With the new ‘Indiana Jones,’ film enters its generative AI era
Well, it’s April 28, 2023, which means only two things for me. First, it means that we are barely two months away from the release of Indiana Jones and the Dial of Destiny, which I will be attending along with my 6-year-old dressed in full Indy regalia (I’ll go as Dr. Henry Jones). According to Deadline, the movie features a whopping 25-minute intro that takes place in 1944, starring Harrison Ford, de-aged by AI.
This news is already exciting and impressive without even seeing the result. After disappointing brief uses of the AI in The Mandalorian and The Book of Boba Fett, it clearly indicates that the producers have total confidence that artificial intelligence is ready to put a spell on the audience. Some 25 minutes of seeing a thirtysomething Indy punching Nazis is a lot to take if it’s not as flawless as the real footage in the original movies. This movie may very well mark the frontier of a new future, a point of no return for film and TV production in which reality and fiction merge seamlessly to create any story you can imagine with total suspension of disbelief.
Oh, and it’s Friday, so it’s time for my weekly generative AI roundup. At the risk of being hyperbolic, this week has blown my eyes through the back of my skull. It’s been really hard to pick just three for today’s menu: Apple’s impressive lighting AI, a head-spinning new text-to-speech synthesis AI, and what may be the very first commercial 100%-created-by-AI product.
Apple’s FaceLit: change the angle and light of your portraits
Apple has been conspicuously absent from the past months’ AI gold rush. It was as if the company was content spending its days listening to Siri knock-knock jokes until Microsoft and OpenAI ate it alive. But as a new research paper shows, we may be in for a wild ride soon. Check this demo:
It’s name is FaceLit and it seems hard to believe we will not see this added to every single Apple device in a few months. According to the researchers, the technology can generate a 3D portrait from your mugshot, for you to change the angle and illumination at any point. It reminds me of SwitchLight (from last week’s roundup) but it’s potentially more impactful because every Apple user may get it built in on their devices.
Bark: the AI voice that stole my heart
I confess I fell in love with Scarlett Johansson’s AI character in Spike Jonze’s Her, perhaps the most prescient movie in the last 50 years. Listening and playing with this new text-to-speech generative AI had the same effect on me. The sighs, the laughter, the breathing . . . It’s so believable that it is unbelievable (as in “I can’t believe they already achieved perfect human voice synthesis”).
I have never heard anything like this before. You can try Bark now on this Hugging Face space, install it on your computer if you have the skills (ask ChatGPT to help), or wait until it inevitably hits your phone or computer in a few weeks, as it is free and open source.
Generative pizza ad: “It’s like family but with more cheese”
Redditor Pizzalater created an ad wholly generated by AI. GPT4 wrote everything, from the script to the tagline. MidJourney made the images, and Runway Gen-2 handled the video. The voice-over isn’t real either: It was made with voice-cloning platform Eleven Labs. And the music is also composed and performed by AI using a platform called Soundraw AI Music.
Sure, it looks like a nightmare dreamed up by Marcel Duchamp and H.R. Giger, but it is impressive. Give it two or three years, and like Tom Graham recently told us, this technology is going to completely destroy Hollywood as we know it. Everyone’s generative AI home movies and ads will be as crystal clear and believable as Indy 5.
(18)
