Thanks to everyone for their comments and votes. Please keep them coming; the more votes we get the more significant the results are.
At this moment, voters prefer the 18MP option by a small margin, with the 24MP option close behind. Only a little over 1/5 of the voters want a 12MP FF.
Let explain my own opinions, and address some of the comments made. I apologise for not naming names, but I didn't want to reply to comments individually, as many interesting points were brought up.
Sensor Resolution
Doubling the pixel count increases linear resolution by 41%, so a 24MP sensor resolves more than a 12MP sensor, but it's not Earth-shattering. At least by numbers. Judging with your own eyes, however, is another thing. Compare two images (crops courtesy of Ken Rockwell) of the same scene taken with a
Nikon D3 (12MP FF) and a
Nikon D3X (24MP FF). I'll let you decide if there is an improvement, and if that improvement is worth it. Full test
here.
However, please bear in mind that these are crops. If both photos were printed at 12x18, or even 20x30, there may not be any difference at all. Again, you have to consider you particular needs—somebody who needs to heavily crop images would benefit from the increased resolution.
Much has been said about lpmm (lines per mm), but the truth is a sensor that out-resolves a lens is not only valuable, but even desirable for those wanting the maximum detail in their images (which doesn't necessarily equate to maximum resolution). A great article about this topic was written by Ctein and published
here.
Print Resolution
The number 300dpi gets thrown around a lot when people talk about printing. While this might be the Golden DPI for inkjet printers, it doesn't apply to the printers at photo labs, which use a completely different process (much like that used to make prints from film). I have made large prints at 100dpi with no sign of pixelation; I have no qualms printing my K10D files at 16x24 (160dpi).
High ISO Performance
Common wisdom says the lower the pixel count, the lower the noise, the better the high-ISO performance. Let us once again look at some examples using our eyes. Crops of ISO 6400 shots on the Nikon D3 and the Nikon D3x (courtesy also of K.R.). You be the judge. Full test
here.
Disclaimer: The shrewd amongst you will note that both crops have the same size, when the cameras have different pixel counts. Reading K.R.'s description doesn't clear things up, so I don't know if he downsampled the D3x image to 12MP, upsampled the D3 image to 24MP, or shot the D3x image at 1.41 times greater distance so the monkey was sampled by the same number of pixels. Either way, the test gives us correct information: How would these images compare if they were printed at the same size? Comparing 100% crops to compare high ISO performance on different pixeled sensors makes no sense. This disclaimer also applies to the resolution tests. For a different take of D3 vs D3x, visit
this page.
Cropping Possibilities
A FF sensor is 2.25 times larger in area than Pentax's APS-C. If you crop an APS-C sized image from the center of the sensors I gave as options you get the following pixel count in your image:
12MP FF = 5.3MP APS-C
18MP FF = 8MP APS-C
24MP FF = 10.7MP APS-C
To answer a common question related to the above, a FF sensor with the same pixel density as as the K-7's 14.6MP sensor would be 32.8MP. One with the same pixel density as the K-x's 12MP sensor would be 27MP.
Price
Does a 24MP sensor cost more than a 12MP sensor? I don't know. I suspect there is more involved work in printing a higher MP chip, but how that might affect cost I can't tell. A faster processor might also be needed to handle the higher number of pixels. Much was written about the Nikon D3x costing 60% more than the D3 ($8k vs $5k) when the only apparent difference was the sensor, with users feeling Nikon was fleecing them, especially when Sony were offering the same sensor for $3k in the A900 and $2K in the A850. Something else to bear in mind is that Nikon's 12MP sensor was developed in-house, making their cost lower in principle, while the 24MP sensor was bought from Sony. But did they really pay $3 more for each Sony sensor than what the 12MP sensor cost them to design and build? Nobody thinks so.
My Opinion
I was surprised at the poll results because I always thought people would want a FF camera for its low-light capabilities. Most of our pictures are only seen on a computer screen at no more than 1000x1500 pixels in size—that's a 1.5MP image. Given that most of us are not printing at 20x30 on an inkjet printer, what need do we have for 24MP? I am starting to wonder what the motives are behind the GUFFCPD (Give Us a FF Camera Pentax Dammit) movement. If you're a landscape photographer, the increased resolution means nothing unless you print
very big; even in a normal poster print the extra pixels would probably be wasted. If you're a wildlife photographer, you're much better off with a K20D or K-7 and their higher pixel density on top of the APS-C crop factor; to get the same FoV with a FF camera would require buying longer lenses that get expensive very fast.
The picture comparisons I pointed to shows that a higher MP sensor can perform just as well at high ISOs as a lower MP can. It is also telling that Nikon squeeze better high ISO IQ out of the Sony sensor than Sony themselves, so it's not just about the sensor but—how the camera processes the image is of utmost importance.
My own vote was for a 12MP. Here are my reasons:
- I want improved high ISO performance.
- 12MP is enough for all my printing needs.
- 24MP images would mean buying twice as much storage space (SD cards and hard drives), and it would also double image transfer and backup times, on top of increasing processing times when using Photoshop.
- Larger images also mean lower fps, possibly smaller cache, and most likely slower processing time in the camera if costs are going to be kept down.
- I wouldn't need or want to use my DA lenses on this camera; I'd use the FF lenses I have and maybe buy 1-2 of the new D-FA lenses Pentax would release if I had the need.
My first reason is a moot point given what I just wrote about MP counts affecting ISO performance, so it doesn't count anymore, but the other reasons are still valid.
Conclusion: My Prediction
I didn't choose the MP numbers haphazardly, despite some thoughts among you to the contrary. Pentax does not make its own sensors and needs to buy them. At the moment, there are only 3 known manufacturers of FF sensors that could be available for purchase:
- Nikon: 12MP CMOS
- Kodak: 18MP CCD
- Sony: 24MP CMOS
Nikon: Sony and Kodak readily sell their sensors (the Nikon D3x uses the Sony sensor while the Leica M9 uses the Kodak sensor). I don't know whether Nikon would be interested in selling their 12MP sensor to Pentax, but given that they have recently updated it in their D3s, it's quite possible they would want to make some money on the side by selling units of the old version to Pentax.
Kodak: The Kodak sensor is a CCD, not CMOS as the others. On the Leica it reaches a maximum ISO of 2500, which according to
some tests is not even usable and lags way behind the Nikon D3x. Pentax have done great things with CCDs at high ISO, so they might be able to squeeze more out of it than Leica has.
Sony: The Sony sensor is probably cheaper, because Sony are in the business of making sensors (remember they manufacture camcorders) and because of volume (right now their FF sensor is in 3 cameras). Plus, Sony have sold sensors to Pentax in the past, and if I may give my opinion, Pentax have done an excellent job of extracting the best IQ from them (most recently: K100D, K10D and K-x).
Samsung: Samsung's two sensors (for the K20D and K-7) have not been as good, especially in high ISO performance (compared to Sony's 12MP APS-C sensor in the K-x). Furthermore, all reports indicate Samsung are not developing a FF sensor and that their cooperative venture with Pentax is over. This is why I haven't mentioned Samsung at any point.
Canon: Canon make all their own DSLR sensors, and as far as I know they have never sold any to other companies. I doubt very much they'll start doing that now which is why I didn't include them in the above list.
And the winner is... Sony!
Here are my reasons for thinking that Pentax's FF DSLR will use the Sony 24MP sensor, in no particular order:
- Pentax have a proven relationship with Sony that has worked out for both of them for almost a decade. These relationships are important in Japanese business culture and are very much present when decisions are made.
- From the engineering side of things, Pentax techs are familiar with Sony electronics by now and will probably extract better performance from a Sony sensor with less effort (which means they can do it in less time for less money).
- And speaking of money, the Sony sensor is likely to be the cheapest on the market for the reasons I outlined earlier. Given the current FF market, Pentax have to aim for a US$2k launch price; anything higher and they'll be dead in the water. If the sensor is the single most expensive component in a DSLR, it makes sense to buy the cheapest one available.
- Pentax have a large catalogue of DA lenses...and an ever-dwindling catalogue of D-FA lenses. Not being a huge company, they cannot release 5-8 D-FA lenses within 6 months of a FF launch (and another 5-8 in the 12 months following). By using a 24MP sensor they can appease (and appeal to) those users with heavy investment in DA glass by offering a 10.7MP crop mode like Nikon does. Tempting DA users to buy a FF body also means more D-FA lens sales down the line when they re-tempt these users with D-FA glass.
Now keep on voting! Despite what I predict, I still want to know what it is
you want, irrespective of whether it will happen or not.
.