ShengShu Technology announced today Vidu Q1, a new generative video model that’s taking on the VFX industry. If you’re not familiar, Vidu in its V1.5 update was the first to introduce Multiple-Entity Consistency – where subjects including objects, people and the environment behave and look consistent frame-by-frame. Sometimes with less developed generative AI solutions, character faces might suddenly change or contort – think the early days of the Will Smith eating spaghetti video.
Vidu Q1 Generates Believable First-to-Last Frame Transitions
With Vidu Q1, the company now delivers higher-quality animation and HD quality footage and audio – specifically for transitions and generating soundtracks and sound effects. And all of this is powered entirely by text prompts. Lowering the barriers to entry for Hollywood-esque productions, Vidu Q1 offers a new feature “First-to-Last Frame” for cinematic-grade transitions. After uploading two images or film stills, which could even be irrelevant to one another, Vidu Q1 can analyze and generate a highly believable transition between the two. The point being that the resulting footage with the transition won’t look jarring. Anyone that’s attempted to be a content creator would know how powerful of a feature this could be in cutting both time and cost for VFX.
“Vidu Q1 marks a pivotal step forward in making video generation smarter, more expressive, and more accessible,” said Yihang Luo, CEO of ShengShu Technology. “This new generative video model brings us closer to our vision of building the next-generation content creation platform – one that empowers anyone to turn imagination into reality with unprecedented ease, creative freedom and cinematic precision.”
Audio Generation Completes the End-to-End Loop for Video Editing
Moreover, Vidu Q1 is rolling out a new audio editing capability that offers users to generate both background music and sound effects, but also allows for the ease of editing or splicing audio into the film through simple text prompts like “10-12 seconds add rain.” The resulting audio is also output in high-fidelity audio of up to 48 kHz – Vidu notes that this is a first for any generative video company, which underlines the major leap in quality from ShengShu Technology (owner of Vidu). In fact, Vidu Q1’s AI can even match the tone and mood of the film so users don’t have to be a musician, and also puts the power of audio engineering in their hands by enabling users to add multiple layers of audio for up to ten seconds.
Vidu Q1 Lands #1 Spot on VBench Generative Video Benchmark
For all the attention to quality, Vidu Q1 has a quantifiable claim. The company announced that it outperformed all competitors, even OpenAI’s Sora and Google Gemini, to land itself in pole position on VBench’s generative video benchmark. And with the momentum behind the company, Vidu is getting closer and closer to Hollywood as the company recently co-announced a sci-fi short film series with Aura Productions, which will use Vidu (including Vidu Q1).
