ShengShu Technology burst into the scene with firsts, like Multiple-Entity Consistency, and the first commercially available generative video platform, Vidu. And the company announced today that they’ve managed to land 10 million users in the first 100 days. But it’s still early and numerous industry-wide issues are being earmarked for improvements. Chief of which includes speed and affordability.
Every time a video is generated, it takes dozens of seconds or even minutes. Generative video in its current format today might be great for generating footage for editors, or social content, but most companies now are looking at speed as a new Pandora’s Box for applications that can’t be possible without improvements in speed and cost. This applies to the programmatic advertising industry, much of which is automated, and even new methods of storytelling.
For example, the company imagines a world when generative video can be used to illustrate a story. But that story is interactive and adapts to every decision you make, like Netflix’s Bandersnatch, but with virtually unlimited endings. Maybe if you were curious as to what might happen if Harry ended up being sorted into Slytherin and had the chance to ‘influence’ the Sorting Hat’s decision, generative video just might make that possible in the future. But for this to happen, speed is of the essence.
This is where ShengShu Technology’s Vidu 2.0 technology enters the picture. It’s a major update that’s arguably a milestone for the industry. Vidu 2.0’s focus is on speedier outputs but at a much lower cost of generating each video clip, to which it credits is possible thanks to its groundbreaking technology, which it calls a “full-stack interference accelerator.”
“Vidu 2.0 features ultra-fast generation speeds, robust multimodal context handling, at a more affordable price that’s easier than ever to use. More importantly, these are the cornerstones that enable real-time content to be co-created by users or businesses, enabling them to immerse and better connect with their audiences,” said Jiayu Tang, CEO and co-founder of ShengShu Technology.
Admittedly Vidu 2.0 doesn’t generate videos instantly, but the company leads the way among competitors as it brings the time it takes to generate clips to under 10 seconds. And they managed to do this at a cost that’s 55% cheaper than the industry average.
To shed some light on these benchmarks, ShengShu Technology explains that the industry average cost of generating a clip is US $0.084 per second. Vidu 2.0 though has managed to bring that down a massive 55% to only $0.0375 per second. Better yet, you might think that if it’s faster, the quality of the video would suffer, but Vidu 2.0 makes sure that doesn’t happen.
As part of the vision behind the Vidu 2.0 update, ShengShu Technology envisions a future where text prompts – some of which come with its own complications and knowhow to get the perfect output – could eventually give way to generating clips with just a single click. This takes the guess work out of attempting to generate complex prompts through trial and error.
Vidu 2.0’s approach to this is with a “Templates” feature that users can choose from among a series of pre-set prompt templates. Templates make the addition of interactive props, or complex actions – like for example attempting to get two specific people among a crowd of five, to shake hands with one another – significantly easier.
