Search AI Products and News

Meta has officially launched a new short video editing function, which is based on its advanced Movie Gen AI technology, providing users with a more convenient and creative video editing experience. Users can now easily replace costumes, scenes, and styles in videos, creating amazing effects that seem to open a door to a world of wonder. This new tool supports over 50 preset effects, allowing users to choose from options that can transform video characters into comic art, marble statues, and even...

Character.AI, which is highly influential in the field of AI chat and role-playing, recently announced a series of multimedia features aimed at enhancing user experience. According to a blog post released by the company on Monday, the new AvatarFX tool will allow users to create custom animated videos, enabling them to bring their AI chat characters into new creative spaces. Additionally, the platform has introduced 'Scenes' and 'Streams' features, allowing users to make videos featuring characters and more.

The Apple Machine Learning Team, in collaboration with Nanjing University and Hong Kong University of Science and Technology, has launched an advanced AI model called Matrix3D. The main function of the model is to reconstruct real objects and scenes from a small number of 2D photos, providing users with high-quality 3D output. Users only need to provide three photos, and Matrix3D can automatically generate detailed 3D reconstruction effects. This process not only simplifies the steps of 3D modeling but also brings new opportunities to various application fields, while promoting the development of AI technology.

In the field of artificial intelligence, 3D vision and spatial understanding are becoming crucial for advancements in embodied AI, autonomous navigation, and virtual reality applications. In March 2025, Hangzhou Qunhe Technology announced the open-sourcing of its self-developed 3D vision large language model, SpatialLM, at the GTC2025 global conference, generating significant industry interest. This model, with its powerful spatial cognition capabilities and cost-effective data processing, offers revolutionary breakthroughs for robot training, architectural design, and AR/VR applications. AIbase is based on the latest information...

Japan's state-backed Rapidus Corp. has begun adjusting its chipmaking equipment and is expected to start test production of advanced semiconductors by the end of this month. This is a crucial step for Rapidus as it pushes to enter the artificial intelligence (AI) components market. The two-year-old startup aims for mass production of 2nm semiconductors by 2027, aiming to rival Taiwan Semiconductor Manufacturing Company (TSMC), a leading semiconductor manufacturer. Image Caption: Image

Midea Group has publicly showcased its first prototype of a humanoid robot. An 11-second video demonstrates the robot's ability to dance, make a heart shape with its hands, and respond to voice commands. Midea's Vice President and CTO, Wei Chang, announced the establishment of a humanoid robot innovation center, signifying increased investment in R&D and a focus on developing commercially viable products in the near term. Wei stated that Midea's embodied intelligence team is concentrating on three key areas: core components for humanoid robots, overall machine development, and the robotization of home appliances. Currently, their focus includes components such as reducers, motors, sensors, and controllers.

Google recently announced a significant update to its Gemini app on iOS and iPadOS, introducing a range of new lock screen widgets. These widgets are designed to provide users with more convenient access to the AI assistant's features, enhancing user experience. The update adds six widgets in total, allowing users to directly access specific Gemini functions, streamlining the workflow. This move by Google is clearly a deliberate attempt to attract more users.

On February 14, 2025, Kunlun Wei Group officially launched the Matrix-Zero world model, marking a significant step for China in the field of spatial intelligence. Matrix-Zero includes two sub-models: a large model for 3D scene generation and a large model for interactive video generation, aiming to reshape digital content creation patterns through AI technology and promote innovation in industries such as film production, game development, and embodied intelligence.

On January 24, Baichuan Intelligence made a significant announcement with the launch of China's first full-scene deep thinking model — Baichuan-M1-preview. This model excels across three dimensions: language, vision, and search reasoning, marking another major breakthrough in the AI field. Baichuan-M1-preview has demonstrated outstanding performance on multiple authoritative evaluation sets, particularly in mathematical reasoning and coding tasks, surpassing many well-known models, including o1-preview. It is worth mentioning that,

For a long time, efficiently generating high-quality, wide-angle 3D scenes from a single image has been a challenge faced by researchers. Traditional methods often rely on multi-view data or require time-consuming per-scene optimization, and they struggle with background quality and reconstruction of unseen areas. Existing technologies, when dealing with single-view 3D scene generation, often produce errors or distortions in occluded areas due to insufficient information, along with blurry backgrounds and difficulties in inferring the geometric structure of unseen regions. While regression-based models can synthesize new views in a feedforward manner, they struggle with complex...

Recently, Jige Technology proposed a new framework called DriveDreamer4D, aimed at leveraging prior knowledge from world models to improve the reconstruction quality of 4D driving scenes. Traditional 4D scene reconstruction methods primarily rely on two main approaches: NeRF and 3DGS. NeRF is like a super painter, capable of rendering a collection of photos into a 3D model using neural networks. On the other hand, 3DGS simulates various objects in the scene using a set of three-dimensional Gaussian functions. However, both methods have their limitations.