In this live stream, Bob Doa Media introduces viewers to Dulu Stream, an AI-powered live video tool that transforms live broadcasts into animated characters in real time. While acknowledging that Dulu Stream is not perfect and has some inconsistencies, the host emphasizes its impressive capabilities, especially the ability to convert live video into animated characters with lip-syncing and movement. The stream showcases the interface, explaining how users can select or upload any image—whether a person, animal, or cartoon—as a character, and how the software handles lip movement and articulation with varying success depending on the character model used.
The host demonstrates the flexibility of Dulu Stream by creating characters from text prompts, such as an elderly wizard or a pageant contestant, and changing backgrounds using Nvidia Broadcast software. The stream highlights differences between two versions of the AI model, Lucy 2.1 and Lucy 2.5, noting that while the newer 2.5 model offers more advanced visual effects and environmental changes, it sometimes sacrifices smoothness and lip-sync accuracy compared to the older 2.1 model. Various examples are shown, including characters in different settings like beaches, arcades, and dungeons, as well as the ability to change clothing and add props in real time.
Throughout the stream, the host experiments with creative and whimsical prompts, transforming the character into a wide range of personas such as a pirate with a monkey, a superhero with elemental powers, a steampunk inventor, and even abstract concepts like a human kaleidoscope or a sandstorm. These demonstrations showcase the AI’s ability to interpret complex and imaginative descriptions, adding visual effects like bubbles, fire, and glowing objects. The host also explores humorous and unusual prompts, including a talking toilet and a ventriloquist dummy, highlighting both the entertaining and sometimes quirky results produced by the AI.
The live stream also touches on technical aspects, such as the challenges of syncing lip movements perfectly, the frame rate limitations, and the potential for integrating third-party voice changers like 11 Labs voices, though native support is not currently available. The host encourages viewer interaction by inviting prompts and suggestions for future live streams, emphasizing the interactive and experimental nature of using Dulu Stream for creative projects, storytelling, and even music videos. The stream serves as both a tutorial and a showcase of the AI’s current capabilities and limitations.
In conclusion, Bob Doa Media’s live stream provides an in-depth and entertaining exploration of Dulu Stream’s AI video technology, demonstrating its potential for real-time character animation and creative expression despite some imperfections. The host’s enthusiasm and willingness to experiment with diverse prompts highlight the tool’s versatility and inspire viewers to imagine new possibilities for AI-driven live content creation. Feedback and ideas from the audience are welcomed to help shape future streams and projects using this innovative technology.
