I’ve read this on several sites ... most of the time as a way to try and justify not moving to a higher resolution camera ... why move from a 12 mp camera to a 21 mp camera ... after all it takes going to 48mp to double the resolution of a camera.

OK, from a purely mathematical perspective this is true.  To double your horizontal and vertical resolution densities, you have to increase your overall pixel density by 4 times.   So what’s wrong with this statement?

It’s pretty simple actually ... it makes an assumption that it takes a doubling of resolution to obtain a benefit.  But the truth is doubling resolution normally has a staggering impact, and far smaller increases are noticeable and even substantial.  You don’t have to double the MPG of a car to get a significant and helpful benefit ... even as little as 10% to 20% is substantial.

So lets talk about this from a little different perspective.  Imagine a scene that contains an infinite amount of edge frequency ... detail so small that no camera has a chance to resolve it up to detail so large that even a 2mp sensor can capture it.  If you had cameras where all other factors were equal except pixel resolution,  an increase in horizontal/vertical pixel count of 10% would record information that was 10% smaller.  This is about the difference between a 30d and a 40d camera ... the 30d has 3504x2336 sensors,  the 40d has 3888x2592 sensors.  This means there are 10.9% more pixels in each direction to record data, yielding a detail threshold that is about 11% better than a 30d.  Of course, this is assuming all else is equal, which usually isn’t the case.

The other main factor is noise, and as pixels get smaller the normal side effect is noise.  One way around this is better sensors, which we are seeing.  Of course the more obvious way is to make a larger chip so the sensor sites don’t have to get smaller.  This is the advantage full frame dSLR’s have and why Nikon has finally admitted they were wrong (for years they felt the APS size sensor was the future), and have introduced the Nikon D3.  I personally believe they are working very hard on a 20-24mp version of the D3 ... they fully understand why they were wrong and why full frame sensors are here to stay.

For the past few years this is how I have looked at increasing resolution ... what is the percentage increase in pixels in each direction.  To increase those dimensions by 100% does require 4 times the pixels, but as I stated before, this much of an increase is HUGE ... if you don’t believe me just check out this comparison of my PhaseOne p45 against my friends Nikon here, then tell me there is no in between.  This is where this argument falls flat on its face.

It’s true as pixel densities increase, it becomes harder and requires a lot more pixels to make a change.  But the concept stays true, and will continue to work until we reach a point in time that the sensor is no longer the limiting factor in resolution.  I believe at some point in the future a sensor may be designed with enough resolution that it exceeds the resolution of any lens.  At that point, increasing sensor size will yield very little benefit, unless new optics are created which again out do the sensor.

So don’t be fooled ... a camera with 20% to 30% more pixels in a single direction, if well made and using good glass, can provide a substantial increase in ability to capture detail ... - it doesn’t take a 400% increase to gain a benefit.