I'm often complaining about the megapixel race. For example I had wished that the Canon 5D mark II had only had 14MP instead of 21, and thus could have had even better low-light performance. And the same goes for recent Canon compacts. But there's a case to be made for the opposite viewpoint.
This article is about DSLRs, but the same principles applies to compact cameras. It seems that higher resolution will compensate for higher noise, so it will even out because the total light gathering area of the sensor is the same. And then you can always downsample the higher resolution image, but you can't invent data from the lower resolution image.
Of course, if you never intend to make prints bigger than, say, 8x10" (20x25cms), then 6MP will do you. But try and find a 6MP camera today.

Article about the future shake-out of the camera market. Pentax and Olympus are endangered species.
"Of course currently the great white hope is the big-sensor compact market, including the Micro Four Thirds platform. The event of speedy on-sensor autofocus and new high-rez screens and electronic viewfinders may give us compact cameras with professional capabilities. And this may make DSLR cameras much less relevant to many photographers. Especially serious enthusiasts, because for them a small-sensor camera might not be good enough, but they still care about bulk and weight. And who knows what amazing cameras we may see in this area in the future."
And later in the article:
...actually, thinking about it, I think the Micro Four Thirds idea may be flawed. When I want a serious and flexible camera, my Nikon D90 is compact enough. And when I want a much smaller camera, I really don't care if it has exchangeable lenses. If I want something really portable, I'm not likely to bulk it up with extra lenses. So I think a fixed-lens (zoom or not) big-sensor compact is a better idea.

My Canon G10 arrived yesterday, and one of the first questions I set out to answer was how it would compare to the Leica D-LUX 4 (Panasonic LX3) at high ISO.  Panasonic, widely praised for stepping out of the megapixel race by keeping the LX3 at 10MP, claims that the benefits of large pixels include greater dynamic range and lower noise at high ISO.  Based on samples at The Imaging Resource and elsewhere, most seem to agree that the LX3 in-camera JPEGs outclass the G10 in-camera JPEGs at high ISO.  However, those who have seen the shootouts on this site know that I am mostly interested in comparing processed RAW files rather than in-camera JPEGs.

In the case of the G10 and LX3, comparisons of processed RAWs are limited by sparse third party support for the RAW files of these two cameras.  In this comparison, I've chosen to use Raw Photo Processor (RPP), which uses a modified version of dcraw.  RPP is good for this comparison in that it does not apply any sharpening besides that which is inherent in the demosaic process, nor does it apply any noise reduction when that option is disabled.  Thus, it's about as close to the RAW data as I can get while presenting a color image comparison for your evaluation.  The downside to RPP is that it does not technically support the D-LUX 4 yet, so the black level and colors needed adjustment.  Overall, I tried to approximate the black level, white level, and color between the two processed files.  The exposure for each photo was the same: f/4, 1/15s.  Both cameras were set ISO 800, a 28mm equivalent angle of view, and 4:3 aspect ratio.  The LX3 angle of view came out slightly wider.  A tripod was used with self timer and image stabilization disabled.  After processing the RAW files, the resulting G10 image was downsized using Photoshop Bicubic to match the LX3 file dimension for crop comparisons.

Pardon the choice of teddy bear test scene, but I was experimenting with a "1 Bear under the lamp" system for comparing dynamic range =).  Here is the overall scene, with the LX3 resized image on the left, and the G10 resized image on the right:



In each of the following 100% crop comparisons, the G10 crop on the right has been downsized to match the output size of the LX3 crop on the left. Differences in color and contrast are due to the RAW processor (and my use thereof) and are not indicative of significant characteristics of these two cameras.













I was surprised to find that there is little difference in either detail or noise between the LX3 and G10 at ISO 800 under these testing conditions. Both files respond very well to subsequent chroma noise removal in Noise Ninja. The RAW files are here in case anyone would like to process them differently:



This is just one preliminary example, but assuming that other examples show the same - that a downsized G10 RAW conversion can match an LX3 RAW conversion at high ISO - then it raises the question as to whether the LX3 is really a better low light camera. The following table shows the minimum f-stop (max relative aperture) at any given focal length and corresponding focal length equivalent for these two cameras:



As you can see, the LX3 has a roughly 2/3 stop speed advantage at any given focal length equivalent. The other thing to consider is which camera has the more effective image stabilization. In comparing the D-LUX 2 (LX1) and G7, I felt that the G7 had the more effective stabilization. The LX3 seems to me to have better stabilization than the LX1, but I don't have a sense yet for how it compares to the G10 in this respect.

Overall, I'd have to say that the LX3 is probably the low light champ, especially with regards to in-camera JPEGs, but the G10 is surprisingly good at high ISO.

Emil Martinec has shared a technical analysis of the Panasonic LX3 sensor at base ISO. He assessed photosite efficiency, read noise, and dynamic range. A graph depicts the LX3 photosite signal to noise ratio compared to that of the Canon G10. Very interesting article if you enjoy this sort of thing.

Link: DPReview forums - LX3 sensor analysis

Nikon expert Thom Hogan has posted an interesting comparison of the Canon G10, Panasonic LX3, and Nikon P6000. As usual, Thom has done an excellent job with the testing and analysis.

One correction: Thom says "The 16:9 aspect ratio is a love it or hate it thing. If you don't love it, this is not the camera for you, as you'll be leaving pixels on the floor and the camera suddenly doesn't have enough pixels to do more than smallish output." This is not true. In fact, the camera uses less total pixels in 16:9 than it uses in 3:2 or 4:3 aspect ratios. The reason for this is explained and illustrated here.

Thom's high ISO results and conclusions are interesting and somewhat surprising. I have a G10 on the way and look forward to testing it against the LX3/D-LUX 4.

Link: bythom - The Coolpix, PowerShot, Lumix Challenge

[Thanks Ted J]