Apple has recently launched LiTo, a new AI model that can reconstruct 3D objects from an image while accurately preserving lighting effects like reflections and highlights. The end product is realistic and better than previous techniques.
The model transforms visual information into numerical data to understand both an object's shape and how light interacts with it. This concept is called latent space.
The process involves two important steps. First, an encoder compresses the image into a compact representation, and then a decoder reconstructs it as a 3D object. The model adds details such as shadows, reflections, and lighting changes throughout the process.
Traditional models focus only on shape or basic surface details, while Apple’s latest AI technology takes it a step further. It combines geometry with lighting behaviour to produce objects that look hyperreal even when viewed from different angles.
Researchers utilized thousands of images of objects from different viewing angles and under different lighting conditions to train the model. The AI system developed its abilities through incremental learning rather than simultaneously processing all information.
Apple’s technology has several applications across industries. For example, it can be used to enhance video games, virtual reality apps, and product design. The AI model can also help create improved user experiences on digital platforms.
The new AI model shows how technology is blurring the lines between reality and synthetic digital creation. If the technology is developed further, it could change how we design and interact with 3D content.
Its prediction capabilities for complete 3D shapes and their lighting effects provided better results across the board. With the pace at which AI is evolving, creating lifelike models from simple images may quickly become common in the near future.