You are basically correct. The main factor affecting noise when you use high ISO settings is the size of each pixel: the smaller it is, the fewer photons it collects during an exposure. Random variation may result in adjacent pixels collecting (say) 100 and 102 photons during an exposure, which will give a very similar result. Much smaller pixels may result, with same natural random variation, in adjacent pixels collecting 8 and 10 photons - i.e. a big difference between parts of the picture that are supposed to be the same. Hence, noise.



So you will get much less noise from a sensor which is (i) bigger, and (ii) subdivided into fewer pixels, so that each pixel is as big as reasonably possible. The sensor in a DSLR can easily be 10 times the size of one in a compact camera, so straight away the noise will be less. For a sensor of a given size, larger pixels (i.e. fewer megapixels on the sensor) will give less noise. So a DSLR like Nikon D40 (big sensor, split into 6MP) will give much better performance at high ISO than a 12MP compact (where the sensor is not only small to start with, but subdivided into an even larger number of pixels)..

The differences between different types of sensor (CMOS, CCD etc) are much less significant than this. poeple buying compact cameras are being tricked by manufacturers into believing that more megapixels means better quality. It doesn't - it means more noise. A 10MP sensor the size of a small fingernail in a compact camera will produce a completely cr*p image at high ISO, whereas the much larger 6MP sensor in a DSLR will produce sharp A3-size even at high ISO settings..

Best wishesMike..

Image sensors of compact cameras have a very low SNR after real ISO200. One exception seems to be Fuji Super CCD that have the same problem but the threshold is ISO400 real (so at least 1 stop better) even if the sensor is not 4 times larger than the others (I mean 1/1.8" sensors while Fuji has 1/1.7" or 1/1.6"). It seems the layout of the sensor permits to the individual pixels to be larger than other sensors..

As a conclusion the main problem with the size of the sensors is the small amount of light that each individual photosite (or pixel) receives..

That's why dSLR, which have sensors about 14 times larger than compact cameras, perform better at high ISO speeds. Also people have seen the drop in high ISO performance when the dSLR went from 6 MP to 10 MP although the progress in the noise reduction reduced the losses..

That's why you never have to look at the high numbers that different manufacturer put as the maximum ISO speed. First of all the numbers are fake (you don't get the same parameters of shutter speed and aperture at a fixed ISO speed) and the results are unusable..

I have or had several cameras from different manufacturers and no ISO speed was equal to the other (taking the same scene with different cameras and fixing the ISO speed and the aperture I got different shutter speeds with a perfect exposed result from each camera)..

As an example (fixed parameters are ISO speed 100 and f/5.6) the same scene was perceived by different cameras with the following shutter speeds:Sony P100 1/500 sCanon A610 1/1000 sCanon S3 1/640 sFujifilm S6500 1/550 s.

Being a bright sunny day and a front lighting of the subject I would be inclined that Canon A610 was closer to the reality. All photos that resulted were properly exposed..

As a conclusion never try to use more than ISO400 on most compacts (ISO800 on Fuji with SuperCCD) or you will be forced to print postcard size (4x6" or 10x15 cm) to get a decent picture..

OIS helps keeping the ISO speed low on appropriate scenes (scenes with very slow or no motion in them).VictorBucuresti, Romaniahttp://s106.photobucket.com/albums/m268/victor_petcu/..

Baloo_buc wrote:.

As an example (fixed parameters are ISO speed 100 and f/5.6) the samescene was perceived by different cameras with the following shutterspeeds:Sony P100 1/500 sCanon A610 1/1000 sCanon S3 1/640 sFujifilm S6500 1/550 sBeing a bright sunny day and a front lighting of the subject I wouldbe inclined that Canon A610 was closer to the reality..

There's a really neat rule of thumb for estimating exposure outdoors on a bright sunny day. At f/16, the correct shutter speed is the numerical inverse of the ISO speed. So using ISO 100, the shutter speed will be 1/100, at ISO 200 you'll need 1/200, and so on. Because it works at f/16 in full sun, it's called the 'Sunny 16' rule..

To convert the exposures for other apertures, count the stops. If you gain a stop somewhere, you have to lose it somewhere else. In your example, the aperture is f/5.6, which is three stops brighter than f/16. (Count 'em - 5.6 -> 8 -> 11 -> 16). So at ISO 100 the likely shutter speed is 1/800 (count the stops again - 1/100 -> 1/200 -> 1/400 -> 1/800). Which is right in the middle of the exposures measured by your four cameras!..

That's why I said the Canon A610 seems to be accurate..

I knew a slightly different Sunny f/16 rule (ISO100, f/16, 1/125 s). On the Internet that version seems to be the base for the EV calculations.VictorBucuresti, Romaniahttp://s106.photobucket.com/albums/m268/victor_petcu/..

What Mike said!.

Mike703 wrote:.

You are basically correct. The main factor affecting noise when youuse high ISO settings is the size of each pixel: the smaller it is,the fewer photons it collects during an exposure. Random variationmay result in adjacent pixels collecting (say) 100 and 102 photonsduring an exposure, which will give a very similar result. Muchsmaller pixels may result, with same natural random variation, inadjacent pixels collecting 8 and 10 photons - i.e. a big differencebetween parts of the picture that are supposed to be the same.Hence, noise. This is a bit of a simplification but it makes thepoint (I hope)..

I would add the following. With good illumination, the mean exposure is 100 photons per photosite. The noise is +/- 1 photon. But at low illumination, the mean exposure is 10 photons per photosite...and the noise is still +/- 1 photon. In really dim light, the mean exposure is 1 photon per photosite and the noise is still +/- 1 photon. What the camera does when you bump the ISO is amplify the signal and the noise.

Example 2 needs 10X amplification: 100 +/- 10 (10% noise). Example 3 needs 100X amplification: 100 +/- 100 (100% noise)!.

So you will get much less noise from a sensor which is (i) bigger,and (ii) subdivided into fewer pixels, so that each pixel is as bigas reasonably possible. The sensor in a DSLR can easily be 10 timesthe size of one in a compact camera, so straight away the noise willbe less. For a sensor of a given size, larger pixels (i.e. fewermegapixels on the sensor) will give less noise. So a DSLR like NikonD40 (big sensor, split into 6MP) will give much better performance athigh ISO than a 12MP compact (where the sensor is not only small tostart with, but subdivided into an even larger number of pixels)..

This clear distinction is becoming blurred (pun intended). Camera manufacturers are increasing the in-camera noise reduction (NR). This does reduce the noise, but also smears the detail, so that pictures start looking like "water color paintings". They try to differentiate between areas that can stand heavy NR (like the sky and dark shadow areas), but increasingly, they fail. Most pix will be sorta OK, but then you'll get one pic that has blurry grass (as an example)..

Mike's advice is exactly correct. I'd say, "Buy the biggest sensor you can afford and the fewest photosites you can put up with!".

The differences between different types of sensor (CMOS, CCD etc) aremuch less significant than this. poeple buying compact cameras arebeing tricked by manufacturers into believing that more megapixelsmeans better quality. It doesn't - it means more noise. A 10MPsensor the size of a small fingernail in a compact camera willproduce a completely cr*p image at high ISO, whereas the much larger6MP sensor in a DSLR will produce sharp A3-size even at high ISOsettings..

Go read this guy's explanation of IQ and how to get it:.

Http://www.here-ugo.com/BridgeBlog/?page_id=11.

Charlie DavisNikon 5700 & Sony R1HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/..

Baloo_buc wrote:.

I knew a slightly different Sunny f/16 rule (ISO100, f/16, 1/125 s)..

That version uses 'whole stop' shutter speeds - 1, 1/2 ... 1/30, 1/60, 1/125, 1/250 etc. But it doesn't really matter because it's all very approximate, and the brightness of 'full sun' obviously varies somewhat..

On the Internet that version seems to be the base for the EVcalculations..

EV numbers are based on whole stop shutter speeds, that's correct. But do remember that the real world shooting conditions that are often shown in conjunction with EV charts are not part of the EV number system, they are just observations to convey to the reader how bright EV-such-and-such is in practice...

Thats the price you pay for a small sensor camera..

A DSLR is far more serious..

Good software is also part of the equation..

Just because they go upto 3200 etc doesn't mean it's usable..

Its decision time now - stay under 400 or consider a bigger sensor camera..

Http://www.robert-harrington.com/articles/sensorsize.htm.

There is much more into on the science of sensors - it may be enlightening, particularly when read next to the advertising hype..

Peter.

Persuasive Marketing Systems -inc Copywriting, Design & Photography..