Tencent Unveils Its Own Text-to-Video AI Tool: A New Rival to OpenAI’s Sora
Tencent has introduced HunyuanVideo, a powerful text-to-video tool with 13 billion parameters, competing with OpenAI’s Sora in the AI-driven video creation race.
Tencent, the big Chinese company known for games and social media, has launched a new tool that can create videos from text. This tool is part of Tencent’s AI model called Hunyuan. It aims to compete with other companies like OpenAI, which made a similar tool called Sora.
HunyuanVideo: A Powerful Tool
Tencent’s new tool is called HunyuanVideo. It can make short, high-quality videos by just using a few words typed in by the user. Tencent says it has more than 13 billion parameters, which means it is the biggest open-source model for making videos. The tool is free for both businesses and individuals to use.
How Do Text-to-Video AI Tools Work?
Text-to-video AI tools, like Tencent’s HunyuanVideo and OpenAI’s Sora, use advanced technology to turn written words into moving images. They rely on a system called a Diffusion Transformer, which helps the AI understand the words and break them down into visual parts. The tool reads the text, processes the details, and creates frames that combine to form a video. It adds realistic effects like lighting, camera angles, and reflections to make the video look lifelike. The more specific the text prompt, the better the AI can generate scenes. However, these tools are still improving, so sometimes not all details from the text show up in the final video.
What Can HunyuanVideo Do?
HunyuanVideo is good at making videos that look very real. For example, one video shows a surfer riding a big wave and spinning around. In another video, hot air balloons rise into the sky while campfires burn below. These videos show how well the tool can change camera angles and make things look lifelike.
Tencent vs. OpenAI
Tencent is not the first company to make a tool like this. OpenAI introduced a similar tool called Sora earlier this year. Tencent’s tool uses the same type of technology called Diffusion Transformer, which is also used in Sora. Other Chinese companies like Kuaishou and Alibaba are also working on text-to-video tools.
Here’s a quick comparison between Text to video AI tools:
| Company | Tool Name | Launch Date | Key Feature |
|---|---|---|---|
| Tencent | HunyuanVideo | 2024 | Free, creates high-quality videos |
| OpenAI | Sora | 2024 | Text-to-video generation |
| Kuaishou | Kling AI Model | 2024 | Short-video creation |
| Alibaba | Tora | In progress | Based on OpenAI’s OpenSora model |
How HunyuanVideo Works
HunyuanVideo can make short, 5-second videos in 720p resolution. Users can type prompts in English or Chinese, and the tool will try to make a video based on those words. For example, one video created by the tool showed a Chinese lady cooking. Another showed a cyberpunk-style animated cat. However, not all actions from the prompts were always included in the final video.
How to Try It
Anyone can try HunyuanVideo through Tencent’s AI chatbot app, Yuanbao. Users can give it text prompts and choose video styles, quality, and size. The app also has advanced settings for camera movements, which make the videos look more professional.
Upgrades to Hunyuan
Tencent is always working to improve its AI tools. Last year, they introduced the Hunyuan model. In March 2024, they added a tool to turn images into videos. By May, they improved their text-to-image tool, making it 20% better than before. Tencent’s launch of HunyuanVideo shows how quickly the race to make better AI tools is growing. This new tool is a strong competitor to OpenAI’s Sora and other similar products, making the future of AI-generated videos exciting.