Tencent has unveiled an innovative tool in the world of digital content creation, a text-to-video generator called Hunyuan Video. This cutting-edge tool is designed to serve both enterprise and individual users, utilizing Tencent’s robust Hunyuan foundation model that boasts an impressive 13 billion parameters.
The Hunyuan Video tool can produce five-second videos at a resolution of up to 1280×720 (720p). Tencent highlights the tool’s Advanced 3D VAE (Variational Autoencoder) architecture, which ensures excellent visual consistency and seamless, natural motion, elevating the quality of short video content like never before.
Central to the tool’s functionality is the Multimodal Language Model (MLLM)-based text encoder, which outshines previous models in terms of text alignment, a crucial aspect that assesses how accurately the generated content reflects the initial prompt. The Hunyuan Video tool achieves a text alignment rate of up to 68.5%, while its visual quality scores are a remarkable 96.4%.
Available as an open-source platform, the tool employs a credit system to generate videos. Users can purchase various credit bundles, with rates set at $9.99 (€9.48 or £7.85) for 150 credits, $44.99 (€42.70 or £35.34) for 750 credits, and $79.99 (€75.93 or £62.83) for 1500 credits. Each video creation requires 15 credits.
AI-driven video generation is rapidly shaping up to be a significant business prospect, attracting not only Tencent but other tech giants like Google, Meta, Amazon, and OpenAI into the fray. Meanwhile, prominent Chinese companies such as Kuaishou and Alibaba are also joining the fast-evolving AI video creation arena, marking a vibrant and rapidly expanding industry sector.
