Fully detailed APCS (active pixel) sensor spec sheet!

[SAR_admin]

 

 

We would like to hear the expert voice on that brand new and world's first Sony sensor! Download the full file at Wetransfer.

 

[Igor]

Hi everyone,

 

I think that the new sensor will not benefit in terms of low light performance. Indeed, it can not capture more light per unit of time since it uses the same RGB filter which rejects 2/3 of light.

 

As to the higher resolution, they can make Bayern sensors with crasy resolution (e.g. 20 Mpix 1/2.6''), so there is no need for that. 

 

No need for interpolation - no moire, nice indeed.

 

If they can ensure its durability (fast moving part ot top of the photodiode array).

 

The minimum total exposure time with that model can not be less than (2x1/192 + 3x1/16) = ca. 1/5 ms, or 1/5000 s. Correspondingly about 5000 fps, as far as the bus/processor/recorder allows. And it could be improved in the future.    

[rr98]

I think that the new sensor will not benefit in terms of low light performance. Indeed, it can not capture more light per unit of time since it uses the same RGB filter which rejects 2/3 of light.

 

I think you are correct when compare the pixel with the same size from APCS sensor and Bayer sensor.  However, base on what I understand, since APCS sensor only need 1 pixel to capture 3 color channels, in theory, it saved the spaces for 3 additional pixels in order to accomplish this on Bayer sensor which allow doubling the pixel size.  

[Igor]

I think you are correct when compare the pixel with the same size from APCS sensor and Bayer sensor.  However, base on what I understand, since APCS sensor only need 1 pixel to capture 3 color channels, in theory, it saved the spaces for 3 additional pixels in order to accomplish this on Bayer sensor which allow doubling the pixel size.  

Thinking generally, with given sensor size and the same RGB filter the amount of light consumed with all pixels (or all Bayer subpixels) will be the same. You can only distribute the light in different ways (spatially in Bayer or temporally in APCS).

 

One big APCS pixel is equivalent to three smaller Bayer subpixels. The big pixel is exposed to each color for 1/3 of the total exposure time whereas the smaller (1/3 by area) subpixels are exposed to their colors for full time. The number of formed electrons will be the same. I do not take into account the useless area of gaps which will be smaller for APCS. 

 

If you make an APCS sensor with small pixels, these small pixels will be exposed to each color for 1/3 of the exposure time. This will result in lower SNR. After downsizing to the above pixel count, though, the SNR will become the same as above.