I got into a great discussion over on Luminous Landscape where we got some really interesting information about sensor design, processors on cameras, and the relative importance of firmware updates, image quality and sensor design.  I came out with what, by now, you may recognize as my typical “show me the beef!” statement that sensors is sensors, and they don’t have a whole lot to do with image quality. 
 

Naturally, I got fed my words, and had to do a little backpedaling.  Of course there have been advances in sensor technology, and the sensors today are a big part of why our image quality is so much better than camera only a few years ago.  And there ARE some big differences in sensors from generation to generation, and especially price to price..  but still...
 

I got inspired by this great link from Dalsa talking about the difference between CMOS sensors and CCDs that’s packed with tidbits of information about sensor design- like, for instance, did you know that although CMOS sensors can be cheaper because they can be mass-produced, their real advantage is that they can easily be tailored to a custom application (read: specific camera) which gives you great performance but also eradicates any cost advantage?  Huh?  Didja? 
 

I put together this matrix of what happens in the process of making a digital image from the pixel, and where it happens.  I also threw in some of the things that can go wrong- either through bad design or just cost-cutting.  Here’s that for you. 
 

So what, after all this, can you conclude?  I’m sticking to my guns.  Sensors is sensors…  but with a few caveats.  First, look at sensors in the generational sense.  A current sensor is going to perform head-and-shoulders better than an old sensor that’s a few generations back in development.  They just know how to get more out of them now, and they're contiuning to eek more out out of them with every new generation.
 

Second, there’s a huge difference in the performance from a straight cost standpoint.  At the bare minimum, it takes a lot of development to get performance out of a sensor, and by most of the figures I’ve been able to find, the numbers look like ½ to ¾ the cost of many cameras is the sensor alone.  There are some pretty significant manufacturing costs too, especially when you talk about bigger sensors.  The math in effect when you're looking at silicon chip manufacturing is staggering.
 

The price of the camera is figured on all the usual factors, naturally the total quantity of sales estimated figures big into it, and that’s why you’re seeing such great performance now from under-$1000 cameras- they can just sell a lot more of them and make more money from them to offset the development costs from a better sensor.  However, there’s just no way you’re going to get the same performance from two sensors in the same generation that are in cameras thousands of dollars apart. 
 

My bottom line?  Really for any camera in any price-point, I’m not too interested in what sensor is being used or why.  I can assume that any major manufacturer can compete at a price point or they’re just not going to be there, and that whatever tradeoffs they have made are going to produce image quality are going to make some people happy.  Not necessarily me. 
 

Add to that the fact that there are some very important steps outside of the sensor that can make significant improvements in image quality, even in firmware updates (I’ve seen those with my own eyes, by the way…) and I’m back to the basic question- can you really make any general statements or come to any conclusions about a camera’s performance from knowing what, or whose sensor it is?  Can you compare two similar cameras based on the sensor, and come to any reliable conclusion about image quality?
 

Nope.