Hacker News Reader

Comment by anonymous344 | Hacker News Reader

awesome project, might not last long if server costs arise.

but where are the images from ? scraped or self generated? how can i make sure that they are not copyrighted?

machine generated works cannot be copyrighted, but its a legal gray area as to whether creating the images violates copyright (in my opinion it is not fair use, as it displaces the market value of the work it is a derivative of)

But the use of each work is very minuscule in sum and is generally, if not always, an original new work. How is a Mona Lisa in the style of Van Gho a violation of the copyright of Mona Lisa or Van Gho? It is a new and unique work. Makes it a legitimate use. In fact, in certain cases the entire work can be used and is considered fair use.

Americans do pronounce his name "Venn Gho", but it is spelled Van Gogh!

Interestingly enough, there are many pronunciations of Van Gogh's name [1] (we French people say "Gog" rhyming with "dog"), but the rarest one is probably the one from his own native Brabant dialect.

[1]: https://en.wikipedia.org/wiki/Vincent_van_Gogh#cite_note-5

I was more so talking about Van Gho the music artist[0], mixed with a bit of Mona Lisa the band[1]. But to each their own

[0] https://www.facebook.com/VanGhoMusic/ [1] https://en.wikipedia.org/wiki/Mona_Lisa_(band)

LOL... I learned something today!

I agree that images produced are not derivatives of any particular copyrighted work, instead they are derivative of all copyrighted work. I don't think OpenAI et al should get a free pass for finding a way to rip everyone off at once, but it's certainly a new frontier in law, hence the gray area.

Well, in copyright law there is the De Minimis Concept, which kinda sorta allows, as you put it, OpenAI to "rip everyone off", despite everyone holding what they originally had.

Completely obvious they are generated. Just zoom in on anything that has text and it’s easy to notice.

Reply on news.ycombinator.com