Sensors, megapixels, focal length and quality output

[GrumpyOne]

Buyers want the most megapixels with the biggest sensor, longest lens but at the cheapest cost. Naturally, something has to give.

Please correct me if I'm wrong here.

The way I understand the way these cameras work is that reducing the sensor size means that there is less detail available for each individual pixel that is shot, which is why DSLRs can take such good pictures, they have a huge sensor advantage.

But the bigger sensor also means that the optical components are necessarily bigger as well, which adds to the expense. So to reach equivalent focal length, you either have a big lens opening onto a big sensor or you have a smaller lens opening onto a smaller sensor.

A 1/2.5 sensor has an area of ~24.7 mm^2
A 1/1.8 sensor has an area of ~38.2 mm^2
A 23.7x15.7 sensor has an area of ~372 mm^2

Assuming two compact cameras had identical hardware and software, with the exception of the sensor size and appropriate sized lens, the 1:1.54 ratio means that the camera with the larger sensor can produce images of the same quality at a resolution that is 54% larger. So, where the one camera would produce barely noise-free images at 6mp at a given set of settings, the camera with the larger sensor could do the same quality with a 9mp image.

Assuming the above is correct, would that also mean that the key to "simulating" a larger sensor is just to reduce the resolution of the image in situations where quality could suffer as a result of the smaller sensor size? If so, at a ratio of 15:1, if a DSLR took an picture at 7mp, the compact with the 1/2.5 sensor could only get the same quality by taking a picture no larger than 800x600.

On target or off-base?

 

[0bli0]

unfortunately, no. the pixel density of the sensor is constant. the problems with inherent noise etc. of the sensor itself will not change. in fact having a smaller image will likely make this more apparent.

one of the biggest problems with small sensors is the effect that aperture has in terms of bokeh (or the out of focus areas) due to something called circle of confusion. a lens with a 2.8 aperture taking an image with a 50mm lens (at a distance of 2 metres) on a sensor that's 66% of the size of a piece of 35mm film is going to render relatively nice out of focus areas on an image printed at 4"x6" or on a 17" monitor. on a camera that has a sensor that is 25% of the size of a piece of 35mm film, you would need to print the same image (taken at the same focal length, distance, and aperture) at around 16.75"x 25.1", and you'd need a ginormous monitor.

for me, with a sufficiently large sensor that is relatively noise free, the most important thing is lens quality. a lens that is poor quality (colour rendition, contrast, sharpness, distortion, etc.) cannot be made better with a better sensor. it can, to some extent be compensated for through processing, but there are definite limits.

if you look back to the Canon 1D - it is a 4 megapixel CMOS camera which used to be the most popular Sports Illustrated magazine shooter's body. generally mated with expensive 'L' lenses, there have been thousands of 1D photos published in SI, including dozens of covers. now, with the various Mark 2 versions and also Nikon D2 line, there are quite a few other choices by SI shooters. in any event, it shows that a quality low 'megapixel' image can actually yield decent quality images.

 

[MarkBarbieri]

Wow, there's a whole lot of stuff mixed together here.

First, there are a lot of bad things that happen when you shrink a sensor. First, each pixel gets hit by fewer photons (bits of light), so the signal part of the signal-to-noise ratio goes down. That means a relatively higher amount of noise in your picutre.

There are a lot of other issues that come into play, including the angle of light hitting the sensor sites, the gaps between sensor sites, the different power requirements of different sensors, etc.

As you stated, you do gain advantages in being able to use smaller optics. Of course, the downside to that advantage is that there is less of a tolerance for flaws in the optics. A spec of dust on the front of a 77mm lens is much less of an issue than a spec of dust on a 20mm lens.

Assuming two compact cameras had identical hardware and software, with the exception of the sensor size and appropriate sized lens, the 1:1.54 ratio means that the camera with the larger sensor can produce images of the same quality at a resolution that is 54% larger. So, where the one camera would produce barely noise-free images at 6mp at a given set of settings, the camera with the larger sensor could do the same quality with a 9mp image.

Assuming the above is correct, would that also mean that the key to "simulating" a larger sensor is just to reduce the resolution of the image in situations where quality could suffer as a result of the smaller sensor size? If so, at a ratio of 15:1, if a DSLR took an picture at 7mp, the compact with the 1/2.5 sensor could only get the same quality by taking a picture no larger than 800x600.

I'm having trouble following you. As a general rule, you can make a higher pixel quality, lower resolution image from a high resolution image by averaging pixels values to reduce the effects of signal noise. That's one reason that web sized photos sometimes look good but when you print them they look noisy.

 

[GrumpyOne]

Thanks, those posts helped.

 

[Groucho]

The concept is sound, but as I understand it, the sensors have a native resolution and the raw sensor data is going to be at that size.

Kind of like an LCD monitor compared to a CRT one - the LCD is always going to display its native resolution no matter what the PC is set to display, whereas a CRT will actually show the true resolution.

But theoretically, yes, a 1/2.5" sensor that's got a native resolution of about 1mp should be about the same quality as a DSLR that has about 15mp (assuming an identically-matched lens.