Researchers from the University of Hong Kong (HKU) led by Professor Yifan Evan Peng of the Faculty of Engineering have developed two methods that help holographic displays more convincingly convey changes in focus, perspective and the appearance of materials. The work combines computer graphics, wave optics - the branch of physics dealing with the behavior of light as a wave - and calibration based on artificial intelligence. The results were obtained in computer simulations and on laboratory display prototypes.
A holographic display reconstructs light waves rather than simply showing a flat image. Thanks to this, it can imitate how, in a real room, near and far objects come into focus at different distances, how surfaces hidden at one angle become visible, and how glare slides across glossy surfaces. But calculating such waves is difficult: many existing methods first describe the scene with a set of images from selected depths or viewpoints, and then convert them into a hologram. Along this path, some information is lost, which makes blur change unnaturally, objects appear incorrectly, or the effect work only from certain points.
In one study, published in the journal Nature Communications, the team starts with polygonal meshes - interconnected surface patches from which three-dimensional models are built in animation, engineering and scientific visualization. This approach preserves the description of the surfaces and depth of the scene, not just a set of flat frames. "Using meshes allows the three-dimensional structure of the scene to be preserved throughout the entire hologram generation process," explained Xiangyu Meng, one of the first authors of both works.
In addition to image brightness, the researchers take into account the amplitude of the light wave - its strength - and the phase, that is, its position in the oscillation cycle. Together they determine how light propagates and forms an image at different distances. Calibration with the help of AI bridges the gap between calculation and the real display: a trained model uses camera measurements to compensate for imperfections in the optical system. In laboratory tests, the method produced smoother changes in focus and sharper details, and objects hidden from one viewpoint became visible from another.
The second work, published in ACM Transactions on Graphics, is devoted to the interaction of light with materials. A mirror reflects its surroundings, glass refracts passing light, and a glossy surface scatters it differently in different directions. The HoloPathTracer method combines path tracing - a graphics technique in which virtual light rays are traced through the scene - with wave optics. The calculation preserves not only brightness but also the phase needed for a hologram. "HoloPathTracer additionally models how light physically interacts with the geometry and materials of the scene," noted Wenbin Zhou, also one of the first authors.
The resulting holograms convey reflections, refractions, shifting glare and indirect lighting together with natural changes in focus and viewpoint. To speed up the work, the team reuses light paths when creating a series of holograms and caches parts of the scene's lighting.
Both methods address different parts of one task: the mesh-based approach provides a calibrated path from conventional three-dimensional models to holograms, while HoloPathTracer extends their capabilities through more complex interaction of light and materials. For now, these are research prototypes. Mass-market real-time devices will require faster computing and improvements to spatial light modulators - the devices that form light in a holographic display - and the surrounding optics. "We believe that this direction will help make holographic experiences more natural and visually realistic for future VR and AR systems," said Peng.