Although computational photography is the hottest camera term right now, next year might bring some intriguing new sensor technology, with Samsung and Canon both unveiling new processors that are “human eye-like” and capable of capturing color photographs in the dark.
Samsung’s new sensor, which was reportedly co-developed with Chinese firm Tecno, is focused on smartphones, as you might think. The Samsung ISOCELL GWB sensor, according to TechTimes, boasts a 64MP resolution and, more crucially, has an uncommon RGBW (red, green, blue, and white) color filter array, which Samsung claims is the closest to human eyesight yet.
What exactly does this imply? While most camera sensors recreate colors in your images using RGB (red, green, blue) sub-pixels, the addition of white sub-pixels appears to increase color accuracy and low-light performance. This is because the ‘white’ sub-pixels (which are actually clear) are supposed to take in all forms of light, not just those in the RGB spectrum, in order to increase the sensor’s overall sensitivity.
Samsung stated in a webinar on this new sensor that its combination of enhanced sensitivity, color accuracy, and 64MP resolution will allow it to capture images that are closer to what the human eye perceives, according to TechTimes.
Of course, this is a fairly subjective claim, and Samsung isn’t the only company to experiment with RGBW filter arrays. We were thrilled with the results of the Huawei P8’s 13MP RGBW sensor (produced by Sony) when it was released in 2015. The Chinese manufacturer just released an RYYB sensor (red, yellow, yellow, blue) for the Huawei P30 Pro.
Unfortunately, the new ISOCELL GWB sensor from Samsung is unlikely to appear in the next Samsung Galaxy S22 series. According to the newest reports from @FrontTron, the S22 and S22 Plus models will have 50MP primary cameras, ruling out this sensor’s technology for Samsung’s future flagship phones at the very least.
Canon’s new night-vision sensor, which Nikkei reports will be mass-produced in 2022, is perhaps more intriguing and creative.
The sensor’s SPAD (Single Photon Avalanche Diode) technology allows it to capture high-quality color pictures in the dark, making it perfect for security cameras, augmented reality (AR), and self-driving automobiles. Because the sensor has a global shutter (which scans the entire sensor at once rather than line-by-line), it might also revolutionize high-speed video capturing.
CMOS designs are used in most camera sensors today (including Samsung’s latest one), while SPAD sensors function in a completely different way. According to Canon, CMOS sensor pixels only monitor the quantity of light that hits each pixel in a specific length of time before converting it to electrons.
SPAD sensors, on the other hand, detect every light particle (or photon) that enters the pixel and convert it into an avalanche of electrons. This implies the sensor is more sensitive, collects less electrical noise, and records more precise data per photon.
This isn’t, however, Canon’s most recent breakthrough. Although SPAD technology has been known for a long time, it has never been feasible to make a sensor with such a high megapixel count – until now. Canon initially created a 1MP SPAD sensor in June 2020, and now, according to Nikkei, it has a 3.2MP version ready for mass production next year.
So, how does this affect cameras? From 2023 onwards, the technology is projected to be employed in security cameras (to replace infrared night vision). Canon also wants to employ their SPAD sensor in conjunction with time-of-flight (ToF) technology to estimate three-dimensional distances, making it suitable for autonomous automobiles. However, during the next few years, SPAD sensors may be found in professional, ultra-high-speed video cameras.
It’s a larger problem for next-generation cameras than it is for phones
The near-simultaneous announcements of new sensors from Samsung and Canon demonstrate that camera hardware is still evolving. On paper, though, there’s little doubt that Canon’s is the more innovative, if more distant, option.
The combination of software and silicon, rather than sensors, has fueled the phone camera revolution of the last few years, which has seen them transform into unparalleled point-and-shoot powerhouses. That will most certainly remain the case, despite Samsung’s promises about its new “human eye-like” ISOCELL GWB processor.
While certain hardware advances are still feasible, especially in areas like liquid lenses and metalenses, Samsung’s sensor is still built on the CMOS technology used in the majority of today’s cameras. Changing the color filter array on a phone camera can have an influence on its performance, especially in low light, as we’ve seen with Huawei phones. However, that is no longer as important as what is now achievable with computational photography’s multi-frame processing.
As we just observed with the Google Pixel 6’s AI camera trickery, the most interesting developments for most people, rather than skilled photographers, may lie less in notions like “accuracy” and “sharpness” and more in creative realms. It’s much more enjoyable to be able to remove undesired background items or replicate panning effects than it is to obtain minor color accuracy improvements.
However, fundamentally altering the way camera sensors function, as Canon’s new SPAD sensor does, might open new capabilities that go beyond computational photography, which relies on the signals provided by today’s CMOS sensors.
This is evident in the restrictions of ‘Night modes,’ however, the SPAD image sensor appears to be capable of capturing high-detail photos with a tenth of the light required by conventional sensors. Because of the combination of night vision technology with depth-sensing 3D cameras, it’ll be focused on security cameras and autonomous cars at first, but it also has fascinating professional photography applications.
The sensor’s global shutter allows it to shoot at a staggering 24,000 frames per second (at 1-bit output), opening up some fascinating new slo-mo possibilities. In addition, the technology can capture photos with a significantly wider dynamic range than traditional cameras. But, for the time being, Canon’s objectives appear to be more industrial, with security cameras and better ‘eyes’ for autonomous cars owing to greater real-time mapping skills looking to be the primary goals.