On June 29, the famous generative AI platformRunwayannounced that its Bunsen video platform, Gen-3Alpha, is opening up to some users thetest.
Gen-3Alpha is the latest product launched by Runway on the 17th of this month. Compared with the previous generation, it achieves substantial improvement in light and shadow, quality, composition, textual semantic restoration, physical simulation, and action consistency/coherence, and refers to OpenAI's Sora.
It should be noted thatGen-3 can't generate a soundtrack, the sounds for all these pieces are added by themselves. Currently, only Google's VideoFX can generate videos with music.
Someone also made a short video with Gen-3 that focuses on the inspirational story of racing, dreaming, and never giving up. The story framing, dribbling, and close-ups are all good for a complete microfilm story.
The following video is a demonstration of Gen-3's powerful textual semantic reduction capabilities, and the author states that her prompt was "hand-drawn pencil art style rabbit fur girl".
The rabbit fur here is a misnomer, it should be "rabbit-eared girl", but the final Gen-3 result is still very correct!.
Runway says it will soon be available to everyone as testing continues.
Gen-3 address: https://runwayml.com/blog/introducing-gen-3-alpha/