Rest in peace Hollywood?
Open source text-to-video model to compete with industry leaders
Text-to-video conversion is still in its infancy, but there is no doubt that the technology is gaining momentum. So let’s take a closer look at Zeroscope, an open-source model that’s currently making waves in the AI scene.
(No time to read. Want to try out the model? Jump to the end of this article)
Indeed, companies like RunwayML already exist with enormous computational and financial resources, and have already released powerful video-to-video and text-to-video tools in Gen1 and Gen2. increase. And some companies, like Google with Phenaki and Meta with Make-A-Video, still keep their video models secret.
But in this emerging industry scenario where big tech companies are already entering, Zeroscope is suddenly entering the industry as a completely free-to-use open source alternative.
This is notable because in the past, communities around open source projects have created innovative workflows that were later used by big tech companies. Take a look at what the Stable Diffusion community developed a year ago and recently implemented by companies like Adobe and Midjourney.
A great strength of the open-source model: Any developer can further improve the model. And Zeroscope could follow a similar path.
And what happens to a technology that is still in its infancy when there is a community pushing development?
dual approach
Following a familiar approach from early model versions of Midjourney, Zeroscope is based on two components.
- zero scope_v2 567wdesigned for rapid content creation at a resolution of 576 x 320 pixels for video concept exploration.
- Zeroscope_v2_XL,a…
