Recent discussions on these forums about the announced Sony sensor prompted me to do a simple calculation of the number of 5-micron pixels for different sensor sizes. I seriously doubt this is anything new. At any rate, here are the numbers:
Format Dimensions(mm) Area (mm^2) Megapixels(5 micron pixels)
---------------------------------------------------------------------------
35mm 36x24 864 34.6
APS-C 1.5x 23.6x15.7 370 14.8
APS-C 1.6x 22.2x14.8 329 13.2
4/3 17.3x13 225 9.0
1/1.7 7.6x5.7 43 1.7
1/2.5 5.76x4.29 25 1.0
A few observations:
1. Point-and-shoots with 1/2.5" sensors ought to have 1 MP if they are to keep noise down!
2. The limit for 1.5x APS-C seems to be just under 15 MP. The Samsung sensor in the K20D is almost at that limit. I hope nobody ventures past 15MP on APS-C sensors.
3. Four-thirds with 10 MP is a bit (about 11%) past the limit, but 10 is much a nicer "marketing number" than 9.
4. For full-frame there seems to be room past past Sony's announced 25MP. Who'll be the first out with a 35MP FF sensor? Hard to tell, but I wouldn't be surprised if a 30 or 35 MP Samsung sensor showed up in a Pentax FF DSLR.
5. I think M.R.'s predictions regarding under $1k full-frame DSLR within 2 years (in his recent "tea leaves") are not far off. If one thinks about it, the mechanical parts cost is essentially the same for 4/3 and APS-C as it is for full-frame. Electronics on the other hand keep getting faster, better and cheaper, and that's what will drive prices down. That is, in the long run, a full-frame DSLR should not cost more to produce than a reduced-frame one. As we know, lower production cost does not translate automatically into lower prices. However, competition in the marketplace will take care of that.