Byte Debunks Sora Rumors: Incomplete Product, Lags Behind Global Peers

On February 20th, according to reports, before Sora exploded onto the video creation track, ByteDance, a domestic company, also launched a groundbreaking video model called Boximator.

Unlike models such as Gen-2 and Pink1.0, Boximator can precisely control the actions of characters or objects in generated videos using text.

In response to this, sources at ByteDance stated that Boximator is a research project on technical methods to control object movements in the video generation field and is not yet ready to be launched as a complete product.

Moreover, there is still a significant gap between Boximator and leading foreign video generation models in terms of image quality, fidelity, video duration, and more.

As reported earlier, OpenAI recently released the first video generation model, Sora, which can generate a high-definition video of 1 minute just by inputting prompts, being considered a game-changer technology in the entire video generation field.

The model can generate complex scenes containing multiple characters and specific types of movements, precisely capturing details of objects and backgrounds.

The official website has already updated 48 video examples where Sora accurately presents video details, demonstrating a profound understanding of the existence status of objects in the real world and creating characters with rich emotions.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.