Please Someone Technically Intelligent Help me Understand The Less MegaPixel Advanta

Started Mar 2, 2012 | Discussions thread
Graystar
Veteran MemberPosts: 8,373
Like?
Re: Please Someone Technically Intelligent Help me Understand The Less MegaPixel Adva
In reply to PhotonTrapper, Mar 4, 2012

PhotonTrapper wrote:

PhotonTrapper wrote:

I would think you want to start being limited by diffraction at max about mi-way through your f stop range.

Graystar wrote:

First, there's no logic at all in crippling your wide-open resolution just because you have an aperture limitation somewhere else. That's thought-less.

http://www.clarkvision.com/articles/sampling1/index.html

Graystar, do your really mean "cripple" your WIDE-OPEN resolution??? Did you rather want to mean the smaller apertures? If diffraction starts to be noticeable at mid range of apertures (like I suggest with 6 megapixels on that sensor size), you are not crippling the wide open side, the wider apertures are not affected by diffraction then.

If the pixel density is a match for diffraction at f/4.8 (for example) then it's not dense enough to capture all the detail at f/2. So you've crippled your ability to resolve image detail when wide-open.

Graystar wrote:

Second, Nyquist limit calculations work great for constant frequencies, of which photographs most definitely are not. Work by Roger Clark demonstrates that detail with dimensions as small as 0.14 of the Airy Disk can be distinguished.

Yes, Nyquist limite is a function of wavelength, and most images deal with the full specturm (red to blue), I agree, but that has little effect on the numbers here because green light is used (middle of spectrum) in the equation.

That has nothing to do with it. The issue isn't the wavelength of light. We're not sampling wavelengths. It's an infinite number of Airy Disks that are being sampled, and there's no rhyme or reason to the placement of these disks. The random nature of the scene is the issue that was raised and reviewed in the Roger Clark article.

With a typical Bayer filter and a 1/1.7" sensor (5.7x7.6mm), your pixel pitch at 20 megapixel would be of the order of 1 micron and your diffraction would exceed your pixel resolution for all apertures (starting at 1.4) for green light (middle spectrum) so 20 megapixels is not realistic for a sensor of that size.

Only if you don't believe the Roger Clark article. Here's an article that compares all the Canon G models from 1 to 10 against each other. On the A3 print test you can clearly see an increase in detail from the G9 to the G10 on the first image. For some reason the other G10 images appear slightly out of focus and that wasn't corrected. Still, it demonstates that 14.7 MP can get you more detail than 12 MP.

http://pixinfo.com/en/articles/canon_powershot_g_evolution.3/

So to saying that 20 MP is not realistic doesn't jive with what's actually happening. Canon went back down to 10 MP due to the demands of a vocal minority, but have now started to increase the resolution again with the 12 MP S100.

.

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