The sensor of Quadra CFA is 16MP. They operate in the following modes:
The 16MP sensor in Quadra CFA is not a standard Bayer mode. Instead, they are pseudo-Bayer patterns in which four similar pixels are together. In previews and snapshots with a resolution of 4 million pixels or less, the Quadra CFA sensor is configured in Ning Bin mode, in which four adjacent pixels are merged into one pixel. Therefore, 4MP is a standard Bayer model, which can be directly processed by camera ISP assembly line.
Nowadays, 4cell 1 chip has both normal high pixel and dark high sensitivity and low noise performance, and has become the main force of mobile phone cameras.
The smaller the mobile phone, the smaller the module and chip size. On the basis of a certain size, high pixel and large pixel have always been a contradiction for mobile phone cameras. However, the high-resolution image quality brought by high pixels and the high-sensitivity and low-noise image quality brought by large pixels are very important.
In order to have both, 4cell 1 chips came into being, also known as Tetra cell, QuadrBayer and Four cell. Based on the classic Bayer array, the chip arranges each color into a combination of four pixels, which successfully switches the camera between high pixels and large pixels.
The operation of synthesizing four small pixels into one large pixel is easy to understand, so how to realize the operation of restoring to a high pixel?
Remosaic technology magically restores high-resolution images. The process from the original 4cell 1 pixel arrangement to the ordinary Bayer structure is called Remosaic.
****Remosaic is usually divided into two ways: software and hardware. The software Remosaic calculates a certain weight ratio according to the distance as the signal value of the pixel through pixel exchange or the connection between the pixel and the surrounding related pixels. Usually, the software re-mosaic algorithm is integrated on the platform side.
Now there are some chips that transform the pixel structure through independent ISP signal processing. Each photosensitive unit can display and output data independently, and Bayer arrangement of normal hardware can be photographed without additional software interpolation.
Of course, there is a big difference in speed between the two. Hardware is much faster than software re-mosaic, and hardware re-mosaic can support full-size Bayer preview. However, whether the mobile phone uses full-size preview needs to consider other factors such as power consumption. The software Remosaic takes a long time to process, and at present it only takes full-size photos.
After re-mosaic, the resolution has been greatly improved for high pixel applications.
You think you only need a four-in-one chip &; Remosaic** can be "both civil and military"? **
Careful friends can find that there are many false colors such as sawtooth grid in the real shot on the left, which are mainly caused by the hardware sensitivity of the chip itself and Remosaic. In order to optimize this phenomenon, "crosstalk calibration" appeared.
Crosstalk calibration to reduce false color interference
Because the chip pixel position of 4cell 1 is different and the sensitivity of each pixel is different, it is usually necessary to introduce crosstalk calibration to eliminate this difference.
Crosstalk calibration tools usually divide the whole image into multiple ROI blocks, calculate the energy of each pixel channel and determine its compensation data, and then the chip uses these calibration data to make the energy distribution in the original uneven state more balanced.
For example, a bottle of good wine needs to be opened at the class reunion ten years later, but it needs to be distributed evenly to every friend when pouring. At this time, we can complement each other by pouring more wine and less wine, so that everyone can enjoy the same good wine.
Changes of energy before and after crosstalk
The reflection of crosstalk calibration compensation in the actual image is mainly to remove the color block interference caused by signal difference, such as grid and sawtooth, which is usually more obvious when shooting a single color block. For example, the following pictures are blue sky and yellow objects respectively, and the color block interference before and after crosstalk calibration will be more obvious.
4cell 1 chip will also involve PD compensation, bad point compensation and so on, and the whole process usually needs to compensate PD point first, then do crosstalk calibration, and then remove the mosaic to become a normal Bayer for application. For the software Remosaic, these steps are usually encapsulated in an algorithm library provided by the chip manufacturer. The mobile phone calls the algorithm library to realize 4cell 1 chip, which has both high resolution in normal state and high sensitivity in dark state, and then you can take beautiful photos.
The application of 4cell 1 is not limited to this, but also has related applications in high dynamics.
4cell 1 unique 3 HDR gives you a different high dynamic experience.
There are many ways to know HDR. 3-HDR is a unique 4-core HDR chip. Each color group is equipped with high, medium and low exposures to achieve different brightness, and finally an HDR picture is synthesized.
Different from ordinary HDR technology, 3 HDR real-time graded exposure technology actually allows the four basic pixels that make up 1 large pixel to collect different exposure information in real time, and synchronize these information once, and realize real-time graded exposure through hardware, which not only makes the front subject and background clear, but also makes the preview effect completely consistent with the actual imaging, and truly achieves what you see is what you get.
Finally, another wave of eggs to see the beautiful photos of Miss Sister before and after using the 3-HDR function! ! !
Reference article:/link? URL = _ Oyp _ xiux 8 og 43 toe wa _ mqhuincqmgnxqqxgkyrahxoth 7 GW 4 qo xtq 2 adhgrsdxs 7 lnxwgh 5 AVL M9 nuyce 2 va & amp； wd = & ampeqid = 84d 28 f 880002 18a 000000045 fcf 17 1f