Show HN: Create a Deep Fake Video from Any Image
myvoiceyourface.com
myvoiceyourface.com
https://github.com/AliaksandrSiarohin/first-order-model
The license for the code specifically states "Non-Commercial".
Is your code something else? Also seems like you've double posted it.
Edit: Your other posts also have a similar look - are you testing the waters for a viable biz model by adding web-app wrappers around ML papers who have published their code on Github? Not being accusatory or anything, just trying to understand.
The problem with NC (IMHO) is that it's too loose. Everything is up to interpretation. It's clear that you don't have a license to use the software, except the CC-NC license, but I think you probably would have to go to court to really determine if the usage was "commercial" or not. For this reason, I won't touch CC-NC works with a 10 foot pole. If you get sued, you are almost certain to go to court. There is probably a good chance that an injunction will be granted while they decide whether or not the usage is "commercial". By the time it's all over, you would have been much better off not using the code. For me, this means that using CC-NC code is pretty much a gamble on whether or not someone will attempt to sue me -- which is pretty darn close to just forgetting about licenses and hoping I don't get sued for infringement.
https://colab.research.google.com/github/AliaksandrSiarohin/...
I think this is just a wrapper of that script
https://myvoiceyourface.com/pay/
$20/mo for 20 videos. While the technical output you've achieved is notable, just curious, who is your target audience? How do you see these videos being used?
Then you also have trolls, "news channels", conspiracy theorists, youtube comedian personalities to name a few more.
The market definitely exists.
Also some manual filtering would probably be very appropriate, as there were some including racial slurs and NSFW audio and visuals.
I've tried cropping the image and the video several times to make sure only the head is visible but it doesn't work.
My image is around 250KB and the video slightly under 9MB. I'm saying this because of the two following lines:
RuntimeError: imageio.mimread() has read over 256 MiB of image data. Stopped to avoid memory problems. Use imageio.get_reader() or memtest=False.
Context, another post from 12 hours ago: https://news.ycombinator.com/item?id=23423372