Huya has launched VAM 1.0, a real-time multimodal digital human model that can engage in conversations, sing, dance, and play games, all from a single photo input. The model is built on a DiT architecture and can operate continuously for over 24 hours, outputting at a resolution of 480x832 at 28 frames per second. Unlike previous AI digital humans that often felt like pre-recorded videos, VAM 1.0 offers genuine real-time interaction, including the ability to handle interruptions, adapt to user preferences for address, and maintain conversational flow. The technology addresses key industry challenges such as temporal stability, interactive capabilities, and computational efficiency, aiming to enhance applications in live streaming, e-commerce, and news broadcasting. AI
IMPACT Sets a new benchmark for real-time interactive AI digital humans, potentially accelerating adoption in live streaming and virtual content creation.
RANK_REASON Product release from a significant AI lab (Huya) with a new model version (VAM 1.0). [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
