Resolution and High ISO use.

[whizkid]

Bob
 
Having read your article unraveling the confusion surrounding some aspects of resolution and the 50d brings me to a question.
Part of the complaint about the 50d and it's 15+ mp sensor is the impact of higher ISO noise levels.. Arguments indicate that it's greater resolution allows some of the better lower MP competion to catch up by ISO 1600-3200. Does your analysis follow that thinking or are they overlooking something as I believe they did when throwing the lens argument around? 

[Bob Atkins]

Since I've not done extensive side by side comparisons I really can't say with any degree of certainty whether a lower pixel count camera might give higher resolution than a 50D at high ISO settings. I can say that on my testing of the 50D against the 40D, at ISO 3200 the 50D appeared to give slightly higher resolution than the 40D.

My gut feeling in general is that it's not very likely. ISO doesn't affect resolution as much with digital as it did with film. With film the grain size increases as ISO goes up, whereas with digital the "grain" size (pixel size) remains the same. There is some resolution loss due to noise reduction techniques which can trade resolution for noise.

If there is an effect it's probably pretty small and I doubt it's something you'd really need to take into account when deciding on a camera.

[whizkid]

It is noise reduction that I meant to be commenting to concerning resolution loss. A lot has been made of the 50d's push into the 15mp range and I hope Canon makes appropriate advances to meet future high mp offerings. I'm actually dismayed that Canon went to a 14.7 mp sensor to two of it's P&S models.  Keep that P&S trend up and a disclaimer may well read "for daylight use only"

[Bob Atkins]

Why? My testing shows that the new 15 MP APS-C sensors are capable of delivering higher resolution at no increased noise when compared to the 10MP and 12MP sensors. Yes, they need more aggressive noise reduction, but they have higher resolution to start with so you can actually use more noise reduction and still come out slightly ahead at high ISO. At lower ISO you don't need the noise reduction and so you can get full benefit of the higher resolution.

[klindup]

This thread set me thinking.  CCD imagers used by astrophotographers cool the chip to reduce noise.  Do you think we shall see DSLRs with cooled chips.  A typical imager is cooled to some 30c below ambient and I have seen some that are cooled to 60 below using peltier junctions.  Some of these devices, at least the expensive ones keep the sensor clean by putting an argon filled quartz cell in front.

Ken Lindup

[Bob Atkins]

Cooling the sensor is too complex for consumer electronics. Even using a Peltier cooler will consume a lot of power, add to the complexity of assembly and there has to be some way to get rid of the heat that the Peltier cooler extracts. Plus all of of this  adds size, weight and cost.

You also have to be sure that you don't get condensation on cooled components, which isn't easy in consumer electronics which may operate in high humidity environments.

On top of that, performance up to ISO 1600 (maybe even ISO 3200) is quite good enough for 90% of users, maybe 95%.

So I do not expect actively cooled sensors in DSLRs.

[whizkid]

Bob

Again your MP vs noise position makes good sense. I hope advances in noise reduction keep abreast of MP increases.  Do you have any concerns about impact to DR and smaller pixel sizes or is that something that is countered as well.

From the other posters comments I recall Canon at one time produced a variant of it's 8mp 20D and called it the 20DA for astrophotographer use. Not a CCD of course and nothing about cooling but there was  something it did to help.

[Bob Atkins]

The 20A had a longer cutoff filter over the sensor so it passed more red/IR to record hydrogen alpha emissions from nebulae. The sensor was the same as that in the 20D and it was not cooled.