New Hot Thread!

Started Apr 28, 2009 | Discussions thread
Flat view
chuxter Forum Pro • Posts: 21,714
New Hot Thread!

2 Days ago, PhotoGo started a new thread about an article re "Why More Megapixels Don't Make Better Pix" by Chris Gaylord of ABC News. Every time somebody talks about this subject, it's quite popular...this thread maxed out in about 18 hours.

Most posters take a hard stand, either for the proposition that there is a negative correlation between the number of pixels at some point, ie, more pixels are good up to some undefined point, then they actually make things worse.

Even though Chris Gaylord clearly qualified his comments by stating, "Such a counter-intuitive snag mostly affects tiny digital cameras, the ones compact enough to fit in your pocket.", posters took him and his "source", Amit Gupta, founder of to task by immediately extending the concept to large sensor cameras.

I got into a discussion with Oly Canikon...we were sorta in agreement/on the same side of the argument. bobn2 jumped in with a challenge (to me) that I was not able to accept (in time). He stated, "I would be very interested on your opinion, as a scientist, on the reasons for there being such a 'sweet spot'."...I had brazenly admitted that I believed in a "sweet spot" w/o referencing any evidence. The best "evidence" is Roger Clark's site:

His Apparent Image Quality graph is often referenced here on dpr in discussions related to IQ vs MP. I'm sure bobn2 has seen it. Roger clearly explains why AIQ dips at high pixel densities:

"When pixels become very small, they hold so few electrons that dynamic range suffers, and this causes the turn down in AIQ at pixel sizes below 2 microns pixel pitch."

In contrast to Roger, bobn2 believes, "...There is a monotonic increase in IQ with pixel density. The slope of the increase diminishes, but never goes negative."

This is the HEART of the differences we have about IQ vs MP. One group believes that IQ continues to improve as the number of pixels increases...the other group believes that IQ improves up to a point and then decreases. The second group can't agree on where the "point" (where IQ downturns) is...every time we try to have a discussion about it, people like bobn2 jump in and distract us.

Roger Clark thinks it's at a pixel pitch of about 5 microns. I tend to locate it at 6 microns. These differences are not very important and they are subjective. Roger clearly states that fact: "Apparent image quality is a subjective measure..."

His AIQ graph is VERY busy and difficult to understand w/o some effort. I offer MY confusing graph to this discussion:

In this graph, I show two curves, one of photosite area vs photosite pitch (area is just the pitch squared) and the other of resolution vs photosite pitch. I assumed that 2.5 pixels would be required to resolve two line pairs (you can do the same graph if you want and use a different assumption...on your own time). I annotated the diagram with the pitch values of some common cameras (both types), mostly Canon & Nikon, with a few Sony models. Again, you can add YOUR camera to the diagram was difficult to get even 46 models on there!

Note that both curves are dotted at each end. That is where reality stops and imagination begins...there are simply no cameras in those regions (at least that I could find in my search). If cameras existed with a pixel pitch greater than 12 microns or less than 1.5 microns, their performance would be as shown...I think.

I added some "clouds" to show that some camera models do better in low light settings...some do better resolving detail...some are good compromises...and some are only good for snapshots (non-discerning audience). I showed that these groupings are broad and overlap.

There is a yellow "SS" circle to show where MY "sweet spot" is. As I said, Roger's is at 5 microns and yours can be wherever you please...nobody is dictating where "your" sweet spot has to be.

BUT, it's clear that there IS a sweet spot! Any sensor with a large photosite WILL behave better in dim light as regards noise and dynamic range. Any sensor with a greater number of photosites WILL behave better as regards resolution. These two factors are opposing, so SOMEWHERE in the "middle" will be an appropriate compromise. The middle HAS to be better than the ends, because at one end you don't have any resolution and at the other end most photosites don't get any photons (the few that do only have ONE...very hard to tell whether the one electron was associated with a photon or was noise)!

There are extraneous issues about lens resolution, time to readout data, etc., but these are not too important to understanding the HEART of the issue, which is whether a sweet spot exists in the middle somewhere.

I hope my graph will help somebody who is confused by the seemingly continual heated arguments. The Bottom Line is that there is NO SINGLE SWEET SPOT, but generally things are better in the middle. If you want/need better low-light performance, then shift the sweet spot toward larger photosites. If you want/need better resolution, then shift the sweet spot toward more photosites.

If you want BOTH, then you need a MF camera with lots of big photosites!

-- hide signature --

Charlie Davis
Nikon 5700, Sony R1, Nikon D300
'I'm from Texas. We have meat in our vegetables.' Trenton Doyle Hancock

 chuxter's gear list:chuxter's gear list
Nikon D810 Nikon D500 Tamron SP 90mm F2.8 Di VC USD 1:1 Macro (F004) Sigma 24-105mm F4 DG OS HSM Tamron SP 150-600mm F5-6.3 Di VC USD +2 more
Flat view
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow