Chinese tech giant Alibaba is working on Tora, a Sora—based video generation tool. This is reported by SCMP.
It is being developed based on the Diffusion Transformer (DiT) architecture, which is the basis for the OpenAI neural network presented in February.
The newspaper notes that experts have created a "trajectory-oriented DiT framework for video generation." This means that the movements created by Tora precisely follow a given trajectory, reproducing the dynamics of the physical world.
"We adapted the OpenSora workflow to convert raw videos into high-quality video—text pairs and used an optical flow estimator to highlight the trajectory," they said.
It is unknown when the new tool will become available to the general public. Previously, other Chinese companies have introduced their own AI-based video generators.
On July 24, Kuaishou's Kling model became available worldwide. Later, the startup Zhipu presented its solution — Ying.
In August, Faceu Technology, owned by the Chinese giant ByteDance (owned by TikTok), launched the Jimeng AI neural network to generate videos from promptov.
Recall that in February, OpenAI introduced a new generative artificial intelligence model Sora, which allows you to convert text into video.
In June, Toys "R" Us faced a wave of criticism after releasing an advertisement created entirely using a tool from OpenAI.