look news india
News India Live, Digital Desk: Google has recently launched its VEO 3 model, and now it’s the Chinese veteran. Today, the company has officially opened its new AI model, Hunyuanvideo-VAVATAR, which allows users to transform only one photo and stable images with an audio clip into dynamic videos. This tool has been developed by Tencent Hunyuan and Tencent Music Entertainment Group.
This release is different from a reason Is: This is not just another animated show shaking head on the soundtrack. Hunyuanavidio-Avatar also considers tone, spirit and even stage presence, which makes it ideal for short video, e-commerce content, advertising and character-based story telling.
How it works: talk to your photo
The process is simple. Users upload a photo and a voice clip (up to 14 seconds). Hunyuanavidio-Avatar then automatically detects emotional tone and visual context from the audio, and creates a vibrant avatar video in response.
The current version supports single-character mode, but TENENT has confirmed that multi-character generation will soon be available. This model works in various styles such as photorialistic, cartoons, 3D and even human verses.
The developers stated, “This system supports any input avatar image to animate in a high-melody and emotional video with simple audio conditions.” This includes changes in facial expression, gesture and even synseed lip movement.
Where it is already being used
According to Tencent Music, this model is already running in many of her entertainment apps. Although the company did not name them all, the ability to sync the characters with the voice of the model makes it naturally suitable for the Karaoke platform, livestream app and digital influencer.
This technique is also designed to support multi-sector dialogues in future updates, which can be a game-chain for creators wishing to create short video skits or educational material without the need of many actors.
Open to developers, but GPU-hungry
Single-character version of Hunyuanavidio-Avatar now official Hunayuan site But the sample is available with weight and infection code. The open-source plan includes support for both cloud-design build, Linux system, dock container and single and multi-GPU setup.
It is said that you will need a serious GPU to run it. Tencent recommends at least 24GB Vram for basic generation, but the best results come with 80GB GPU. It supports both Cuda 11.8 and 12.4 setups.
What will happen next?
The team is soon planning to open multi-sector mode, which will make the users make full views with many talk avatars. Tencent has indicated that this version will include face -ware masking and advanced emotional control among the characters using audio signals.