Lemon Slice, a digital avatar generation company, is expanding the capabilities of AI chatbots by introducing a video layer through its new diffusion model. This technology allows for the creation of customizable digital avatars from a single image, enabling more interactive AI agents within apps.
The Lemon Slice-2 model enables the generation of dynamic digital avatars that can engage with users, addressing tasks ranging from customer queries to mental health support. Co-founder Lina Colucci highlighted the importance of video interactivity, drawing inspiration from tools like ChatGPT.
Lemon Slice’s solution, offered through an API and embeddable widget, features a 20-billion-parameter model capable of livestreaming videos at 20 frames per second on a single GPU. Users can customize avatar backgrounds, styles, and appearances to cater to diverse requirements.
Furthermore, Lemon Slice focuses on creating both human-like and non-human avatars, leveraging ElevenLabs’ technology for generating avatar voices. By utilizing a general-purpose diffusion model, the company aims to differentiate itself from competitors and deliver superior avatar solutions.
Source: TechCrunch
Leave a Reply