"The LYTIA 901 features a Quad-Quad Bayer Coding layout, where 16 pixels of the same colour are grouped. This structure is said to allow the sensor to capture high-resolution detail while maintaining strong light sensitivity. During regular photography, the 16-pixel cluster behaves like a single larger pixel, which helps in low-light conditions. When zooming, the sensor uses a remosaicing process to restore a normal pixel arrangement and recover fine detail."
"Sony has integrated an AI-based remosaicing circuit directly inside the sensor. This is the first time such a system has been built into a mobile CMOS sensor, according to the company. The AI processing helps reconstruct difficult high-frequency details such as small patterns and text. This logic also helps enable rapid processing and supports up to 30 frames per second video capture at 4x zoom in 4K resolution."
"For dynamic range, the LYTIA 901 uses several HDR systems. Dual Conversion Gain HDR is claimed to read the sensor at different gain levels within a single frame, while Fine12bit ADC is said to increase the bit depth used during conversion. Sony also includes Hybrid Frame HDR, which blends short-exposure frames with DCG data. Together, these technologies achieve a dynamic range of more than 100dB, which helps retain detail in both bright highlights and deep shadows."
The LYTIA 901 is a 200-megapixel, 1/1.12-type smartphone image sensor with 0.7-micron pixels designed for next-generation flagship phones. It uses a Quad-Quad Bayer Coding layout that groups 16 same-colour pixels to boost light sensitivity and capture high-resolution detail, behaving as a larger pixel in low light and employing remosaicing to recover detail when zooming. An AI-based remosaicing circuit is integrated on-sensor to reconstruct fine high-frequency details and enable rapid processing, supporting up to 30 fps 4K video at 4x zoom. Multiple HDR methods including Dual Conversion Gain, Fine12bit ADC, and Hybrid Frame HDR deliver over 100dB dynamic range.
Read at Gadgets 360
Unable to calculate read time
Collection
[
|
...
]