This first-of-its-kind video model from a Chinese AI company that can accurately control characters at the minute level is here | WAIC 2024

SenseTime has done something great again, and it can generate a one-minute video with just a photo.

On July 4, Vimi, the first large controllable character video generation model for C-end users created by SenseTime, was selected as the "Treasure of the Hall", the highest honor for the World Artificial Intelligence Conference (WAIC) exhibition display. The most innovative exhibit at the conference.

Based on SenseTime's new large model, Vimi can generate a character video that is consistent with the target action through just a photo of any style, and supports a variety of driving methods, which can use existing character videos, animations, sounds, text, etc. Driven by multiple elements.

With the rapid development of large models and generative AI technology, it is no longer difficult to make still photos "alive".

However, existing products on the market still have many problems in practical applications, such as inaccurate movement and expression control, poor effect stability, and limited video duration, which limit the creative space of video creators.

In order to break through these bottlenecks, SenseTime has launched a large controllable character video generation model – Vimi.

Unlike picture expression control technology that can only control head expression movements, Vimi can not only achieve precise expression control of characters, but also control the natural body changes of the characters in the photos in the half-body area, and automatically generate hair, hair, and hair that match the characters. Costume and background changes.

At the same time, Vimi can also reasonably generate light and shadow changes, making character movements and visual effects smooth and natural, and the picture harmonious and beautiful.

More importantly, Vimi can stably generate single-shot character videos up to 1 minute without deterioration or distortion over time, fully meeting the demand for long-term stable videos in entertainment and interaction scenarios.

Ririxin officially stated that Vimi will be fully open to C-end users, especially to meet the entertainment and creation needs of female users.

Users only need to upload high-definition pictures of people from different angles, and digital avatars and photo videos of different styles can be automatically generated.

The video characters generated by Vimi are no longer just dull facial movements, but are matched with gestures, limbs, hair, etc. to form more complete and unified character movements, allowing creators to edit and recreate based on the generated video materials.

For users who like to take selfies, Vimi supports a variety of entertainment and interactive scenarios such as chatting, singing, dancing, etc.; for users who are keen on emoticons, Vimi can generate a variety of interesting character emoticons through a single photo, with various ways of playing and achieving creative freedom.

In addition, Vimi also provides a variety of generation styles such as aesthetic photo style and fantasy style, allowing users to travel through different dimensions and enjoy blockbuster-like immersive visual effects.

At present, 80% of the videos in the industry have characters as the main content. At the same time, with the explosion of short videos and live broadcast platforms, the demand for videos with characters as the main content has increased exponentially.

However, there is a lack of AIGC products on the market that can control character video generation, which leads to an urgent need to improve the efficiency of content creators.

It can be said that the advent of Vimi has just filled this gap, providing a simple and easy-to-use creation tool for the majority of C-side video creators, which can be widely used in the creation of character short videos.

More details of Vimi will be unveiled at the "Love Without Boundaries·Xiang Xinli" artificial intelligence forum hosted by SenseTime on July 5. APPSO will also follow up with reports.

Currently, Vimi has opened reservations on the official website of SenseTime, and you are welcome to experience it.

Attached is the reservation address portal: https://vme-int.softsugar.com/questionnaire/

# Welcome to follow the official WeChat public account of Aifaner: Aifaner (WeChat ID: ifanr). More exciting content will be provided to you as soon as possible.

Ai Faner | Original link · View comments · Sina Weibo