D-ID is Mixing a Generative AI Cocktail
CEO Gil Perry breaks down a different way to create digital people
D-ID started out as de-identification for face-ID technologies. The company developed technologies to alter photos of people that thwarted algorithms of facial recognition systems. That knowledge, along with some early experience in generative adversarial networks (GAN), came in handy when the company began animating still photographs and developing digital people.
I sat down with D-ID CEO and co-founder Gil Perry to discuss the company’s origin, the technology behind its Creative Reality Studio product, and the recent success integrating GPT-3 and Stable Diffusion. We also discuss what he learned about viral software hits from his work at Meerkat, D-ID’s role in generating 100 million animated photos for MyHeritage, and the recent adoption of its own Creative Reality studio.
You can listen to my full interview with Perry in your favorite podcast app or on the web through the buttons below. He shares a great deal about the company, the technology, and how the market is trending today.
A Different Digital People Experience
If you try out the Creative Reality Studio, you will quickly recognize it as a different type of solution from other virtual human video generators. You can select from a gallery of digital people or avatars. You might also be surprised that you can upload your own headshot, and it will automatically animate your photo to speak the script you enter or have GPT-3 write for you.
The other feature that you will likey notice is how quickly the studio renders your video. Perry mentions that the videos are generated in 60 frames per second, which is faster than real-time. It is noticeably faster than other solutions.
Let me know if you do create your own short video and give me your opinion on the experience. If you do, I recommend you upload your own headshot or create a virtual persons photo using the Stable Diffusion integration. The studio allows you to make a few videos for free as a trial.