Researchers have developed a new optical screen that converts invisible infrared light into visible light, while making the converted images more than 1000 times brighter – offering a potential alternative to the expensive detectors used in today’s infrared cameras.

New research led by the ARC Centre of Excellence for Transformative Meta-Optical Systems (TMOS) at the University of Melbourne demonstrates a new way to make invisible infrared light visible without relying on the expensive detector technology used in today’s infrared cameras.

Infrared is a type of light our eyes can’t see, while carrying information about heat and objects in the environment. It’s widely used for environmental monitoring, industrial inspection, medicine and security.

However, infrared cameras remain expensive because they rely on specialised detectors that are costly to manufacture and often require cooling.

The international team of researchers demonstrated a new way to see infrared light that could one day make these systems smaller, lighter and far less expensive.

Lead author Dr Nima Sefidmooye Azar completed the work while at TMOS at the University of Melbourne. Now based at the University of Queensland, he says the team developed an optical screen – a ‘metasurface’ – covered in tiny structures thousands of times smaller than the width of a human hair.

The screen converts invisible infrared light directly into visible light while making the converted image more than 1000 times brighter. It works by controlling how light behaves, concentrating invisible infrared light where it is needed to make the image brighter.

“Infrared imaging has enormous potential, but today’s cameras generally depend on expensive semiconductor detectors,” says Dr Sefidmooye Azar.

“Our platform offers a completely different approach by converting infrared images into visible ones using a compact optical screen,” he says.

How they did it

The researchers coated the patterned surface with tiny particles made of rare-earth materials. These particles absorb infrared light and re-emit it as visible light.

The metasurface was engineered to trap and concentrate incoming infrared light, dramatically strengthening the interaction between light and the nanoparticles.

The approach made the converted images more than 1000 times brighter than nanoparticles could achieve on their own.

Coauthor Professor Kenneth Crozier, from TMOS at the University of Melbourne, says the biggest challenge was that making the image brighter often made it blurrier.

Unlike previous approaches, the new design makes the image brighter while keeping it sharp – two goals that have traditionally been difficult to achieve at the same time.

“The amount of brightening would change depending on the angle the light arrived from, causing fine details to disappear,” Prof. Crozier says.

The team overcame this by designing a new type of metasurface – a ‘flat-band’ dielectric metasurface – that works consistently even when light arrives from different angles.

“We designed the surface so it makes infrared images much brighter without making them blurry,” says Prof. Crozier.

The final picture

The team successfully demonstrated high-resolution infrared-to-visible imaging, producing bright, high-contrast visible images from infrared patterns while retaining fine spatial detail.

Dr Sefidmooye Azar says although the current prototype requires active infrared illumination, continued advances in nanoparticle materials and metasurface engineering could dramatically reduce the amount of infrared light needed, bringing practical applications closer.

Because the device is compatible with scalable manufacturing methods such as nanoimprint lithography, he says it could eventually enable large-area, low-cost infrared imaging devices.

“Our platform combines strong upconversion enhancement, angular robustness and polarisation independence in a single compact device,” says Dr Sefidmooye Azar.

“The approach offers a pathway toward lightweight and detector-free infrared imaging technologies.”

While more development is needed before the technology reaches commercial products, the researchers believe it could eventually be applied broadly.

“Potential applications include night vision, remote sensing, biomedical imaging and compact infrared cameras compatible with standard visible imaging technologies,” says Dr Sefidmooye Azar.

This research was published in the journal Light: Science and Applications. It was supported by the Defense Advanced Research Projects Agency (DARPA, USA), the Australian Research Council, the U.S. Department of Energy, and the National Science Foundation (USA).

Enhanced infrared-to-visible upconversion imaging via metasurface–lanthanide nanoparticle hybrid screens

Nima Sefidmooye Azar, Matthew Parry et al.

Light: Science & Applications, 2026, doi.org/10.1038/s41377-026-02449-5

ABSTRACT: Accessing the rich information carried by infrared light typically relies on bulky, complex optoelectronic systems. Lanthanide-based upconverting nanoparticles (UCNPs) offer a compelling alternative by converting infrared light into visible photons through nonlinear anti-Stokes processes. However, achieving strong upconversion under the low excitation intensities relevant to infrared vision remains challenging, motivating strategies to enhance light–matter interaction. Here, we demonstrate enhanced infrared-to-visible upconversion imaging enabled by integrating alloyed Yb/Er UCNPs with a resonant dielectric metasurface. The metasurface supports an optical resonance aligned with the UCNP excitation band, leading to over three orders of magnitude enhancement in upconversion emission. Crucially, flat-band angular dispersion of this resonance enables uniform enhancement across incident angles relevant to imaging, thereby preserving spatial frequency content and yielding sharp, high-contrast images. In light of ongoing advances in lanthanide-based materials, this metasurface–UCNP hybrid screen provides a promising platform for compact, detector-free, and scalable infrared imaging technologies based on optical upconversion.