From self expression to gaming and customer
service Magic Unicorn provides superior, optimizable, and scalable 3D avatar technologies
powered by deep learning and advanced computer graphics algorithms.
Built using the proprietary data set of over
20,000 3D face scans collected by Magic Unicorn over 3 years.
Avatar can be automatically created by taking a
single selfie.
All processing is done by a host device.
No server load, no privacy concerns.
It takes less than a second on a dated low-budget
mobile device to fit a face shape, skin color, hairstyle, eyebrows, eyelashes, beard, mustache,
and many more elements.
No need for tiresome manual setup.
Everything is customizable though. There is an
extensive library of facial features and accessories. Each of them is perfectly fitted to any
face shape by the special algorithm.
There are hands for better expression
And bodies
And they are customizable too
We use a FACS-based animation system which
encodes face deformation as separate muscle movements. It can be driven by several means
Everything is done on the fly by our custom built
video-to-animation system. Key features:
Face and landmarks detection (mobile and PC)
Real-time high quality facial expressions tracking at 60 fps
Head position and angle detection
Small model size
RAM and GPU memory efficient
By face tracking
Our speech-to-animation technology is the only
one available working on mobile at the moment.
Key features:
Generalizes to different languages and people
Real-time high quality voice-based facial expressions animation on mobile devices
Head angle and wink voice-based animations
Small model size
RAM and GPU memory efficient
By audio
Animator-friendly face and body rigs are
available for Maya and Blender with a single click export scripts.
By hand
There is a large set of personalized stickers
All are animated
Some work for two people
Non-human characters are also supported and
provide some degree of personalization
Streaming features
1
Duplex stream audio
with realtime facial animation metadata for avatar
2
Opus codec for hi
quality in-game voice chat
3
Serverside relay packet
validation
4
Possible both
peer-to-peer and unicast way streaming
5
Broadcast physics
events for distributed physics calculations