At ISE Barcelona, Ivan Lau, Co-Founder and CEO of Pantheon Lab, presented the essential techniques developed by his company to create lifelike digital humans for retail.
Anyone who has spent hours creating their perfect digital self in a video game knows the importance of getting it just right. Ivan Lau also knows a thing or two about virtual humans, and at ISE 2025 in Barcelona, he gave a fascinating look at the work of Patheon Lab in Hong Kong at his talk titled “The Next Frontier: Digital Humans and Generative AI in Retail Digital Signage.” Pantheon specializes in virtual shopping assistants, product recommenders, and interactive brand ambassadors — all designed to give customers a more personalized experience while shopping or frequenting locations such as post offices, transit stations, and even office buildings.
Here are five visual techniques that Lau states as key for creating the perfect digital human.
1. Synthesis
To create a convincing virtual face, Pantheon Lab combines large quantities of real human faces. Eye placement, nose size, smile lines, and any other facial characteristic that makes us human, are all mashed together to create an average of all the humans’ traits. The result is a digital avatar that passes as human based on all the familiar cues we perceive.
2. Animation
When talking to a real human, you might not consciously notice the small subtle gestures that make us… well, human. The slight shift of our eyeballs, twitches in our faces, and subtle muscle movements are things that are usually clearly absent when talking to an artificial life form. Pantheon, however, includes these slight gestures in their digital humans, making a more convincing personal assistant.
3. Expression & Reaction
When you ask someone a question, how do you expect their face to react? You probably don’t think about it too often… until you’re interacting with an AI. Pantheon uses a facial expression library that pulls from a plethora of life-like reactions and expressions to ensure customers are getting what they expect from a human — whether living and breathing or digital.
4. Lip Synchronization
Just like their library of human faces, Pantheon also uses real human voices for each virtual shopping assistant. All lip, mouth, and jaw movements are synchronized and accurately mapped to a real human speaking. The added bonus of equipping each virtual assistant with multilingual capacity is another game changer for a digital human.
Recommended: Yes, Parade Performances are Lip-Synced — Here’s Exactly Why.
5. Dynamic Body Gestures
In addition to making sure facial expressions and lip movement are realistic, the movement of the virtual shopping assistant’s body is key. Many digital humans fail to make the subtle movements in their shoulders, arms, and hands while they speak, but Pantheon’s body gesture database takes care of that. Based on contextual clues, the virtual assistants complete a variety of gestures that enhance their realism and add to their already impressive facial movements. ed AI and real-time analytics to transform the way businesses connect, engage and succeed.”
(February 12, 2025). Dana Jelter – Manager, Digital Content, AVIXA. Retrieved from https://xchange.avixa.org/posts/ise-2025-5-techniques-to-create-a-convincing-digital-human?channel_id=ai-in-av
