I don’t like the look of yours.
1 Like
“Salade de gilets jaunes intersidérale” ( the first words that came to my mind in French)
1 Like
the magic of technology
My interpretation of the IP section pasted above was that it was referring to the app itself and the technology surrounding it, not the images it generates.
1 Like
I guess the question is whether “content” refers to the images generated as well as the app. I’d love to get clarification from the company on what the terms are. Some of these images are genuinely f-ing amazing; it would be awesome if we could use them for our own (potentially commercial) purposes.
Shoutout to the Neuromancer quote!!!
1 Like
Yeah, “original content” is a bit ambiguous. But I take that to mean the graphics used in the app, not the graphics it generates.
I’d be interested in seeing how they’d police it if the generated images were copyrighted, especially since the same word used multiple times generates slightly different results. Maybe they have some kind of encoded watermark. In that case, it’d just be a matter of using the image as a base for another image, ie. either painting over it or running it through another AI service that makes it look like a painting lol
1 Like
I have been involved in a commercial copyright claim. If you don’t want to have a very bad time, you will end up spending a lot of money on lawyers.
Best would be to contact the company and tell them you make sick underground techno and ask for licensing terms while emphasizing your diet of ramen.
1 Like
I imagine all files generated are kept on file to be fair, wouldn’t be a massive amount of storage. There’s gotta be a reason they don’t make it into an offline app. And a reason it’s free
If the process is deterministic, they may only need to keep the prompt strong and the version of the algorithm that created the image. But storage of still images is basically free.
It work in french.
1 Like
I see all these interpretations and I understand why AI decision making on IVRs, dialog bots and navigation is so terrible.
2 Likes
I do wonder how it works though. It takes some clear influence from an external source it’s not just being randomly generated based on other peoples input and machine learning. You can write any celebrity’s name in and get something that looks vaguely like them. I get it could be trained on massive data sets but it seems too non specific. like it’s been trained on everything from mundane objects to spiderman to boris Johnson. I was expecting some human spider hybrid but got accurate Spider-Man colours on the side of a sort of building
2 Likes
There’s a good point raised here:
If wombo.art is in fact extracting artwork from google or other random image searches on the internet… which I expect they are and this thread seems to confirm. Then I think they would have a “very” hard time defending a copyright claim if you were to use the artwork generated.
1 Like
Interesting thought about machine learning. How does one exactly generate massive datasets to feed to AIs without using an immeasurable number of third party sources? you can hook it up to 10000000 pictures of Spider-Man through the internet but you can hardly provide it with 1000000 pictures of Spider-Man I produced myself or acquired the rights to use
1 Like
Also, how does the legal stuff work, exactly, when it’s clear that the algorithm uses other, previously copywritten images to generate the end product? For example, using some very specific prompt words, I was able to generate an image that plainly obviously used one of my own photographs to make it. EDIT: someone had something to say about exactly that on the Reddit linked above…
1 Like