Software and Services

Video: Creating hyper-realistic historical characters with AI

20 January 2022

Researchers at MIT have used artificial intelligence (AI) to create hyper-realistic digital characters of famous people from history such as Albert Einstein, Martin Luther King, Jr., Mona Lisa and more.

The move comes as concerns over AI creating digital characters may lead to misleading or even dangerous deepfake content.

MIT wanted to show how AI could be used in a positive manner and specifically created hyper-realistic characters of pioneers in history. But it could also be used to preserve people in time or chart how someone ages over time such as a child.

MIT compiled an open-source character generation pipeline that combines AI models for facial gestures, voice and motion as well as a variety of audio and video outputs. The pipeline marks the output with a traceable watermark to distinguish it from authentic video content and to show how it was generated to prevent malicious use.

Researchers hope that the open-source system will allow teachers, students, healthcare workers and professionals to explore how the tools could help in respective fields. MIT said the results could improve health and well-being or contribute to personalized education.

Using the system, MIT was able to create a synthetic version of Johann Sebastian Bach, which had a live conversation with cellist YoYo Ma in MIT’s lab during a musical interfaces class.

Real-world uses

MIT said that digital, hyper-realistic characters could be used in real-world situations. For example, with the growing shortage of mental health professionals, AI characters could deliver therapy to people with social anxiety or even those with deeper mental health concerns.

Additionally, the technology can be used to anonymize faces in videos while preserving facial expressions and emotions, which could be useful in sessions where people want to share personal information such as health or trauma experiences but don’t want to reveal their identity. It could also be useful for whistleblowers or witness accounts.

MIT students in a deepfakes class were able to create an animated figure in a historical Chinese painting and were able to create a dating breakup simulator.

The full research can be found in the journal Nature Machine Intelligence.

To contact the author of this article, email PBrown@globalspec.com


Powered by CR4, the Engineering Community

Discussion – 1 comment

By posting a comment you confirm that you have read and accept our Posting Rules and Terms of Use.
Re: Video: Creating hyper-realistic historical characters with AI
#1
2022-Feb-08 7:33 AM

The setting of ethical standards and taking precautions with this software, like with any criminal enterprise, reaches to the limits of the basically moral of society. They are of none effect on designing and unscrupulous actors of nefarious intent. As shown in other contexts, this may be contemporary real or fictional characters.

It is for this reason that mankind needs to institute immediately in every national and international law, the absolute prohibition of the assignment of rights of personhood, individual , corporate, otherwise to AI manifestations of any kind, whether physical, cyborg, automaton, android, virtual, hardware, software, by legal fiction or any other definition, or by majority vote of governance or the people respectively.

Engineering Newsletter Signup
Get the GlobalSpec
Stay up to date on:
Features the top stories, latest news, charts, insights and more on the end-to-end electronics value chain.
Advertisement
Weekly Newsletter
Get news, research, and analysis
on the Electronics industry in your
inbox every week - for FREE
Sign up for our FREE eNewsletter
Advertisement