F-stop limits with 39MP sensors

[BJL]

The recent article at this site suggests that at f/11 in 36x48mm format, diffraction limits resolution in mid-spectrum green to about that of a 31MP sensor, and by f/16 the limit is 28MP for blue, less for longer wavelengths. See Table 3 of http://luminous-landscape.com/tutorials/resolution.shtml

If so, about f/10 would be the upper limit for getting maximum possible resolution from such a sensor.


To users of 39MP sensors:

- How well does this theory match your experience?

- What f-stop limits are you setting on yourself when you wish to avoid detectable loss of sharpness to diffraction?

- Or are you perhaps not worrying about squeezing every last pixel's worth of detail out of your 39  million of them, instead letting the res. fall where it may once DOF is adequately controlled?

[Mort54]

The recent article at this site suggests that at f/11 in 36x48mm format, diffraction limits resolution in mid-spectrum green to about that of a 31MP sensor, and by f/16 the limit is 28MP for blue, less for longer wavelengths.

I use a P45+ on a Mamiya AFD II with a basic set of Mamiya lenses. I haven't done any scientific comparisons, so my comments are anecdotal in nature.

In general I'd agree that sharpness starts to decline by f/16, and maybe even sooner, depending on the lens. BUT, most (and probably all) of my lenses are optically optimal at somewhere between f/5.6 to f/11, depending on the lens. So it's hard to tell if the loss of sharpness is due to diffraction, or to the fact that I'm using a lens at a less than optimal f stop. It's probably a little of both. In any case, the decrease in sharpness isn't very noticeable at f/16. I seldom use smaller apertures than f/16, so I can't really comment about f/22 and smaller. You really have to look closely to see the degradation at f/16, and I certainly don't hesitate to use f/16 if I need it for depth of field.

One thing I have been doing more of lately is focus bracketing, and then blending in PS. I'm doing this not so much to minimize the diffraction effects, but to get improved depth of field. With a really close foreground and a nice background that's crying out for lots of detail, it's hard to satisfy both without tilt lenses. So that's where focus bracketing comes in. When I do this, I typically also drop back on the aperture - I'll usually use f/8 or maybe f/11. Obviously, focus bracketing, like stiching, comes with its own set of limitations.

[Bill Caulfeild-Browne]

The recent article at this site suggests that at f/11 in 36x48mm format, diffraction limits resolution in mid-spectrum green to about that of a 31MP sensor, and by f/16 the limit is 28MP for blue, less for longer wavelengths. See Table 3 of http://luminous-landscape.com/tutorials/resolution.shtml

If so, about f/10 would be the upper limit for getting maximum possible resolution from such a sensor.
To users of 39MP sensors:

- How well does this theory match your experience?

- What f-stop limits are you setting on yourself when you wish to avoid detectable loss of sharpness to diffraction?

- Or are you perhaps not worrying about squeezing every last pixel's worth of detail out of your 39  million of them, instead letting the res. fall where it may once DOF is adequately controlled?
[a href=\"index.php?act=findpost&pid=205318\"][{POST_SNAPBACK}][/a]

I've tested all my Mamiya lenses for their best apertures. Generally, f11 is the optimum on my P45+. (There are some exceptions - I find the 300mm APO just as sharp at 5.6 as 11, and the 28mm seems at least as sharp at f16 as at f11.). In fact, I'd say the Table 3 you quote is right on - with one caveat.

Diffraction limits assume a "perfect" lens - which none of us own. There are enough anomalies even in the best of lenses that while theoretical diffraction limits are reached at f11, you'll often not be able to tell the practical difference at f16.

Personally, I'll only use f22 if I really really gotta, but f16 is fine. Still, I generally leave the camera set at f11 for landscape work and only change if there's a good reason.

Incidentally, I usually read the MFD vs DSLR debate with great amusement (why compare such different tools?) but I notice nobody seems to acknowledge the advantage a larger format has when using wide angle lenses. Try a MFD with, say, a 28 mm lens compared to a DSLR at about 20 mm. The difference in detail captured is quite significant to my eyes.

bill

[HarperPhotos]

Incidentally, I usually read the MFD vs DSLR debate with great amusement (why compare such different tools?) but I notice nobody seems to acknowledge the advantage a larger format has when using wide angle lenses. Try a MFD with, say, a 28 mm lens compared to a DSLR at about 20 mm. The difference in detail captured is quite significant to my eyes.

Oh no Bill,

Ray might be trolling around, if he sees this comment he is sure to come out from under his bridge.

Cheers

Simon

[marc gerritsen]

Obviously, focus bracketing, like stiching, comes with its own set of limitations.
[a href=\"index.php?act=findpost&pid=205369\"][{POST_SNAPBACK}][/a]

One disadvantage of focus bracketing is that when you try to blend 2 photos from a different focus plane is that they do not perfectly match, the further apart the f-plane is, the less they match,

I usually shoot f13 on a H3D39 and usually use a 35mm, so that gives me a reasonable depth of field and I tend to keep things more focused in the background then in the foreground

I did a test  with f30 f13 and f6.8
font focus middle focus and distant focus

100% crops of each of the 9 photos have been collated in this file

have a look

cheers
m*

[Khun_K]

I tend to believe it is diffraction.  When I was using P25 with my Contax 645, for example with 120/4 Apo-Makro Palanar, I can shoot at f/22 with good sharpness but when I move up to P45 and P45+, I lost one stop and can only get good sharpness at f/16, or not over 1/2 stop further from f/16. I am believing if one day we use for example 50-60mp back, we may need to further open to f/11.

The recent article at this site suggests that at f/11 in 36x48mm format, diffraction limits resolution in mid-spectrum green to about that of a 31MP sensor, and by f/16 the limit is 28MP for blue, less for longer wavelengths. See Table 3 of http://luminous-landscape.com/tutorials/resolution.shtml

If so, about f/10 would be the upper limit for getting maximum possible resolution from such a sensor.
To users of 39MP sensors:

- How well does this theory match your experience?

- What f-stop limits are you setting on yourself when you wish to avoid detectable loss of sharpness to diffraction?

- Or are you perhaps not worrying about squeezing every last pixel's worth of detail out of your 39  million of them, instead letting the res. fall where it may once DOF is adequately controlled?
[a href=\"index.php?act=findpost&pid=205318\"][{POST_SNAPBACK}][/a]

[Frank Doorhof]

Let's just hope we don't go into the area of the 1DsIII where diffraction kicks in at anything above f5.6

f16 is still workable but f5.6 is terrible for a studio camera.
Most of my work is shot between f8-f16 for MF and for the DSLR on f5.6-f11

Another reason to just stay with 22mp backs and make them perform even better for DR and higher ISOs.

[jonstewart]

I have much the same experience as Mort.

No discernable issues when you really need to use f16. P45/Mamiya.

It might be of interest than I recently chose a Schneider 38 Super Angulon, over the 35XL, since the fall off in image quality from f11-f16 appeared much gentler in the 38 than the 35, with this sensor.

Cheers

[Bill Caulfeild-Browne]

Incidentally, I usually read the MFD vs DSLR debate with great amusement (why compare such different tools?) but I notice nobody seems to acknowledge the advantage a larger format has when using wide angle lenses. Try a MFD with, say, a 28 mm lens compared to a DSLR at about 20 mm. The difference in detail captured is quite significant to my eyes.

Oh no Bill,

Ray might be trolling around, if he sees this comment he is sure to come out from under his bridge.

Cheers

Simon
[a href=\"index.php?act=findpost&pid=205397\"][{POST_SNAPBACK}][/a]

Whoops! Sorry...what was I thinking?
Bill

[Mort54]

One disadvantage of focus bracketing is that when you try to blend 2 photos from a different focus plane is that they do not perfectly match, the further apart the f-plane is, the less they match,

Yes, that is one of the disadvantages. Changing the focal plane seems to change the focal length ever so slightly. One trick to minimize this effect is to stack your bracketed images in Photoshop and run "Align Images" on them. This resamples the top images to align with the bottom image. Then do your blending. It works pretty well. There are other tricks, but they tend to be image dependent.

Mort

[ternst]

Have any of you guys found Helicon Focus to work with landscape images? Works pretty well with macros, but I've never gotten great results with wider shots.

Tim Ernst in Arkansas

[Mort54]

It might be of interest than I recently chose a Schneider 38 Super Angulon, over the 35XL, since the fall off in image quality from f11-f16 appeared much gentler in the 38 than the 35, with this sensor.

Hi Jon. This is a little off-topic. I've been trying to talk myself into supplementing or replacing my Mamiya AFD II kit with something like an Alpa 12TC. My goal is to put the best possible glass in front of my P45+ for landscape shooting. It sounds like you have both a Mamiya system, as well as some other system that accepts a Schneider 38 Super Angulon. So I'm looking for any opinions you have between Mamiya glass and Schneider glass, in terms of sharpness, micro-contrast, color rendition, bokeh, and anything else you can think of.

Mort.

P.S. I am trying to arrange a comparison shoot with my dealer, but gettting hold of a Cambo plus Schneider demo system has been a challenge. So while I'm waiting for that to happen (if it ever does), I'm looking for opinions.

P.P.S. One of the things that turns me off about the wide Schnieders, compared to the Rodenstock HRs, is the fact that the Scheiders seem to need a center filter to handle the vignetting. Do you find that the center filter is needed, if you aren't using shift? I'm guessing that the MFDB is only "seeing" the center portion of the image circle, and that most of the vignetting is occuring outside of this center portion. Would love some confirmation on my guess.

[Jack Flesher]

Don't have much to add, just pretty much agree with what's been said. I shoot the P45+ on the Mamiya, and with my glass I see maybe a slight amount of detail loss going from f8 to f11, but the added DoF usually more than makes up for that.  I would agree that f16 shows a more significant falloff, so I only use that when I really need the added DoF -- kind of an emergency aperture.  

Re the 35XL v 38SA.  I have a Horseman with the Rodenstock 35 HR digital.  I can tell you that lens is stunning at f8, f11 and f16, appearing hardly any different in resolution at all at any of those apertures, and only shows a modest loss of detail at f22 -- still very usable.  Diffraction kicks in bigtime and it's visibly softer at f32 though, so I don't use it beyond f22 and usually hover in the f8 to f16 range most of the time.

As far as micro-contrast differences between Mamiya and this Rodenstock lens, frankly there is not much difference. The Mamiya glass I use is surprisingly good, with very good micro contrast and not overly harsh global contrast.  Resolution wise, the 55, 80 and 150 Mamiya AF lenses all are laser sharp in the f5.6-f11 range, and quite usably sharp wide open.  Color, contrast and overall image character appear very similar between the Rodenstock 35 and the aforementioned Mamiya glass.    

Cheers,

[juicy]

The recent article at this site suggests that at f/11 in 36x48mm format, diffraction limits resolution in mid-spectrum green to about that of a 31MP sensor, and by f/16 the limit is 28MP for blue, less for longer wavelengths. See Table 3 of http://luminous-landscape.com/tutorials/resolution.shtml

If so, about f/10 would be the upper limit for getting maximum possible resolution from such a sensor.
To users of 39MP sensors:

- How well does this theory match your experience?

- What f-stop limits are you setting on yourself when you wish to avoid detectable loss of sharpness to diffraction?

- Or are you perhaps not worrying about squeezing every last pixel's worth of detail out of your 39  million of them, instead letting the res. fall where it may once DOF is adequately controlled?
[{POST_SNAPBACK}][/a]

This was an interesting comparison by Michael using a P45 and 180mm f/5.6 Rodenstock APO Sironar HR with Linhof:
[a href=\"http://www.luminous-landscape.com/tutorials/understanding-series/u-diffraction.shtml]http://www.luminous-landscape.com/tutorial...ffraction.shtml[/url]

Cheers,
J

[Jack Flesher]

Jim Collum posted some aperture examples in this thread from a 55 Rodenstock Digital.  The differences in overall performance and diffraction loss at various apertures are interesting: http://forum.getdpi.com/forum/showthread.php?t=2457

[jonstewart]

Hi Jon. This is a little off-topic. I've been trying to talk myself into supplementing or replacing my Mamiya AFD II kit with something like an Alpa 12TC. My goal is to put the best possible glass in front of my P45+ for landscape shooting. It sounds like you have both a Mamiya system, as well as some other system that accepts a Schneider 38 Super Angulon. So I'm looking for any opinions you have between Mamiya glass and Schneider glass, in terms of sharpness, micro-contrast, color rendition, bokeh, and anything else you can think of.

Hi Mort, have a AFDII with primes and 55-110, and finally decided on a Silvestri Bicam, so I have shift camera, dare I say like the XY (!), that can can simultaneously move in both directions; rise/fall is the standard camera movement, and shift is with the universal stitching back), plus, stick on the maxi flexi bellows and use longer lenses in bayonet mount as a compact view camera.


Found the 35 image circle wasn't big enough for the simultaneous movements on this camera. Image quality seemed little different. Had the Rodenstock Sironon (sp) 45 and IQ was'nt much different (but that was very informal comparison)

P.P.S. One of the things that turns me off about the wide Schnieders, compared to the Rodenstock HRs, is the fact that the Scheiders seem to need a center filter to handle the vignetting. Do you find that the center filter is needed, if you aren't using shift? I'm guessing that the MFDB is only "seeing" the center portion of the image circle, and that most of the vignetting is occuring outside of this center portion. Would love some confirmation on my guess.
[a href=\"index.php?act=findpost&pid=205499\"][{POST_SNAPBACK}][/a]

My testing indicated that the 35 did require the CF for best results, but others have reported correcting in in PP (which wouldn't necessarily be my choice!) This is caused by falloff toward the edge; not sure this is exactly the same as vignetting? In any case, the IQ at centre and edge is not much different - must take time to reexamine that assertion.

As I think I posted, the 35 clearly hits the edge (wouldn't even describe it as vignetting, too hard!) when you rise or fall over about 10mm while shifted left or right about 16mm (that is the left and right positions on the stitching back)

With the sensor in portrait orientation on the Bicam, you have a total of about 70mm width used (ie the diameter of the image circle you can see), so even on rise or fall 15, at full width, vignetting is not noticeable at f11 and particularly at f16 with the 38sa, so I think your assertion is correct. If you have a bigger image circle, you're not going near enough the edges for vignetting to be noticeable.

I have my own 38 and CF coming (I had a loaner), and I'll be happy (as part of my testing anyway) to take and up you some comparison shots (as P45 Raw files, then you can judge for yourself). If I don't PM you within the week, PM me!

I'm very happy with the Bicam; While not being a Alpa, it offers a lot for the price.

[Mort54]

If I don't PM you within the week, PM me!

Thanks Jon. I'll keep my eyes peeled.

[BJL]

Thanks for all the responses, and a correction: my subtitle should have said that f/10 [not f/8] is the predicted threshold for diffraction to start reducing resolution with the 6.8 micron pixels of the 39MP Kodak sensors.

And the data provided in this thread fits this fairly well: the fall off starting between f/8 and f/11, far more noticeable from f/11 to f/16, but of course with f/16 perfectly usable when DOF (sharpness away from the focal plane) outweighs optimal sharpness at the focal plane.

By the way, this is also what Olympus said years ago, recommending maximum f-stop of f/11 with the E-1 and its 6.8 micron pixel Kodak sensor, with S and P modes not going beyond f/11.


So it should be about f/8 and below for optimal sharpness with the 6 micron pixel spacing of the new Kodak 50MP sensor, along with sharp enough lenses.

If so, one question is how often does f/8 give adequate DOF for your 48x36mm format work? (Because when DOF needs are greater, the upgrade from 39MP might give little or no resolution advantage.)

[Ray]

One comment that I think should be made in respect of this topic is that MF lenses traditionally have had lower resolution at F5.6 and F8 than their 35mm counterparts.

This might have changed with the new breed of Rodenstock and Digitar lenses designed for DBs.

If an MF lens with high resolution sensor such as the P45 or Aptus 75 appears to be virtually as sharp at F16 as at F8, then that's a fair indication that the lens used is not particularly sharp at F8.

The 5D has a pixel pitch roughly the same as the P25; marginally greater in fact. The first real world shots I compared at F16 and F8 were taken at 70mm using the Canon 24-105/F4, which is a reasonably sharp zoom but not the sharpest.

I was surprised I could discern no resolution difference at all in the plane of focus, that mattered, between the F8 shot and the F16 shot of the same scene. However, when I later compared my 50/1.8 with the 24-105 at 50mm and F8, the 50/1.8 was noticeably sharper, but not by a great deal.

[yaya]

One comment that I think should be made in respect of this topic is that MF lenses traditionally have had lower resolution at F5.6 and F8 than their 35mm counterparts.

This might have changed with the new breed of Rodenstock and Digitar lenses designed for DBs.

If an MF lens with high resolution sensor such as the P45 or Aptus 75 appears to be virtually as sharp at F16 as at F8, then that's a fair indication that the lens used is not particularly sharp at F8.

The 5D has a pixel pitch roughly the same as the P25; marginally greater in fact. The first real world shots I compared at F16 and F8 were taken at 70mm using the Canon 24-105/F4, which is a reasonably sharp zoom but not the sharpest.

I was surprised I could discern no resolution difference at all in the plane of focus, that mattered, between the F8 shot and the F16 shot of the same scene. However, when I later compared my 50/1.8 with the 24-105 at 50mm and F8, the 50/1.8 was noticeably sharper, but not by a great deal.
[a href=\"index.php?act=findpost&pid=206848\"][{POST_SNAPBACK}][/a]

Wait a minute!!!! Does that mean that you have actually tested a P25?