News for hackers. Stuff that matters.
Re:Two recent studies on judging authentic vs. deepfake images by goto11 on Thursday March 10, 2022 @10:48AM (#30628222)
Fully animated actors can play just as effective and convincing as flesh-and-blood actors, as shown by Disney or Pixar. But is also requires immense amounts ow works and talent - neither which is cheap.

Star Wars certainly didn't use deepfakes because they though it was cheaper or easier.

Re:Two recent studies on judging authentic vs. deepfake images by wutbrodo on Friday March 11, 2022 @09:44AM (#30640493)
I agree that it won't obsolete actors as GP says, but it will absolutely be cheaper and easier. The expense of acting doesn't come from the craft being expensive or rare. It comes from the market power that results when the strongest connection we feel to a performance is inseparable from the performer (ie their face). There's a reason that editors and cinematographers don't have fans outside of film nerds,and it's not that because their craft is trivial or that they don't contribute to the film. Technology costs can be reduced, but market power is what allows them to collect tens (or hundreds!) of millions of dollars. Commoditization via deepfakes turns them into just another employee.

You can look at voice acting as an example. Before Robin Williams' performance in Aladdin started the trend of recognizable celebrities as voice actors, there was a thriving community of voice-only actors in big-budget movies, hired for their craft instead of their brand. And yet, you probably can't name a single one, despite being able to name plenty of actors from non-animated films of the era. They were also inherently replaceable, as memorably parodied by the simpsons: https://youtu.be/g5ffaTv4ajg

It's notable that the studios switched to celebrity actors precisely for the brand that they brought from their non-voice roles.

Deepfakes are pretty much identical to this animated case, allowing it to be extended to physical movies. The recognizable portion of a performance becomes a few model files instead of an actor's face, and the job of a (replaceable) human actor is to play the role of the performer, not just the character.

Re:Two recent studies on judging authentic vs. deepfake images by goto11 on Tuesday March 15, 2022 @07:51PM (#30693494)
You mean you will hire a cheap unknown actor and then add Tom Cruise's face via deepfake, and then market the movie as a Tom Cruise movie? Even if that was legal (which of course it isn't), surely the marketing value of the actors name would disappear if every low-budget movie could use deepfakes of the actor.
Re:Two recent studies on judging authentic vs. deepfake images by wutbrodo on Friday March 25, 2022 @10:56AM (#30802929)
No, I'm suggesting that the tentpole performer will have an entirely synthetic identity.
Re:Two recent studies on judging authentic vs. deepfake images by krapp on Thursday March 10, 2022 @10:52AM (#30628275)
They aren't cheap now, but that's only because the automation and AI pipelines aren't ready yet.
Re:Two recent studies on judging authentic vs. deepfake images by goto11 on Thursday March 10, 2022 @11:02AM (#30628409)
You still need someone to - you know - act.

Saying deepfakes makes actors superfluous is like saying synthesizers makes musicians superfluous.

Re:Two recent studies on judging authentic vs. deepfake images by ivarv on Thursday March 10, 2022 @11:50AM (#30629048)
there's the deepfake visual representation that we're discussing, but what's to say there isn't work being done on the acting side of things? I think there are broad horizons to what can be automated - especially in the world of pure media.
Re:Two recent studies on judging authentic vs. deepfake images by goto11 on Thursday March 10, 2022 @12:04PM (#30629264)
You mean some kind of AI which can read and understand a script and then develop a persona with physical appearance, voice, mannerisms, interpretation of emotional state and interactions with other characters etc.?

I guess such an AI would also be able to write and direct the movie.

Re:Two recent studies on judging authentic vs. deepfake images by krapp on Thursday March 10, 2022 @12:20PM (#30629487)
Realistically there would be a pipeline with various companies and AIs involved, each specializing in a specific aspect of virtualizion (the way effects houses work today,) not a single AI doing everything.

But yes, everything you mention up to and including writing and directing can and will be automated to the fullest degree possible in time.

Re:Two recent studies on judging authentic vs. deepfake images by jayd16 on Thursday March 10, 2022 @11:26AM (#30628714)
Deepfake tech isn't going to get you AI actors. It's just not the same thing at all.
Re:Two recent studies on judging authentic vs. deepfake images by krapp on Thursday March 10, 2022 @12:19PM (#30629460)
It's a piece of the puzzle that will.
Re:Two recent studies on judging authentic vs. deepfake images by skeeter2020 on Thursday March 10, 2022 @11:02AM (#30628417)
this doesn't address the talent behind the images, both cast and crew. Why would optimizing the production workflow make either of those cheaper? In more general software development payroll costs are higher than the rest of costs combined, seems the same scenario in more animation as well.
Re:Two recent studies on judging authentic vs. deepfake images by SketchySeaBeast on Thursday March 10, 2022 @12:12PM (#30629375)
And even now Pixar and Disney are really just melding real life and animation - they are working off real performances and choreography to generate those images. Still need to have people in the pipeline somewhere to have things look right.