IN Brief:
- Metalenz has acquired Israeli imaging developer PxE Imaging, adding Holocoder technology to its optical sensing portfolio.
- The technology replaces conventional colour filters and microlens arrays with a diffractive layer integrated above a CMOS image sensor.
- Metalenz claims 2.5 times greater light transmission than conventional Bayer-filter arrangements, with image reconstruction handled by a dedicated processing stage.
Metalenz has acquired Israeli imaging technology developer PxE Imaging, extending its optical sensing platform into visible-light imaging and depth measurement using semiconductor-integrated diffractive optics.
The acquisition, announced on 11 November 2026, adds PxE's Holocoder technology to Metalenz's existing work on metasurface optics and polarisation sensing. The financial terms were not disclosed, and the companies have not announced a production date for a device combining their technologies.
Holocoder replaces the conventional colour filter and microlens layers above a CMOS image sensor with a patterned optical structure. It uses the wave properties of incoming light to encode information relating to wavelength and depth, enabling colour, infrared, and spatial information to be derived from a single camera.
Conventional colour image sensors usually employ a Bayer filter array, in which neighbouring pixels receive different portions of the visible spectrum. Red, green, and blue filters selectively transmit light before an image processor reconstructs the colour values required for each image location.
Although this arrangement is established across consumer and industrial imaging, the filters absorb light outside their selected wavelength ranges. Metalenz states that conventional Bayer filters can absorb approximately 60% of incoming light, restricting the amount of optical energy available to the sensing pixels.
PxE's approach substitutes a transparent diffractive structure that redirects light according to its wavelength and wavefront characteristics. Rather than rejecting unwanted wavelengths through absorption, the structure creates an encoded pattern from which image information can subsequently be recovered.
According to Metalenz, eliminating the conventional colour filter allows approximately 2.5 times more light to reach the sensor. The company also claims improved signal-to-noise performance relative to traditional Bayer-based cameras, although the comparison depends on the measurement conditions and complete imaging architecture.
Increasing the light collected by a sensor can improve the available measurement signal under restricted illumination. The resulting image quality will also depend on detector noise, exposure time, optical losses, sensor calibration, and the processing applied to the captured information.
The patterned optical layer is only one part of the system. Metalenz describes a dedicated holographic image signal processor that interprets the encoded sensor output and produces colour, infrared, and depth information.
This distinction is important because transferring optical encoding into the hardware does not eliminate image processing. Instead, the architecture aims to reduce computationally demanding operations associated with reconstructing information from multiple independent sensors.
The image signal processor uses a deterministic reconstruction method. Its performance will depend on the optical design, the characteristics of the CMOS sensor, and the implementation of the reconstruction pipeline, including the memory and computing resources available within the finished product.
Depth information is particularly relevant to machine vision. Conventional imaging systems may use stereo cameras, time-of-flight sensors, or structured illumination to estimate the position of objects in three dimensions. These arrangements introduce additional optical or electronic components, as well as calibration requirements between separate sensing elements.
Holocoder instead encodes information associated with wavefront curvature within the same optical path used for conventional image capture. This provides a possible route to obtaining depth information without installing a separate active ranging module, although measurement accuracy and operating range must be established for the intended configuration.
Using a single image sensor can also reduce the alignment requirements associated with combining information from multiple cameras. Colour and depth are generated from the same optical capture, limiting the need to reconcile measurements taken from different viewpoints or at slightly different times.
The optical arrangement nevertheless introduces calibration requirements of its own. Manufacturing variation, changes in temperature, and differences between the patterned layer and the underlying pixels can influence the encoded image and the reconstruction process.
Metalenz already develops flat optical structures that can be integrated with semiconductor image sensors. Its PolarEyes platform uses polarisation information to distinguish physical characteristics of surfaces and materials, providing data beyond the brightness and colour available from conventional imaging.
The acquired technology adds another sensing method to that portfolio. In principle, colour, depth, and polarisation could contribute complementary information to robotic perception or industrial inspection systems, although Metalenz has not demonstrated a commercially available sensor integrating all these capabilities.
The company reports that more than 300 million metasurface optical components have been shipped for consumer applications. That manufacturing history establishes experience with semiconductor-scale optical production, but it does not mean that the Holocoder platform has reached equivalent production volumes.
Holocoder and Metalenz's existing commercial metasurface products currently remain separate technologies. Further engineering would be required to combine their functions in one optical architecture, including work on fabrication, calibration, electrical integration, and system-level performance.
Industrial machine vision applications may benefit from receiving additional information about object geometry and surface characteristics through compact hardware. The effectiveness of that information will still depend on the image processing and application algorithms used to recognise objects or guide equipment.
Metalenz identifies robotics, autonomous vehicles, security imaging, and other edge sensing applications as potential markets. Detailed sensor specifications, independent performance measurements, and evaluation arrangements have not been publicly disclosed; the company is providing technical information to prospective developers through direct enquiries.
The acquisition brings PxE's optical design and imaging expertise into Metalenz's existing manufacturing and development organisation. Commercial qualification will need to establish whether the combined resources can deliver the proposed light sensitivity and multimodal sensing capabilities within the size, power, reliability, and cost requirements of production systems.



