Lots of good information on this thread - some of it a bit incomplete and some a bit misleading, but I think it's possible to sort it out:
All things being equal (they often are not) more resolution is better than less resolution. What needs to be considered is how much resolution is required to properly define the boundaries of fine detail in the capture.
Sometimes that can be done very well with a 1.5 megapixel sensor, and sometimes 6 megapixels is insufficient. If the boundaries of fine detail are properly defined with a 1.5 megapixel capture, noise levels are low, dynamic range is sufficient and colors are proper, then poster sized prints may easily be made with good interpolation software.
If fine detail is so small or distant that even a six megapixel capture doesn't properly define it, and this detail is necessary, then enlarging to the point of revealing the flaw will prove unsatisfactory.
How about some common examples: Let's look at a head an shoulders portrait. Do we "need" high resolution capture to get a satisfactory 8x10 print? Usually, a two megapixel capture is more than sufficient to get a beautiful 8x10 when the subject is this large and the available pixels are vested in a reasonably small area of "geography". There are a couple reasons why this is so. First, portraits are intentionally made somewhat "soft" to enhance the natural beauty and hide flaws. Do we "really" want to see enlarged pores, tiny blackheads, whiteheads, etc? Second, features which actually "count," such as eyes, noses, etc., are defined by sufficient capture pixels that interpolation algorithms can do their jobs and not make "mistakes." So the end result is a pleasing enlargement which doesn't over emphasize flaws and properly defines important features such as eyes, noses, lips and even hair.
Now let's look at the polar opposite: How about a detailed landscape where the foreground consists of several hundred feet of beautiful trees with multi-color leaves turning to Fall shades and the background is rugged mountains with myriad rocks, maybe a bird or two in the sky, etc. What would happen if we were to capture this scene with a two megapixel sensor and enlarge the print. There would not be sufficient pixels alloted to the tiny geography in which leaves of distant trees were painted. The birds in the sky might consist of only a few (half dozen) "marker" pixels, etc. From a distance or in a small print, the human brain would do its own interpolation and we would be quite happy to "know" that those few marker pixels were "leaves" or "birds." But if we enlarge this to poster size, our brains will no longer be "fooled" as they are in oil or watercolor paintings. The interpolation algorithm will instantly reveal that these little "dots" in the sky are just that, dots. Not birds, which we want to believe, but a nice faithfull reproduction of "marker pixels" so that we are ultimately dissatisfied with the print.
Could a six megapixel capture suffice in this case? Maybe - and maybe it would take a large format film to do it, or we would have to view the poster sized print from a greater distance to again allow our brains to work as they have been "trained" to do.
Lower resolution captures are often "better" than higher resolution captures because "all things are not equal". We must choose the appropriate tool for the job. I frequently get much better prints from my 3 megapixel EOS-D30 than I do from my 5 megapixel DSC-F707. Why? Because the D30 makes much "cleaner" captures with near zero noise. Are they always "better" than the 5 megapixel Sony captures? Absolutely not! If I were shooting equipment or books or something with super fine detail and enlarging it greatly, the resolution deficit would immediately be apparent with the D30 capture and the F707 capture would be superior.
The bottom line is that when all other factors are equal, more resolution is generally better. On the other hand, this is rarely the case so no definitive answer can be given except on a case by case basis.
Lin
http://204.42.233.244