Reply to thread

Message

<blockquote data-quote="Hacker News" data-source="post: 76531" data-attributes="member: 365">There are websites that offer free images but either require attribution of the source of the image or else require paying a premium membership or subscription.Presumably these companies are scraping public websites to check if their images are being used without attribution.If someone was to take the image and run it through stable diffusion to generate a new image but using that image as a source, should this also require attribution if it was just used as a starting point?I'm sincerely curious on peoples thoughts from both an ethical AND legal perspective (with all the usual disclaimers)For example, one perspective is that the generated image may not resemble the original image but in a sense was used to get to that point, similar to how an artist may see a copyrighted image and decide on a creative spin on that image.A further perspective is that stablediffusion may have been trained on copyrighted images in the first place even though it may not exactly reproduce an image in it's training corpus.<hr />Comments URL: <a href="https://news.ycombinator.com/item?id=33866466" target="_blank">https://news.ycombinator.com/item?id=33866466</a>Points: 11# Comments: 5<a href="https://news.ycombinator.com/item?id=33866466" target="_blank">Continue reading...</a></blockquote>

[QUOTE="Hacker News, post: 76531, member: 365"] There are websites that offer free images but either require attribution of the source of the image or else require paying a premium membership or subscription. Presumably these companies are scraping public websites to check if their images are being used without attribution. If someone was to take the image and run it through stable diffusion to generate a new image but using that image as a source, should this also require attribution if it was just used as a starting point? I'm sincerely curious on peoples thoughts from both an ethical AND legal perspective (with all the usual disclaimers) For example, one perspective is that the generated image may not resemble the original image but in a sense was used to get to that point, similar to how an artist may see a copyrighted image and decide on a creative spin on that image. A further perspective is that stablediffusion may have been trained on copyrighted images in the first place even though it may not exactly reproduce an image in it's training corpus. [HR][/HR] Comments URL: [URL]https://news.ycombinator.com/item?id=33866466[/URL] Points: 11 # Comments: 5 [url="https://news.ycombinator.com/item?id=33866466"]Continue reading...[/url] [/QUOTE]

Verification