D-ID creative studio has developed a new live portrait tool. It allows users to create AI-generated presenters for marketing and educational content and more.
The live portrait tool combines the conversational text generation of GPT-3, animation powered by Stable Diffusion, and the choice of a set of AI voices or the ability of users to upload their own audio.
The Future of Podcasting?
D-ID's tool potentially transforms the way creators and businesses can create content, in theory negating the need to spend hours filming and recording. Users can simply repurpose existing photographs of themselves with a voice recording or an AI generated voice.
Creators who misspeak during a recording for a podcast or forget to mention a key topic no longer need to re-record but can have live portraits fill in the blanks.
In practice the lip and head movements in real life subjects are still a little jarring and it’s still possible to tell the image has been manipulated by a computer.
A Storytellers Dream
The live portrait tool works much better with their own AI generated presenters. It also works with AI images generated using Midjourney and DALL-E giving digital artists the ability to animate their work and tell a story.
Other use cases for the Live Portrait tool include automated chatbots presented by an animated picture of a real person, something the company has capitalized on with their other chatbot tool Chat D-ID.
D-ID have previously licensed their software to family history site My Heritage. Their Deep Nostalgia tool, allows budding genealogists to upload photos of their ancestors and bring them back to life.
According to My Heritage this has now been used over 100 million times. Their next generation LiveStory tool lets users not only animate their ancestors but allows them to tell a story about their experiences.
Ethics and Privacy
All of D-ID’s tools prevent users from uploading pictures of real life figures to prevent the abuse of deepfakes. I tried to upload a photograph of British King George V and to pair it with his real life Christmas radio broadcast from 1935 to no avail. It seems this policy applies even to public figures, long deceased.
In their ethics statement, D-ID also pledges not to ‘knowingly license the use of our platform to political parties. Nor will we knowingly work with pornography publishers or terrorist organizations, gun or arms manufacturers.’
D-ID was founded in 2017 by Israeli computer scientists Gil Perry, Sella Blondheim and Eliran Kuta and is backed by venture capitalists.