Back to Basics of Sensor: Can anyone throw some light?

[achanth]

Can anyone point to where can I understand more about sensor sizes and the quality of image? I know that Panasonic cameras has 1/1.7" sensor, but the camera suffers from picture noise..whereas other cameras (Olympus, Canon, etc) has 1/2.5", yet the image quality is some much better (i.e. less noise). This may be due to processing of the image as it converts picture into Jpeg form. I refer to the non-DSLR cameras. Why is this so?

 

[westwest2]

what lens these camera use play a part too...

 

[ykia]

There's a good write up in dpreview here...

http://www.dpreview.com/learn/?/Glossary/Camera_System/sensor_sizes_01.htm

 

[achanth]

Thank you both for your helful comments.

 

[yanyewkay]

All similiar based sensors will behave more or less the same, the only difference is with the in-camera processing (ie:built in software) that deals with the noise and colour problems.
In summary,
- all CCD based cameras will suffer badly from noise and it is very hard to push the ISO up without visible noise. However, they are have good reponse to light and is very easy to work with.
- SuperCCD made a variation to the CCD sensor by rearranging the sensor pattern is is able to achieve very high ISO (ISO6400 for a P&S) at significantly lesser noise
- CMOS based have lower power consumption and therefore are less hot and also result in a sensor that produces lesser noise. However, in general, they produce lower contrast pictures (think web cam) but Canon has modified the receptors in a way to overcome this problem and is the (only?) exception to a good CMOS sensor.
- Forveon (not sure how it's spelt) gets rid of the bayer pattern fliter and each light receptor is able to collect light for all 3 primary colours (RGB). Being so, each receptor can now be made bigger and more details can be recorded. Whereas in all other sensors with bayer patterns, usually 4 pixels (RGGB) are used to create 1 pixel on the output photo.

Also, generally speaking, bigger sensor size means bigger light receptor sites which will give lower noise and better dynamic range but with the pixel race going on, bigger sensors are now occupied by MORE pixel sites.. which means the size of each receptor is will also be smaller.

 

[lsisaxon]

Can anyone point to where can I understand more about sensor sizes and the quality of image? I know that Panasonic cameras has 1/1.7" sensor, but the camera suffers from picture noise..whereas other cameras (Olympus, Canon, etc) has 1/2.5", yet the image quality is some much better (i.e. less noise). This may be due to processing of the image as it converts picture into Jpeg form. I refer to the non-DSLR cameras. Why is this so?

Are you comparing same pixel count and whether they are shot at the same light condition and ISO also?

 

[lsisaxon]

Whereas in all other sensors with bayer patterns, usually 4 pixels (RGGB) are used to create 1 pixel on the output photo.

This is not true of the Bayer sensors. There is an algorithm to extract 10MP of RGB from 5mp of G and 2.5mp of R and 2.5mp of B. So in Bayer sensor, the luminance resolution is higher than the colour resolution.

 

[yanyewkay]

This is not true of the Bayer sensors. There is an algorithm to extract 10MP of RGB from 5mp of G and 2.5mp of R and 2.5mp of B. So in Bayer sensor, the luminance resolution is higher than the colour resolution.

yah.. that's correct.. 4 photosite (arranged in a 2x2 matrix) create the 1st pixel on the photo.. then 2 photosites are added and 2 discarded to make another grouop 4 photosite and create another pixel on the picture so we have reused info from 2 photosite and from the 2nd pixel on, it only takes 2 new photosite to create 1 pixel. that's only going in 1 axis direction, when we move down the other axis another photosite step, 1 row of the photosite info are reused again...