing didn't build the algorithm, the building had to come first. To claim that the ALGORITHM should be considered illegal because they used images online when TESTING it is roughly like saying that any art that an artist puts out after looking at a photo or another painting that he hasn't personally paid for should be illegal. But regardless, what they released had zero data in it, because what they released wasn't the model. And given that they were trying to train a model, I would bet they'd go with the pre-tagged content over having to download and tag a bunch themselves. But as a software developer myself, its just as likely that they used any number of sites with libraries of licensed AND TAGGED placeholder images. Is it possible that model was trained using data "stolen" from publicly available images? Possibly. Is it possible that they used images on the internet to train their internal testing model on? 100%. You don't seem to understand what Stable Diffusion actually is. So you are saying that the developers of stable diffusion developed their algorithm without testing it with stolen data?
0 Comments
Read More
Leave a Reply. |