Synthesia AI Avatars
Synthesia is set to enhance its AI-generated avatars with realistic body movements and gesticulating hands. The upcoming full-body avatars will be capable of actions such as singing while holding a microphone, dancing, and even walking across a room. This upgrade will enable the avatars to convey a broader range of emotions, including excitement, fear, and nervousness. According to CEO Victor Riparbelli, Synthesia plans to introduce these advanced avatars by the end of the year.
"It's incredibly impressive. No one else has achieved this level of realism," remarks Jack Saunders, a researcher at the University of Bath who was not involved in Synthesia's project.
Saunders, who previewed the full-body avatars, acknowledged their quality despite minor issues, such as hands occasionally "slicing" into each other. However, he noted, "Chances are, you won't be looking closely enough to notice these small errors."
Synthesia initially introduced its hyperrealistic AI avatars, also known as deepfakes, in April. These avatars leverage large language models to synchronize expressions and vocal tone with the sentiment of spoken text. Diffusion models, commonly used in AI-generated images and videos, create the avatars' appearance. Currently, the avatars are limited to torso-up representations, which can detract from their otherwis
Before the full-body avatars are released, Synthesia is launching an intermediate version of AI avatars that feature hands and can be filmed from multiple angles. The earlier versions were only available in portrait mode and visible from the front.
Other startups, like Hour One, have also introduced avatars with hands. However, Synthesia’s version, which I tested in a research preview and is set to launch in late July, boasts slightly more realistic hand movements and improved lip-synching.e striking realism.
To develop the full-body avatars, Synthesia is constructing an even more advanced AI model. Users will need to visit a studio to record their body movements.
Before the full-body avatars are released, Synthesia is launching an intermediate version of AI avatars that feature hands and can be filmed from multiple angles. The earlier versions were only available in portrait mode and visible from the front.
Other startups, like Hour One, have also introduced avatars with hands. However, Synthesia’s version, which I tested in a research preview and is set to launch in late July, boasts slightly more realistic hand movements and improved lip-synching.
How to Create Synthesia AI Video Click Here
