*** Since the discussion on this subject was originally brought up at this thread, I thought I'd split it as it is an interesting subject and continuing in that thread would've been distracting to the OP's request. Drew



Drew I think the discussion about compensating for noise by simply shrinking higher MP images is a BS one, and yes I have compared real life images. And I know you disagreed with me on this before. Furthermore these have been similar images taken with similar framing of the image topside and underwater (not later manipulated and magnified just to look for noise). I have actually run test strips in black and white and color in my 4800 printer to demonstrate this to myself. Cameras used have been a D200, D700, D3 and D3x. The effect of high ISO noise that is more important than just seeing grain, is actual degradation of the image, which becomes particularly evident in the shadows. The result is muddiness of the shadows and loss of finer gradations of tone within the shadows, this is NOT improved by more MP. Muddy grey is muddy grey, no matter how many MP you have.
The striking thing about the abilities of the newer high ISO cameras (already evident in the 5D, and lost a little in the 5DMKII) is to really be able to SEE into the shadows in low light in a way that is closer to the way our eyes are able to, and much less about noise graininess in my opinion. (Except in as much as noise degrades the shadows). If one looks at movies like Nocturne, and the D3s bear movie, what's striking is that these cameras can actually see in the dark much more like our eyes do. Taking lesser capable cameras and simply turning up the ISO simply does not achieve the same thing irrespective of magnification or MP. At lower ISO like 800 or so, this is less evident and the higher MP argument holds water, but at 1600, then 3200 and 6400, it simply does not stand up to scrutiny. I've seen this very clearly comparing a D3 and D3x, beyond 800 ISO having more MP is of no value in dark water, available light etc. I saw this clearly demonstrated in Cocos.
I also do a lot of dance and performance photography, frequently using 3200 and 6400 without flash of course, where shadow penetration, detail in the shadows, is the issue, not whether there is graininess within the image.
Underwater I have proved this to myself as well seeing strikingly different images ( particularly in the shadows) comparing a D200 and D700 at 800 and 1600 ISO. In well lit areas, the difference is far less evident.
I am not knocking MP, I too am waiting for a higher MP camera, but I'm not convinced it will be the best camera for all situations.
Unfortunately, the ability to crop higher MP images does not always hold water, underwater. (Sorry for the bad pun) Often the need to crop will be because the animal is too far away, in which case degradation of the image will have more to do with water interface.
Obviously we will never agree on this, and I promise this is my last post on this thread as we have certainly flogged it to death, but to get back to the original main point here, is it worth switching systems when either Nikon or Canon come out with new models. In my opinion the answer is no with presently available cameras unless 1024 video is the critical factor here and now. Or unless you really need to count eel spots.

Just to address the discussion about higher pixel count and high ISO noise that Drew and I have beaten to death; I think the reality is in the results despite the opinions of the experts quoted.
http://www.dpreview.com/reviews/nikond3s/page30.asp
Pretty clear that the Canon 1DSMKIV clearly is inferior to the Nikon D3s once 3200 to 6400 ISO is reached, and higher resolution does not compensate for the increase in noise with smaller pixels. Not important for underwater of course.

Sorry Jeff but that is the wrong camera since the 1D"S" mark IV isn't out yet.
You are comparing the 1D Mark IV, which is a 1.3 crop camera to the D3s. One of the basic principles of comparing pixel density is same sensor size. Once you compare different sensor sizes, the math is skewed due to the physics of a larger sensor. Now if you were to compare it to the D3X or 1Ds Mark III, I think you'll be seeing differences in resolution and how processor manages noise reduction. It depends on many things including RAW software.

I'm splitting the thread so we don't take it way off topic.

Sorry, meant the 1DMkIV. Just pointing out that despite the hype that the 1DMKIV may be equivalent or better at high ISO than it's competing Nikon D3S camera as a sports / photojournalism camera, that does not appear to be the case at least in the high ISO department. They specifically point out that actual resolution starts to deteriorate of the Canon relative to the Nikon at 6400 and beyond despite the higher pixel count.

I would have to say that if professionals choose a brand for whatever reason and it doesn't perform to their expectations, then they will change. There is a significant increase in Nikon users at sporting events. Has Canon lost market share? Definitely. That's what happens what a company screws up on a product/s, look at Toyota! At the same time, pro shooters may not want to shoot at 6400 or higher. I think it was Doug Perrine who won't shoot over ISO800 on his D700 because he doesn't like the noise. Noise performance is only one aspect of a camera; AF performance, fps, lenses available etc. If subjective criteria is going to be the variables, then there's nothing to talk about, is there?
Pixel density and noise performance correlates to sensor size, design, processor etc. Like I've suggested, check out Daniel Browning and the other guys I've mentioned. The science is there.

I would love to have a camera that does everything :-) On Saturday night I shot a dance production with the 1DsIII at ISO1600. I would have loved to have a MkIV to shoot 3200 or even 6400 and wouldn't have minded the crop one bit. But I wouldn't want the MkIV as my sole camera as I'd much rather dive w/ the sMkIII. It's too bad that they can't come up w/ an interchangeable sensor camera to help w/ this!

Cheers
James

So after all of this what was the conclusion for shooting in the ISO 100-400 range where most of us mere mortals shoot - is it worthwhile to have more megapixels for things like producing big prints etc?

James I think the 1DMkIV can shoot at 3200, just not as well as the D3s, especially in terms of chroma noise. If one had to choose based on ISO performance alone, and not worry any other criteria, then the D3s is the choice.
The interchangeable sensor is what Scarlet is trying to do. And we'll see how their sensor plays out compared to the big boys at Canon.

As for megapixels for big prints, I don't think anyone can deny it certainly helps. The subjective nature of 'good enough' comes into play. Can a 8 mp image be used in a 10 ft poster or even the side of a building? Sure... but the point is there are cameras like the Phase Ones, Leaftus and Hassenblads which shoot up to 60mp for magazine spreads. Tools for a trade.

Shooting for big prints was why I changed, and the extra MP also gives me the chance to crop a little for composition if I want to.

I agree Drew. If Nikon is going to just stick with 12 megapixel FF as their sensor size/density of choice then every time they iterate, they'll probably get one stop or two better ISo sensitivity and quality. They've definitely shown this w/ the D3s

If Canon wants a leg up again on the sports camera, they need good AF, and at least as good or better IQ. That means they better get their act together if they want to stick w/ the 1.3x crop and higher pixel density that comes w/ that AND a jump to 16 megapixels.

Cheers
James