Ad
Ad
Ad
Pages: « 1 [2]   Bottom of Page
Print
Author Topic: The superiority of the 20D pixel pitch  (Read 13027 times)
AJSJones
Sr. Member
****
Offline Offline

Posts: 353



« Reply #20 on: July 29, 2006, 05:44:07 PM »
ReplyReply

Quote
Don't forget that the 5D and 20D have much stronger AA filters.

1Ds produces VERY sharp single-pixels.
[{POST_SNAPBACK}][/a]

Then you'd expect to see more moire with the 1Ds   .  I've seen very nice images from the 1Ds, but I [a href=\"http://www.geocities.com/andyandmichele/index.htm]found[/url] a while back, that even the 10D captured finer details due to its finer pixel pitch (but obviously fewer total details), leading me to conclude that the "Strength" of the AA filter in both was similar. (A bit better definition of the enemy troops on the horizon!)

 I'd love to see a good methodology for comparison of AA "strengths" so we'd really know how they stack up.

Andy
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8939


« Reply #21 on: July 29, 2006, 07:42:52 PM »
ReplyReply

Quote
The way people respond to these types of comparisons really shows what kind of photography they do.  Anyone who has ever done much wildlife photography knows that the longest telephotos become feeble when the subject is either small or distant, and people with this kind of experience know what its like to have a large frame of nothing.
[a href=\"index.php?act=findpost&pid=71938\"][{POST_SNAPBACK}][/a]

John,
Of course, if you start adding conditions then the thought experiment will be expanded and the choice maybe different. If you had just one shot left, on the basis that all the other shots on the card contained essential information and could not be deleted, and, if the angle of view of the lens was such that the FoV with both cameras (from your stuck position) was not cropped by the window frame and the interior of the room, then the larger format camera would provide more general information and any loss of specific information might not be relevant.

However, as you can see in some of the charts I've posted, the difference between the 20D and the 5D can mean that the name 'Norman Koren' is either legible or illegible, or in the case of the thought experiment, the type and model of a tank partially obscured by foliage, certain or uncertain.

However, the reason I did these experiments was not just to confirm that the higher pixel density 20D could resolve more detail with the same lens from the same distance, but to confirm that it could so so even at f22.

Having recently come across the following chart at Roger Clark's site, which relates diffraction spot size to f stop number, it would seem to me that resolution at f22 with a small pixel camera such as the 20D should be very poor.

Unfortunately, since I don't have any in-depth knowledge about much at all, I'm at a loss to explain how a whopping big Airy Disk 28.5 microns in diameter can permit a camera with a pixel pitch of around 6.4 microns to do so well at f22.

[attachment=855:attachment]
Logged
John Sheehy
Sr. Member
****
Offline Offline

Posts: 838


« Reply #22 on: July 29, 2006, 11:14:28 PM »
ReplyReply

Quote
However, the reason I did these experiments was not just to confirm that the higher pixel density 20D could resolve more detail with the same lens from the same distance, but to confirm that it could so so even at f22.

Having recently come across the following chart at Roger Clark's site, which relates diffraction spot size to f stop number, it would seem to me that resolution at f22 with a small pixel camera such as the 20D should be very poor.

Unfortunately, since I don't have any in-depth knowledge about much at all, I'm at a loss to explain how a whopping big Airy Disk 28.5 microns in diameter can permit a camera with a pixel pitch of around 6.4 microns to do so well at f22.

[attachment=855:attachment]
[a href=\"index.php?act=findpost&pid=72092\"][{POST_SNAPBACK}][/a]

Maybe most of the area of the disk only represents a small percentage of light intensity.

My opinion is that this whole idea of "diffraction limits" is ill-conceived or ill-applied.

I don't believe an arbitrary application because there is some correct arithmetic associated with it!

It is very apparent from my experience that this stuff does not apply, unmodified, to digital photography.  An airy disk does not pixelate - it only reduces pixel-to-pixel contrast, effectively raising the exposure index of high-frequency detail, but sharpening restores it.

It is also apparent that images from WA lenses get softer from diffraction at much lower f-stops than longer lenses.  I see my 10-22 zoom getting softer above f/7.1 but my telephotos don't start getting soft until much higher.
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2866



« Reply #23 on: July 30, 2006, 09:13:37 AM »
ReplyReply

Quote
However, the reason I did these experiments was not just to confirm that the higher pixel density 20D could resolve more detail with the same lens from the same distance, but to confirm that it could so so even at f22.

Having recently come across the following chart at Roger Clark's site, which relates diffraction spot size to f stop number, it would seem to me that resolution at f22 with a small pixel camera such as the 20D should be very poor.

Unfortunately, since I don't have any in-depth knowledge about much at all, I'm at a loss to explain how a whopping big Airy Disk 28.5 microns in diameter can permit a camera with a pixel pitch of around 6.4 microns to do so well at f22.

[attachment=855:attachment]
[{POST_SNAPBACK}][/a]

Ray and John,

Another table from Roger Clark's site adds some useful information. As John mentions, as the aperture decreases, the diffraction spot becomes larger and contrast is lost. However, resolution at low contrast still is present. The Rayleigh resolution criterion is MTF at around 10% and contrast can be restored by sharpening, as Ray demonstrated in his comparisons at f/8 and f/22. However, I find it difficult to believe that it would not be better to have good contrast in the first place, and it would be best to evaluate the entire image before coming to conclusions about the success of the sharpening approach.

[attachment=856:attachment]

Another approach to image restoration is with the Richardson-Lucy deconvolution algorithm. As you may remember, this method was used to rescue images from the Hubble telescope before the optics could be fixed. As I recall, the problem was spherical abberation. Unfortunately, this algorithm is not available (to the best of my knowledge) as a plugin for Photoshop.

[a href=\"http://www.clarkvision.com/imagedetail/image-restoration1/index.html]http://www.clarkvision.com/imagedetail/ima...ion1/index.html[/url]

Here is an interesting thread on the subject:

http://www.howtofixcomputers.com/bb/ftopic...05-0-asc-0.html
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8939


« Reply #24 on: July 30, 2006, 09:20:22 AM »
ReplyReply

Quote
Maybe most of the area of the disk only represents a small percentage of light intensity.


Considering that absolute resolution at f22 is only a few lines less than at f8, it seems a surprisingly large percentage to me. However, in principle that sounds like a good explanation to me.

Quote
It is also apparent that images from WA lenses get softer from diffraction at much lower f-stops than longer lenses.  I see my 10-22 zoom getting softer above f/7.1 but my telephotos don't start getting soft until much higher.


I've always thought the Canon 10-22 was a rather soft lens. I had to test 3 of them before finding one that was almost as sharp at 15mm and f8 as my Sigma 15-30. Maybe I should compare these 2 lenses at a wider aperture. The Sigma seems to behave, between f8 and f16, pretty much as other lenses do, ie. no significant fall-off in resolution.
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8939


« Reply #25 on: July 30, 2006, 09:42:51 AM »
ReplyReply

Quote
However, resolution at low contrast still is present. The Rayleigh resolution criterion is MTF at around 10% and contrast can be restored by sharpening, as Ray demonstrated in his comparisons at f/8 and f/22. However, I find it difficult to believe that it would not be better to have good contrast in the first place, and it would be best to evaluate the entire image before coming to conclusions about the success of the sharpening approach.


I wouldn't argue with that, Bill   . What I think is happening here is that the diffraction spot sizes in Roger's tables are for ideal lenses fully diffraction limited at the f stops quoted. He probably mentions that somewhere. As one stops up from f22, the diffraction spot size decreases, but not in inverse proportion to the MTF. Other aberrations detract from the improvement. A lens's sweet spot, whether it is at f4, f5.6 or f8, usually delivers only very marginally greater MTF than at the next stop down, but it's enough to get some people excited.
Logged
Pages: « 1 [2]   Top of Page
Print
Jump to:  

Ad
Ad
Ad