Too many megapixels? Six is enough for an 8x10.

Started Apr 26, 2009 | Discussions thread
VTL Regular Member • Posts: 118
Re: [2/2] Noise power is a function of spatial frequency

bobn2 wrote:

This diffraction issue is commonly misunderstood. Assuming that you
stop down the aperture to achieve depth of field, then no format is
more diffraction limited than any other.

Yes, I know (as my statement vaguely alluded to).

Thus, if you want to use this argument as
a limit for pixel count, it applies equally to all sensor sizes, and
argues against high counts, not small pixels.

Well, I was more or less arguing against high pixel counts -- not against small pixels, per se. What I wanted to note was that different formats run into certain physical limitations at different points.

Personally, i don't think it is a strong argument. given that
diffraction blurring is inherently linked with deep DoF, I don't see
why we shouldn't take advantage of the additional resolution
available at shallower DoF's.

I agree on the whole. However, P&Ss can't achieve very shallow DoF, and start out very close to "diffraction limited" (a slightly meaningless phrase) and with quite large DoF. If they could achieve shallower DoF (and move away from "diffraction limited"), then more pixels would be useful -- but they can't.

On the other hand, even if a DSLR is "diffraction limited" at f/8 or f/5.6, there are many lenses with larger apertures (it's a sad day when a lenses doesn't manage an aperture larger than f/5.6!). Even there, one will eventually run into problems, since wide-aperture lenses don't perform so well wide-open (typically). (We're obviously still far from this point, though. But, even then, diffraction issues do pose a limit in that the extra resolution only becomes useful under certain circumstances.)

This is a similar issue. The camera shake argument is one against
high pixel counts and affects all format equally (at least
optically), but why shouldn't we have the extra resolution in the
cases when we can use short exposures or tripods?

This is not entirely true. Angular (rotational) motion affects all formats equally. Translational motion doesn't. (Moreover, the mass of a camera -- which is related to the size of the format -- adds additional dampening, remembering that the human holding the camera isn't scaled.)

I find the results from a 12MP camera significantly better than ones
from a 8MP camera, I would expect a 24MP camera to be better still,
whatever the format.

I would say it depends on the camera. A 12 Mp DSLR is significantly better than an 8 Mp one (though the practical difference isn't as large as one imagines). For P&Ss, it's more of a mixed bag -- for some, with lousy lenses, the quality hardly increases at all.

Moreover, for P&Ss, the tiny sensors (and consequently exceptionally
small tolerances required) and relatively low prices mean that the
gains in IQ above 8 Mp are very small. [etc.]

It would be interesting to produce some evidence backing up these
statements. I find that a lot of people say this kind of thing, but
never substantiate it with real evidence, or when they do it's based
on flawed comparisons, such as 100% crops.

It would be, though making fair comparisons is rather difficult (given, first of all, generational differences in technology, not to mention different designs, lenses, etc.). I don't have the equipment (e.g., cameras of the same generation) to do such tests.

However, one can learn something from, e.g., tests of lenses. MTF charts do give an indication of how useful extra resolution will be. For DSLRs, resolving much beyond what I indicated gets more and more difficult. (This is not to say that there's no improvement -- as I said, all other things being equal, more pixels is always better -- but to say that the returns seriously diminish.

I disagree, for people whose photography is serious enough to warrant
an SLR (which doesn't include all SLR owners) the difference between
8MP and 15MP is significant and easily detectable.

I would say it includes only a fraction of DSLR owners (and only for a tiny fraction of the pictures they take!). And I'm willing to admit that there's plenty of room for improvement above 8 Mp, but I'm not entirely convinced that the different between 12 and 15 Mp is so large (evidently, measured in linear terms, it isn't).

For me, and my current collection of lenses (well above basic kit), 12 is enough, 15 is okay (though adding a bit to workflow pains), and 20 is probably too much. This is not to say that 20 wouldn't provide better quality, just that the gains are too small to justify the pains.

Post (hide subjects) Posted by
VTL
VTL
VTL
igb
igb
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow