News for hackers. Stuff that matters.
Re:Family fun with deepfakes by taoistextremist on Friday February 02, 2018 @11:11AM (#16291632)
In the realm of innocent fun with this stuff, just how much time would it take to insert myself into the entirety of a movie? I'm just imagining how fun it would be to invite friends around to watch something we're all excited about, only to convincingly insert my face into the main character's.
Re:Family fun with deepfakes by tomaskafka on Friday February 02, 2018 @11:16AM (#16291684)
Or, once advert industry learns about it, how about replacing all the things on the table in Audrey Hepburn movie with BRAND NEW LEMONMELON FLAVOURED OREOS! :)

Ideally on the fly, with each box sold to the highest bidder, adsense-style.

Re:Family fun with deepfakes by pjc50 on Friday February 02, 2018 @11:36AM (#16291855)
I've already encountered a minor version of this with the film "Demolition Man": when I first saw it on UK TV, it mentioned Pizza Hut at one point, but more recently it was on again and that line was overdubbed really badly to say "Taco Bell".

It's a weird thing to watch an film turn out differently to how you rememebered it when you're not told it's a different cut. Feels like low-grade gaslighting.

(Another example is the translation of "天下" - Tianxia - in the film Hero, which is both critical to the film and translated differently in different versions)

Re:Family fun with deepfakes by jayrobin on Friday February 02, 2018 @11:49AM (#16291989)
The original version was Taco Bell - Pizza Hut was dubbed over in international releases (the latter has a much greater international presence).
Re:Family fun with deepfakes by logfromblammo on Friday February 02, 2018 @11:32AM (#16291807)
I have also thought about this, and whether I could live with myself after making a huge pile of money to relentlessly annoy every other human on Earth.

I conceived it as video bloggers setting up a dynamic replacement volume in their scenes, and compositing a 3-D scene for the sponsored product into it as the stream goes out. One viewer might see an open Coke bottle. Another might see a half-eaten package of Oreos. Another might see framed photographs of adoptable rescue pets. All the vlogger does is avoid entering the ad-volume and the software does the rest.

Inserting into old movies would require a bit more sophistication, I think. You have to discern the geometry of the scene, model it, then discover the insertable volumes, match the camera movement in software, and finally use a shader to match the object textures to the film grain.

But let's not do this. Please.

Re:Family fun with deepfakes by yjftsjthsd-h on Friday February 02, 2018 @11:46AM (#16291944)
> Inserting into old movies would require a bit more sophistication, I think. You have to discern the geometry of the scene, model it, then discover the insertable volumes, match the camera movement in software, and finally use a shader to match the object textures to the film grain.

Only if you're changing volume. We can start with swapping flat surfaces (say, cereal boxes).

Re:Family fun with deepfakes by logfromblammo on Friday February 02, 2018 @02:54PM (#16293699)
A cereal box face is still two triangles (or a quad). It's simple geometry, but still geometry. If you can do two triangles, you could do 12. If you kept it simple, you could get away with image planes and occlusion masks, I guess.

I'm guessing this would dovetail nicely with up-converting 2-D movies to work with VR rigs. If you can reconstruct a scene well enough to fake some depth, it wouldn't be difficult to insert extra models into it.

The dystopic endgame would be to automatically generate the models from the 2-D video and randomly fill some of the volumes that don't interfere with the existing scene with advertising models.

Re:Family fun with deepfakes by brandnewlow on Friday February 02, 2018 @12:04PM (#16292117)
There was a YC company in my S11 batch that had this, 6 years ago. Not sure why they didn't pursue it, but I think in part it was too hard to get brands on board.
Re:Family fun with deepfakes by taoistextremist on Friday February 02, 2018 @12:05PM (#16292121)
This would create a new argument in support of piracy.
Re:Family fun with deepfakes by cr0sh on Friday February 02, 2018 @03:21PM (#16293885)
Many, many hours - depending on what you have for a system.

These small clips can take several hours to render; my (unwarranted) assumption is that the machine being used likely has the equivalent (minimum) of a 1060 or 1070 (at least, that's what I'd use).

Now - if you have the resources to own or build a multi-GPU machine with scads of RAM and the very best CPU - it will still take hours, but you might be able to get it down to "less than a day's worth".

That's my best guess based on what I have seen so far and my own limited personal experience with deep learning and ML tech (nothing involving face swapping or such - more mundane things thru MOOCs I've taken in the past, using my 750ti SC as the GPU with tensorflow).

Re:Family fun with deepfakes by Spectral on Friday February 02, 2018 @03:58PM (#16294158)
This might be a thing in the future, when the technology becomes fast enough that you watch movies live...starring yourself.