HESAI color lidar

Press release (April 2026): https://www.hesaitech.com/hesai-unveils-picasso-6d-full-color-spad-soc-next-gen-etx-and-innovations-in-spatial-intelligence-and-physical-ai/

Hesai Unveils Picasso 6D Full-Color SPAD-SoC, Next-Gen ETX, and Innovations in Spatial Intelligence and Physical AI

Hesai Technology (Nasdaq: HSAI; HKEX: 2525), a global leader in intelligent technology and 3D perception, hosted its 2026 Technology Open Day on April 17, showcasing its latest generation of lidar technologies spanning full-color perception, ultra-high-resolution imaging, and spatial intelligence applications.

A key highlight was Picasso, the world’s first 6D full-color ultra-sensitive lidar SPAD-SoC, enabling native fusion of spatial and color perception at the ASIC level. Built on this platform, Hesai’s ETX lidar series has been upgraded to support up to 4320 channels and is expected to enter mass production in the second half of 2026. The company also previewed Kosmo, a new spatial intelligence AI hardware product, and outlined a new direction in robotic actuation modules for physical AI systems. 

At the core of today’s announcement is Picasso, Hesai’s fifth-generation 6D full-color ultra-sensitive lidar ASIC platform, designed to enable vehicles to see both geometry and color natively on a single chip.

Unlike conventional lidar systems that capture only geometric information such as XYZ, Picasso integrates RGB sensing and time-of-flight (ToF) ranging at the chip-level, enabling the direct generation of colorized point clouds without the need for complex sensor fusion. This 6D full-color capability delivers precise spatial and temporal alignment, resulting in richer environmental understanding and significantly enhancing the perception capabilities of next-generation autonomous systems. 

With photon detection efficiency (PDE) exceeding 40%, Picasso delivers industry-leading sensitivity, enabling stronger long-range performance, better detection of small objects, and improved performance in low-light environments within the same power envelope.

Rather than simply combining camera imagery with lidar point clouds, Picasso enables true ASIC-level fusion, with each point inherently carrying color information. This allows autonomous systems to move from spatial perception to real-time understanding of the driving environment, improving performance in different scenarios such as traffic signals, lane markings, and construction zones. 

The introduction of Picasso marks a major step in lidar development, moving beyond incremental improvements in resolution toward a new generation of full-color perception. This breakthrough is supported by Hesai’s long-term vertical integration strategy and in-house development across core lidar technologies. Following the launch of Fermi C500, a RISC-V-based high-performance lidar controller SoC that integrates MCU, FPGA functions, and ADC into a single chip, the company has established full-stack in-house ASIC development capabilities. This technology foundation has already translated into large-scale commercialization, with Hesai’s upgraded ATX lidar achieving full-scale mass production and securing over 6 million units in orders from leading OEMs.

 

ETX is expected to enter mass production in the second half of 2026, with planned deployment across multiple flagship vehicle models, supporting the large-scale adoption of L3 autonomous driving systems. 

Sony TSMC partnership

Link: https://www.sony-semicon.com/en/news/2026/2026050801.html

Sony Semiconductor Solutions and TSMC Enter Preliminary Agreement for Next-Generation Image Sensor Strategic Partnership 

Atsugi, Japan — Sony Semiconductor Solutions Corporation (“Sony”) and Taiwan Semiconductor Manufacturing Company Limited (“TSMC”) (TWSE: 2330, NYSE: TSM) today announced the signing of a non-binding memorandum of understanding (“MOU”) to form a strategic partnership for the development and manufacturing of next-generation image sensors.

Under the proposed partnership, Sony and TSMC intend to establish a joint venture (“JV”), with Sony being the majority and controlling shareholder,to set up development and production lines in Sony’s newly constructed fab in Koshi City, Kumamoto Prefecture. Through the JV, both companies expect to leverage Sony's expertise in sensor design alongside TSMC’s strengths in process technology and manufacturing excellence as part of a broader collaboration aimed at enhancing image sensor performance.

With the MOU signed, Sony and TSMC are discussing potential investments by the JV. These investments, along with new capital investment by Sony in its existing plant in Nagasaki, are being considered on the premise that they would be implemented in phases based on market demand, and that they would receive support from the Japanese government.

This partnership also seeks to explore and address emerging opportunities in physical AI applications, such as automotive and robotics, paving the way for future innovations and expanded technological advancements. The establishment of the JV remains subject to the execution of a definitive legally binding agreement regarding the partnership and the satisfaction of customary closing conditions. 

“Building on the trust cultivated through our long-standing collaboration with TSMC, I am delighted that we have reached an agreement to advance our partnership to a new stage,” said Shinji Sashida, President and CEO of Sony Semiconductor Solutions Corporation.“This JV is a significant initiative that brings together the strengths of both companies and aims to drive further advancement in technology and business within the next-generation image sensor field. Building on this JV, Sony intends to further strengthen its business operations with a focus on creating high added value. Rooted in the Sony Spirit that has guided us since our founding, we intend to continue to take on the challenge of creating new markets through unprecedented ideas and distinctive technologies.”

“Sony has been our long-time partner in the CMOS image sensor business. We are excited to elevate our collaboration to the next level, which represents a key step forward in driving future sensing technology in the AI era,” said TSMC Senior Vice President and Deputy Co-COO, Dr. Kevin Zhang. “This partnership underscores our shared commitment and mutual vision of leveraging cutting-edge technologies and innovative solutions to deliver leading sensing technology and products. We look forward to working closely together to achieve impactful results and create lasting value for all stakeholders.” 

Soliciting volunteer blog maintainers

Image Sensors Blog is soliciting 1-2 volunteers to help me maintain our weekly posting schedule. We have a large readership spanning industry and academia. This is a great opportunity esp. for an early-career academic (graduate student/postdoc) to provide valuable service to the image sensors community. Please email me (see contact listed in the "Blog Authors" sidebar) with a short introduction and how you see yourself contributing to the blog.

Many thanks to Dave Gilblom who is now retiring from his role as one of the blog maintainers. Thanks, Dave, for the countless hours of hard work poring over image sensor documentation, and for spearheading the Job Postings and Conference List features on the blog! 

Harvest Imaging Forum 2026 Nov 5-6, Registrations Open

Advanced Packaging and Multi-Dimensional Systems Integration for Image Sensors

November 5 & 6, 2026
Antwerp, Belgium

Abstract
Packaging technology for image sensors is an underappreciated issue in the systems integration process, greatly affecting performance, scalability, physical area, and cost. The multi-dimensional package as well as the underlying technology can greatly affect the overall electrical, optical, mechanical, and thermal characteristics of the integrated system. These systems have become multi-dimensional in nature, composed of several interacting layers and multiple sensors with the imagers vertically integrated within the overall system. Now that image sensors are driven more than ever to extreme performance capabilities, novel packaging options and technologies have become increasingly important. An important trend to support large scale sensors integrated into systems are the widely applied stacking and chiplet technologies in which several layers of active silicon are placed above each other. In this way, highly compact, high performance imaging systems are becoming possible. Packaging of systems including image sensors has become much more than placing a single silicon die into a package and attaching some wires.

The landscape of image sensor applications is widening at an accelerating rate, requiring new and different packaging solutions. This forum will review this key issue as well as include a series of highly related topics. This year’s Harvest Imaging Forum is focusing on the existing field of multi-dimensional packaging technology for image sensors. Related topics such as on-chip interconnect, power delivery, synchronization, network communication, thermal behavior, and heterogeneous integration will also be discussed. The focus will be on both current and near term capabilities. These next generation technologies are now becoming available, allowing the development and integration of large scale systems composed of multiple layers and chiplets to produce large scale systems operating at increasingly higher levels of performance. 

https://harvestimaging.com/forum_registration_2026.php 

ST low-power global shutter CIS

[ST Press Release from April 2026.]

STMicroelectronics brings always-on vision to next-generation personal electronics with new ultralow-power image sensors

VD55G4 and VD65G4 extend the ST BrightSense portfolio with compact, ultralow-power, microcontroller-compatible sensors for wearables, AR/VR, and smart devices

These new global-shutter image sensors can consume up to 10x less, in operating mode at around 800×700 resolution and 10 frames per second

Geneva, Switzerland — April 28, 2026 — STMicroelectronics (NYSE: STM), a global semiconductor leader serving customers across the spectrum of electronics applications, introduces a new generation of ultralow-power global-shutter image sensors that deliver high-quality, always-on vision to compact devices operating on batteries or harvested energy. The VD55G4 (monochrome) and VD65G4 (RGB color) sensors, part of the ST BrightSense portfolio, are now available to early adopters, enabling customers to start designing their next generation of smart, ultralow-power vision devices today.
Designed for the next wave of personal electronics and smart devices, the new sensors serve applications including wearables, AR/VR and XR headsets, smart home appliances and medical devices. They are engineered to deliver rich visual context and AI-ready data under tight constraints on power, size, and cost. The sensors combine an ultralow-power detect-and-wake architecture with a very small global-shutter optical format and interfaces optimized for low-power microcontrollers and cost-effective systems on chips (SoCs).

“Always‑on vision is becoming essential for the next generation of personal electronics, from smart glasses and AR/VR headsets to intelligent home appliances and medical devices. With VD55G4 and VD65G4, we are bringing this capability to smaller, lighter products that must run for a long time on a tiny battery. These new sensors help our customers create more intuitive and responsive experiences, extend battery life, and bring embedded vision and edge AI into everyday devices,” said Alexandre Balmefrezol, Executive Vice President and General Manager of the Imaging Sub-Group at STMicroelectronics.

From wearables and AR/VR to smart appliances
VD55G4 and VD65G4 bring always‑on vision to products that must stay small, light, and extremely power‑efficient. Building on the ST BrightSense family, they add a color option, faster response for interactive use cases, and simple connectivity to low‑power microcontrollers, making it easier to add vision to space‑ and cost‑constrained designs.

In wearables, the sensors enable all‑day, always‑aware features such as glance detection, presence sensing, and contextual alerts, while fitting into very compact designs and working directly with microcontroller‑based platforms.

For AR/VR and XR headsets, they combine low power and high‑quality capture to support accurate tracking and spatial awareness, helping extend battery life without compromising comfort.

In smart home appliances, IoT devices, and medical products, the sensors allow more intelligence to run locally on the device itself, reducing cloud dependence and standby power. Their tiny size and energy efficiency also make them well suited to solar‑ or energy‑harvesting‑powered vision nodes.

Ultralow‑power design that consumes up to 10x less
Thanks to an optimized sensor architecture and dedicated always‑on mode, VD55G4 and VD65G4 can consume up to 10 times less power than conventional global‑shutter sensors. They can watch for changes in a scene and wake up the main processor only when needed, shifting from continuous streaming to event‑driven operation. This enables all‑day, always‑on experiences, longer battery life, and practical vision systems powered by small batteries or energy harvesting.

Their very small footprint and integrated image processing simplify design and reduce system cost, while supporting responsive, AI‑ready vision features in a wide range of edge devices. 

Ex-Samsung CTO founds image sensor startup

Link: https://it.chosun.com/news/articleView.html?idxno=2023092160385 

"Lee Je-seok, Vice President and Chief Technology Officer (CTO) of the Sensor Business Team, also resigned recently after 32 years at the company. Lee is considered a "living witness" of Samsung’s image sensor brand, ISOCELL, and is credited with leading the development of the world's first 100-megapixel and 200-megapixel sensors, as well as securing a spot in Apple’s supply chain.

Following his resignation, Lee is reportedly preparing to launch an image sensor startup. On social media, he posted an image signaling a "stealth mode" startup with global experts, stating he would "redefine the future of imaging in January 2027." Industry observers are closely watching as a key designer of Samsung's proprietary technology could emerge as a potential competitor."

Conference List - December 2026

18th International Conference on Sensing Technology (ICST2026) - 7-9 December 2026 - Sydney, Australia - Website

International Technical Exhibition on Image Technology and Equipment (ITE) - 2-4 December 2026 - Yokohama, Japan - Website

IEEE International Electron Devices Meeting - 12-16 December 2026 - San Francisco, CA, USA - Website

If you know about additional local conferences, please add them as comments.

Return to Conference List index

Canon MS-510 low light camera

Press release: https://www.usa.canon.com/newsroom/2026/20260415-camera

Canon U.S.A. Sets New Standard in Low-Light Imaging with the Launch of the MS-510 Multi-Purpose Camera at NAB 2026

Featuring a new Canon 1-inch SPAD sensor, the MS-510 delivers impressive low-light and long-range performance for ultra-low-light environments 

MELVILLE, N.Y., April 15, 2026 – Canon U.S.A., Inc., a leader in digital imaging solutions, today announced the release of the MS-510, an innovative multi-purpose camera engineered to meet the rigorous demands of ultra-low-light full-color video shooting. Building on the success of the MS-500, the new MS-510 is designed to deliver high-fidelity captures of everything from nocturnal wildlife and natural nightscapes to high-security areas such as seaports, public infrastructure facilities, and national borders where "seeing the unseen" is a mission-critical requirement.

In response to increasing demands for excellent performance in near-total darkness, the MS-510 features a new Canon 1-inch Single-Photon Avalanche Diode (SPAD) sensor with the world’s highest number of pixels at approximately 3.2 million with improved Near-Infrared (NIR) sensitivity.

By leveraging this technology, the camera achieves a minimum subject illumination of 0.0006 lux, a significant improvement over the 0.001 lux of its predecessor. This allows for bright, full-color imaging in ultra-low-light environments where traditional cameras can fail. Furthermore, the MS-510 boasts improved NIR capabilities, providing a tactical advantage for night-mode operations and long-distance detection.

The MS-510 is not just a camera; it is a sophisticated tool for critical infrastructure monitoring and law enforcement. Key features include:

•  Broadcast-Grade Optics: Equipped with the broadcast industry standard B4 mount, the MS-510 has a built in magnifying optical system offering compatibility with Canon’s extensive lineup of 2/3-inch ultra-telephoto broadcast zoom lenses, enabling crystal-clear identification over vast distances.
•  Advanced Image Processing: Integrated Haze Compensation automatically reduces the interference of mist and haze while adjusting contrast, and Smart Shade Control corrects for highlights and shadows to maintain image integrity in challenging lighting conditions.
•  Customizable Imaging Profiles: Users can create up to 20 customized image quality settings or utilize Custom Picture Presets to prioritize wide dynamic range or maximum noise reduction depending on the environment.

Designed to fit into existing security ecosystems, the MS-510 shares common protocols with the MS-500, including NU, Pelco-D®, ONVIF®, and WebView. It offers versatile output options, including 3G/HD-SDI for high-quality video feeds and a LAN terminal for IP-based camera control and distribution.

Miscellaneous videos: Samsung vs. Sony; Eric Fossum interview; Eyeo webinar; Gpixel

Why Samsung Can't Beat Sony's Image Sensors:

Interview with Eric Fossum:

Eyeo webinar:

 China’s Newest Tech Billionaire Made His Fortune From Developing Image Sensor Chips For Robotics: 

 

Foveated imaging with optical folding

Jinwen Wei and Liangcai Cao, "Compact Neural Pancake Camera for High-Perceptual-Quality Foveated Imaging," ACS Photonics (2026).

Link: https://doi.org/10.1021/acsphotonics.6c00691 

Pancake catadioptric optics utilize optical folding to effectively reduce the optical-path thickness in virtual-reality display systems. However, limited optical throughput and optical degradation make image reconstruction for pancake cameras severely ill-posed, hindering broader imaging applications. In this article, we propose a neural pancake camera with adaptive-prior deconvolution, achieving compact, high-perceptual-quality imaging. By introducing latent-space projection, adaptive prior deconvolution alleviates the trade-off between pixel fidelity and perceptual quality and addresses the excessive smoothing inherent in conventional pixelwise optimization. The proposed neural pancake camera reduces the ratio of axial length to physical aperture diameter by 3.2 times compared with other flat cameras with high imaging quality. Experiments and ablation studies substantiate that the proposed adaptive prior deconvolution improves perceptual quality by 70%, as measured by CLIP-IQA, while also outperforming the state-of-the-art deep learning models on pixel-level fidelity. As a representative application of the proposed neural Pancake camera, this work further showcases bioinspired foveated imaging, highlighting its potential for bandwidth-efficient imaging in next-generation edge and portable devices. 

 

Figure 1. Pipeline of the proposed neural Pancake camera. (a) Conceptual illustration of the proposed compact Pancake camera scheme. (b) The inherent trade-off between perceptual quality and pixel-level quality of computational imaging. This work proposes adaptive prior deconvolution to promote the perceptual imaging quality of Pancake cameras while improving pixel fidelity. (c) Overview of the training process of the end-to-end adaptive prior deconvolution. The framework operates by optimizing the learnable deconvolution network while leveraging a latent natural-image manifold prior, anchoring the restoration output to the natural image manifold to reconcile the perceptual-pixel trade-off.

 

Figure 3. Quantitative evaluation and visualization of the adaptive-prior deconvolution. (a) Schematic illustration of the inference process based on the adaptive-manifold prior. (b) Quantitative performance profiles of averaged MUSIQ, SSIM, and PP-IQA across inference steps. (c) Visual comparisons demonstrating the effects of the low prior, the adaptively selected proper prior, and the over-prior.

 

 

Figure 4. Demonstration of neural Pancake camera-based foveated imaging. (a) Schematic illustration of foveated imaging in the human visual system. (b) Profiles of high-frequency content proportion and relative acuity versus field of view. (c) Comparison of reconstructed and raw-captured images across different fields of view.