Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks

junyanz.github.io

Post reply on HN
Print view View original

21–30 of 146 posts

Re: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks

#21

Post by pdelbarba »

king_magic wrote: post #7

This is really incredible. Maybe a neat idea: artist draws/paints frames for animation, frames are converted into semi-photorealistic images through this software, and assembled into a movie.

Welcome to HN...

I wish they had a more samples of that. I'd be interested to see where the threshold is in terms of detail to get something vaguely photorealistic

edit: in the other post there's a link to the paper with some good examples. I think the amount of detail you'd need would be pretty prohibitive in terms of artistic effort. On the other hand, it would be interesting to use this on something like a rough CG rendering to try to correct some of the shading imbalance and such.

Re: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks

#22

Post by haloboy777 »

lettergram wrote: post #16

At this point, I'm wondering "What's real any more..." Seriously, if this continues, I don't know how to keep up with this field. I spend at least an hour a day just reading about the work that has been done (i.e. reading the research).

I guess you already have read about Prisma App. Here is the research paper it is based on.. -> https://arxiv.org/pdf/1508.06576.pdf

Re: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks

#23

Post by jimmies »

While this is undoubtedly very impressive, I think it's just another logical step to what we've been seeing so far:

- In the past, you needed to have a pianist at home to perform you a song, with the music box and then the phonograph you don't need to hire anyone anymore. It's probably not as good as a live performance (maybe?), but it's good enough for many people, and much much cheaper, faster, and available.

- You needed advanced knowledge and equipments at home to produce magazine-style tri-fold leaflets or wedding invitations, with modern word processors you can use a template and be alright. It's probably not as good as a professionally customized design (maybe?), but it's good enough for many people, and much much cheaper, faster, and available.

- You used to hire a photographer or an artist to have your portrait photographed/painted, now you can do with your NVIDIA card at home. It's probably not as good as a professionally painted one, but it's good enough for many people, and much much cheaper, faster, and available.

Re: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks

#27

Post by jordache »

TIL - remap green pixels to grey/white and you go from summer to a contrived notion of winter.

This done easily in Photoshop.

- select green pixels, smooth it a bit, then paint white over it.. - apply a blue cast on it

https://s17.postimg.org/q68dz04sf/test.jpg

Re: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks

#28

Post by soylentcola »

coreyp_1 wrote: post #2

I thought that the iPhone photo to DSLR was interesting. (The others are, too, but I thought that this effect was particularly well done.)

I guess I'm a bit jaded since adding depth of field via software is a bit more common these days but overall, this project is still really impressive. Even with obvious artifacts it's not something I'd have considered possible 5 or 10 years ago.

Re: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks

#29

Post by eanzenberg »

thriftwy wrote: post #6

The next step is "turning pencil drawings into photos" and using it to fabricate evidence on grand scale. Why bother catching politicians doing something when you can just draw them in? Will wreak havoc on societies with weak politics/reporting culture.

So... photoshop?
Post reply on HN