Ad
Ad
Ad
Pages: « 1 2 3 [4] 5 »   Bottom of Page
Print
Author Topic: D800 and E... poor for landscape use?  (Read 13996 times)
dimapant
Newbie
*
Offline Offline

Posts: 27


WWW
« Reply #60 on: July 02, 2012, 11:20:07 AM »
ReplyReply

I have D 700 and D 7000 and with a friend, having D 800, we went together taking pictures in Santa Luce, Tuscany, Italy, in a very nice morning, nothing stretching the ISO or dinamica range capability of both cameras, just a very normal day with very normal light .

The reason of that joint taking of pictures was that I am an amateur photographer, taking generic pictures, including landscapes, the D 700 is not certainly the best camera for landscape (low resolution) and I WAS, not anymore, willing to purchase the D 800 and I was asking him to let my try the camera.

Myself took around 100 shots with that D 800, and during that time, I changed the same my lenses from D 800 to D 700, so taking the same shots, at distance of just minutes, same light, same everything, including the lenses, my  14 – 24 mm, my 70-200 VRII and the 24 -70 on both cameras.

I mostly print in A3 and A3+, and I did all the work, from camera to print, on 6 of those D 800 shots, and 3 on D 700, printing both, made with D 700 and D800, quite similar picture on my Epson 3880, small printing size, A3 (around 30 x 40 cm).

I am taking picture and printing myself since 42 years, I am an amateur photographer, on digital cameras from 4 years, expert in digital postproduction, applying proper Capture Sharpening in ACR, Artistic Sharpening as Mr. B. Frazer dictate in his “Real World image sharpening”, with limitation at the extremes of the range, on black and highlights, and applying Print Sharpening, together with proper interpolation, including grain, with Alien Skin Blow UP  and the expert people watching at my prints, all of them, they say that the quality of my printing is quite high and quite OK.

Myself and  others, we basically cannot detect any difference,  with naked eye,  from prints on shots taken by D 700 and D 800, basically the same picture, on A3, observed them  from half meter distance, whilst if you go and check with a magnifier or check carefully at short distance, around 20 cm  and you have good eyes, some more very tiny details can be seen on D 800, but the entire picture, in a real view, look pretty the same.

But……there a “but”, and, in my very personal opinion, it is a very important “but”.

The Nikon zoom lenses that I have tested on D 800, 14-24, 24-70, 70-200 VRII, all of them, unless you close well the diaphragm at around f 6.3 – f 8,  they are not very exciting  on that camera ,not at all, the borders are weak with 14 -24 and 24 -70, better with 70 -200, and with that camera, you can detect a minimum misalignment of a zoom, which is almost normal in a zoom and which cannot basically detected on D 700.

Moreover, if you stop very down the lenses at f 16 – f 20, which is normal for picture with wide angles  for a very deep   DOF, a with the subject in the very closed distance and the landscape in the back, and this pictures are very  normal in landscape shooting, the pictures taken with D 800 is not so sharp like the pictures taken by D 700, and already in 30X 40 cm printing you can see this, no gain at all in respect of D 700, if not a loss.

On this prints, with deep DOF with D 800 and D 700, Details of ACR and Smart Sharpen in PP where working in deconvolution.

I am working mostly with zoom, after almost 40 years with primes on film, I went on zoom on digital: I will not change back to primes for a camera to shot at large apertures, whilst if you close the diaphragm around f 16 or more, the picture will be not better  than the picture taken with D 700 on that relatively small A3 print size.

D 800 could certainly be better in large formats printing, it looks even better on monitor, but monitor is monitor and printing is printing: on A3 it looks pretty  like the D 700, and at high number of F stop even not at the same level of D 700.

This is my modest and limited in number of prints experience with D 800 + 14 -24 , 24 -70 and 70 -200 VR II,  just my experience, and I will not shift from D 700 to D 800 as the majority of my prints are in A3.

Best regards

Alessandro
Logged
free1000
Sr. Member
****
Offline Offline

Posts: 397


WWW
« Reply #61 on: July 02, 2012, 02:36:49 PM »
ReplyReply


Myself and  others, we basically cannot detect any difference,  with naked eye,  from prints on shots taken by D 700 and D 800, basically the same picture, on A3, observed them  from half meter distance, whilst if you go and check with a magnifier or check carefully at short distance, around 20 cm  and you have good eyes, some more very tiny details can be seen on D 800, but the entire picture, in a real view, look pretty the same.

This is what I was expecting at that kind of print size. 

Quote
But……there a “but”, and, in my very personal opinion, it is a very important “but”.

The Nikon zoom lenses that I have tested on D 800, 14-24, 24-70, 70-200 VRII, all of them, unless you close well the diaphragm at around f 6.3 – f 8,  they are not very exciting  on that camera ,not at all, the borders are weak with 14 -24 and 24 -70, better with 70 -200, and with that camera, you can detect a minimum misalignment of a zoom, which is almost normal in a zoom and which cannot basically detected on D 700.

Moreover, if you stop very down the lenses at f 16 – f 20, which is normal for picture with wide angles  for a very deep   DOF, a with the subject in the very closed distance and the landscape in the back, and this pictures are very  normal in landscape shooting, the pictures taken with D 800 is not so sharp like the pictures taken by D 700, and already in 30X 40 cm printing you can see this, no gain at all in respect of D 700, if not a loss.

Interesting.  Thanks for sharing.
   
Logged

@foliobook
Foliobook professional photography folio for iPad
www.foliobook.mobi
kers
Sr. Member
****
Offline Offline

Posts: 707


WWW
« Reply #62 on: July 02, 2012, 03:22:11 PM »
ReplyReply

....I am working mostly with zoom, after almost 40 years with primes on film, I went on zoom on digital: I will not change back to primes for a camera to shot at large apertures, whilst if you close the diaphragm around f 16 or more, the picture will be not better  than the picture taken with D 700 on that relatively small A3 print size....

....D 800 could certainly be better in large formats printing, it looks even better on monitor, but monitor is monitor and printing is printing: on A3 it looks pretty  like the D 700, and at high number of F stop even not at the same level of D 700....

Alessandro

Alessandro, thank you for your input..
You are right to say that 12mp is more than enough to cover A3, and since the iso sensitivity is not that much different between the D700 and D800 the only difference is the greater dynamic range of the d800.
Other benefits are and the ability to shoot video, the internal cleaning system and the fact it is much more silent when shot in liveview. I think those are very strong points.
The zooms you refer to were introduced with the d3- 12MP..
In that respect they still do very well on 36MP - especially the 14-24 if you focus it in liveview. Even the extreme corners at d11 look good.
But obviously you need the new Nikkor primes or Zeiss Primes to get the most out of the sensor.. prints at 150 dpi = 124 cm look very good, even close. (So there is also the benefit you can crop without any loss to A3)
At that size if you use d1,4 the contrast between sharp and unsharp is even more delicate - if you are sensitive to it...
Logged

Pieter Kers
www.beeld.nu
Ray
Sr. Member
****
Offline Offline

Posts: 8812


« Reply #63 on: July 02, 2012, 10:14:12 PM »
ReplyReply

Hi Ray,

Not sure where the D3x came from, but attached is the comparison between the D3x and the D800 (once expressed in cy/mm and once in cy/pixel resolution), from an optical point of view. Obviously the D800 has more resolution, but the optical response at the same aperture is of course identical. Only if one were to crop the D800 to the size of the D3x (no idea why one would do that, other than for argument's sake), then the per pixel response would change, but in favor of the larger sensel pitch camera (less diffraction per pixel due to lower resolution).

These are OTFs (diffraction+defocus, CoC blur assumed to be 1.5x sensel pitch), so a perfect camera is assumed. The actual MTF curves could never exceed these, but will be somewhat lower, depending on things like AA-filter and sensel aperture.

This hopefully also demonstrates where the confusion comes from that some people experience, because they fail to translate to output quality related cy/mm resolution.

Cheers,
Bart

Hi Bart,

Now all you need to do is take some real-world shots to see how the results correspond with your theoretical charts.  Grin

There's no doubt in my mind that a lens used at F16 on a 38.4mp full-frame camera with the same filters and same quality pixels as a 50D will produce marginally, but noticeably better detail (when examined in pixel-peeping mode), than the same lens at F16 used on a 25.6mp full-frame camera of the quality of a 40D.

However, such differences are of little practical significance and would only be noticeable on really huge prints, and even then would probably be noticeable only by the experienced eye.

I think I've posted the following image before, showing the 10mp 40D used at f8, compared with the 15mp 50D at F16. The main difference I see here is the obvious color moiré in the 40D shot which is absent from the 50D shot. What seems to be happening here is that any loss in detail from the lens, due to the greater diffraction at F16, has been more or less compensated by the 50% increase in pixel count of the 50D, resulting in approximately equal resolution for both images.

The nett effect is that the F16 shot is better than the F8 shot if good DoF is required. There's no loss of resolution at the plane of focus; image quality is generally better due to the absence of moiré, and DoF is significantly greater. Three cheers for the high-pixel-count sensor!  Grin
Logged
NashvilleMike
Full Member
***
Offline Offline

Posts: 171


« Reply #64 on: July 02, 2012, 11:44:26 PM »
ReplyReply

This is what I was expecting at that kind of print size. 

Interesting.  Thanks for sharing.
   

This brings up an interesting question: at what print size do the advantages of a D800 (or 800E) take hold?

My own experience to add to the data pool is as follows:

I've got a D800E, and previously my main body was a D700. In a past life I also shot 35mm film, medium format film, and 4x5 film, so I have a pretty good idea of what qualities you get from the larger formats. I do not have any personal experience with MFDB.

In a studio session with the D800E, after printing, it was obviously and immediately noticeable to both myself and my model that 13x19" prints from the D800E, printed on Moab Slickrock Metallic on an Epson 3800, were better, at normal viewing distances, than those from the D700. Noticeably so, but not vastly so. However, I'd be shocked if anyone couldn't clearly see the differences at that print size. Fabric textures were quite a bit better defined and more "realistic", while skin and other aspects of the image that didn't contain so much fine detail, the differences weren't as noticeable as one might expect. But there was no doubt which print was better, none at all. If we use a subjective grading/description from "identical / subtle / moderate/ significant/ vast", I'd grade the difference between moderate and significant myself, although of course this is not a technical measurement of the print detail. Interestingly, this was shooting at F/10 and F/11, which is clearly into the so called diffraction zone of the body, and thus not using "all" of the resolution of the body. Lenses were the same 24-70/2.8, 35/1.4G, and 85/1.8G that I used on the D700.

As for landscape: I've also been working through various lens tests on pseudo landscape test scenes, and on test prints made at 17x22" comparing to the D700, the differences were quite significant with this subject matter. But I've not printed at 11x17" (which is what I believe A3 is closest to in the US system of paper sizes). So it's interesting to hear that someone didn't see any differences that mattered to him at this size, yet in my own evaluation (and my models), going one paper size up from that, there were clearly obvious differences to both of us. There is no doubt which body I'll be using primarily and it's not the 700, as I mostly print at 16x20" and secondarily, at 13x19". From an image quality point of view, I'm extremely happy with the D800E files, even more than I expected to be.

So the big question is: where is the line of demarcation in terms of print size, where it's not advantageous to use the D800E? When I get back off the road and have some time, I may need to do a lot more testing to see, as I've not printed anything smaller than 13x19 yet - I'd rather be shooting than testing, even if one has to do the latter to gain knowledge of their gear.

As an additional note: One thing I've discovered during my (ongoing) lens evaluation (which I believe one has to do over a longer term and with multiple passes in different scenarios) is that with the D800E, it's really not enough just to broadly categorize a lens as "good" or "bad". I've found that it's very much more a "this lens is good in this scenario, at this focal length, and at this distance range" as opposed to generally stating "the 14-24 is bad". For example, doing properly controlled testing on my 14-24 at various focal lengths and comparing it to my 24/1.4G prime, I found that a focal length of 24mm and longer subject distances are the lenses weakness; the 24/1.4G here is clearly better in the corners than the 14-24 here. However, for a close interior shot where the subject distance is quite a bit closer, the 14-24 performs much more acceptably (although not excellently) in the corners at 24mm. Distance to subject matters as lenses often are optimized for one range over the other. And then when you get down in the wider part of the lenses range, specifically around 16mm or so, the lens is actually quite spectacular, with very good corners once stopped down. So this lens is either "not so good" (24mm / infinity / corners) or "acceptable" (24mm / 10 feet distance / corners) or "quite good" (16mm, any distance, all of frame). Same thing with the 24-70; while the lens is obviously tweaked for center sharpness over absolute deep corner performance, at 35mm for landscape, it's quite capable of holding it's own with the 35/1.4G at F/7.1 - I had to run the tests several times to confirm this, and in the studio, where the subject matter (for me) is not corner dependent, it's excellent. But the 24-70 at 24mm - at pretty much any distance, in the corners - yuch. (A highly technical term, I know!).  My 70-200 VR2 at 105mm beats every lens I have in house at F/7.1 except my 200/2G at the closer and moderate distance ranges, including the 105/2 DC. So anyone who says the zoom is "bad on a D800E" would be, in this case, incorrect. But at the same time, I haven't run through all the focal lengths on that lens yet as I'm traveling and quite busy at the moment, but I'm sure there is a "weak" combination of focal length and distance on it as well where it won't be excellent like it is at 105mm. The D800E is very much a critic of lenses, telling you the good, the bad, and the proverbial ugly, about each lens and how it performs at differing distances, apertures, as well as focal lengths (if a zoom).

-m

Logged
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1273



WWW
« Reply #65 on: July 03, 2012, 03:26:48 AM »
ReplyReply

Not necessarily, Guillermo.

Yes, because at f/16 the D800 is diffraction limited, and this means a loss in resolution with respect to shooting at wider apertures.
Logged

BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3012


« Reply #66 on: July 03, 2012, 03:30:19 AM »
ReplyReply

Hi Bart,

Now all you need to do is take some real-world shots to see how the results correspond with your theoretical charts.  Grin

I would have, if I had a D800 and a D3x, but I don't. Besides, the 'theoretical' part is just physics. You can be assured that in practice it won't be different unless the experiment was flawed, e.g. with significant defocus error or different input magnification.

Quote
There's no doubt in my mind that a lens used at F16 on a 38.4mp full-frame camera with the same filters and same quality pixels as a 50D will produce marginally, but noticeably better detail (when examined in pixel-peeping mode), than the same lens at F16 used on a 25.6mp full-frame camera of the quality of a 40D.

However, such differences are of little practical significance and would only be noticeable on really huge prints, and even then would probably be noticeable only by the experienced eye.

The message to take home from the charts I showed is still the same.
  • Sensel density determines on sensor resolution.
  • Physical sensor dimensions determine the required output magnification factor.
  • On sensor resolution divided by magnification factor determines output resolution.

Separate from that,

  • Aperture (and wavelength) determines diffraction.
  • Smaller sensels take smaller samples of the same diffraction at a given aperture.
  • Having more sensels allows to use a lower output magnification factor.

It's the output magnification that ultimately determines how much resolution we will end up with, and on the input side that is determined by the sensel density and sensor array dimensions together. Aperture just sets the overall physical resolution limits on the input side, and the subsequent output magnification, well, magnifies that.

Cheers,
Bart
« Last Edit: July 03, 2012, 03:40:40 AM by BartvanderWolf » Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3012


« Reply #67 on: July 03, 2012, 06:10:31 AM »
ReplyReply

This brings up an interesting question: at what print size do the advantages of a D800 (or 800E) take hold?
[...]
Quote
So the big question is: where is the line of demarcation in terms of print size, where it's not advantageous to use the D800E? When I get back off the road and have some time, I may need to do a lot more testing to see, as I've not printed anything smaller than 13x19 yet - I'd rather be shooting than testing, even if one has to do the latter to gain knowledge of their gear.

Hi Mike,

The answer may be easier to find than you think. The D700 has a limiting on sensor resolution (at the Nyquist frequency) of 59.1 cycles/mm. That is determined by it's fat sensels with a 8.46 micron pitch. If we follow the common assumption that 5 cy/mm (254 PPI) represents good, and 8 cy/mm (406 PPI) excellent, output print quality then we can divide the on sensor resolution by the magnification factor that results in such output resolution, i.e. 11.82x or 7.39x.

Multiplying the physical sensor array dimensions (36x23.9mm) by those magnification factors would give 426mm x 282mm (approx. 16.8 x 11.1 inch) output size at good quality, and 266mm x 177mm (approx. 10.5 x 7.0 inch) output size at excellent quality.

The D800E wouldn't do any better (assuming both files were optimally capture sharpened) at those output sizes with comparable postprocessing, since the output resolution would be the same, limited by the output medium. However, the D800 could still achieve the same resolution at an even larger output size, say 76% larger.

The D800E files would allow higher quality creative sharpening because there are more pixels available for the same feature sizes, so it would be possible to produce higher quality with an optimal workflow, but that's due to processing differently, which risks turning it into an apples and oranges comparison. The quality potential for the D800 is higher, so that will obviously show when exploited to the max.

There can be some differences in the overall look of the output if Capture sharpening is not specifically adapted to the respective sensors and apertures used, but that should not be newsworthy. I've illustrated the situation with yet another chart (as attached), but in practice this should not give any different output results than described above. It does show that the OTF will drop off faster with narrower apertures due to diffraction for the D800 sensor, but that can be compensated for with capture sharpening.

Cheers,
Bart
« Last Edit: July 03, 2012, 06:44:00 AM by BartvanderWolf » Logged
free1000
Sr. Member
****
Offline Offline

Posts: 397


WWW
« Reply #68 on: July 03, 2012, 06:56:04 AM »
ReplyReply

In a studio session with the D800E, after printing, it was obviously and immediately noticeable to both myself and my model that 13x19" prints from the D800E, printed on Moab Slickrock Metallic on an Epson 3800, were better, at normal viewing distances, than those from the D700. Noticeably so, but not vastly so. However, I'd be shocked if anyone couldn't clearly see the differences at that print size. Fabric textures were quite a bit better defined and more "realistic", while skin and other aspects of the image that didn't contain so much fine detail, the differences weren't as noticeable as one might expect. But there was no doubt which print was better, none at all. 

I also think that paper choice, and RIP will have quite a big effect on the appearance.  Some papers show much more detail. I don't know your paper choice but it sounds like it might be fairly detailed.  The same prints on a watercolour paper might be less different.  I don't use RIPs myself as I've not really preferred the results in the past, but I know a lot of people use them.
Logged

@foliobook
Foliobook professional photography folio for iPad
www.foliobook.mobi
kers
Sr. Member
****
Offline Offline

Posts: 707


WWW
« Reply #69 on: July 03, 2012, 07:01:55 AM »
ReplyReply

here two 100% crops - shots with the D800e nikkor 24MMPCE for me the sharpest 24mm Nikkor-

there is a pixel peeping better sharpness at d5,6 - i add a tif . a jpeg spoils the test...

« Last Edit: July 03, 2012, 07:09:05 AM by kers » Logged

Pieter Kers
www.beeld.nu
bjanes
Sr. Member
****
Offline Offline

Posts: 2714



« Reply #70 on: July 03, 2012, 07:43:20 AM »
ReplyReply

Hi Bart,

To illustrate some of these points I shot your resolution chart with my 800e and performed measurements using your method with the sinusoidal star chart and Imatest. Rendering was done with ACR 7.1 usisng PV2012 and no sharpening. The 60mm f/2.8 AFS was used with a weighted tripod, live view focusing, and mirror lockup.The optimum aperture for this lens and camera system is about 5.6. The red circle is at a radius of 92 pixels, which is the Nyquist for the sensor (103 cy/mm).









Alaising is prominent at the optimum aperture, but is reduced along with resolution to the smaller apertures.

Here are the results of the resolution by your method. As per a recent discussion, this method appears to measure resolution near the Rayleigh limit. The graph shows the Rayleigh limit for each aperture, and the resolution at 50% MTF for a perfect lens.



At the optimum aperture, the camera resolves to the Rayleigh limit which is MTF 20% according to your calculations, but usually stated to be about 9%. However, the MTF50 resolution is considerably below the diffraction limit, but could be improved with deconvolution sharpening. At f/16 the camera system does resolve near the Rayleigh limit, but MTF50 without any sharpening as measured by Imatest is considerably less but is near the diffraction limit. Sharpening of the f/5.6 image with ACR using a detail of 100% for deconvolution and a radius of 0.8 and amount of 50% regains considerable MTF and resolves at the Nyquist limit, but the MTF at Nyquist is quite high, indicating considerable aliasing. The image is likely oversharpened, but no overshoot that would indicate halos is noted in the edge plot.



Your comments are welcome.

Regards,

Bill



Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8812


« Reply #71 on: July 03, 2012, 09:21:20 AM »
ReplyReply

Yes, because at f/16 the D800 is diffraction limited, and this means a loss in resolution with respect to shooting at wider apertures.


I thought it was only lenses that can be diffraction limited, Guillermo. Camera bodies could be pixel limited though. F22 is the aperture at which diffraction is so great that my 15mp 50D provides no resolution advantage whatsoever compared with my 10mp 40D, using tripod and LiveView of course.

Comparing images from both cameras at F16, the 50D image has very marginally more detail, but noticeable only at 200% magnification on the monitor. Comparing images at increasingly wider apertures, the differences become more obvious.

The pixel densities of the Canon 40D and 50D are very similar to the pixel densities of the D3X and D800 respectively. I would expect that for landscape work, a photographer using a D800 could expect at least the same resolution at F16 as he would get with the D3X at F11, and possibly even at F8.
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8812


« Reply #72 on: July 03, 2012, 09:42:13 AM »
ReplyReply


It's the output magnification that ultimately determines how much resolution we will end up with, and on the input side that is determined by the sensel density and sensor array dimensions together. Aperture just sets the overall physical resolution limits on the input side, and the subsequent output magnification, well, magnifies that.

Cheers,
Bart

Of course. Have we already forgotten Michael's comparison between the Canon G10 and the Phase One P45 at A3+ print size? If one prints no larger than A3+, The D800 is still useful for its cropping potential. One could use it as a 2x cropped-format camera. 9mp is sufficient for an A3+ print. A 300mm lens becomes effectively a 600mm lens when one needs it, yet a 12mm lens still provides the full FoV of a 12mm lens on FF when one needs it.
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3012


« Reply #73 on: July 03, 2012, 10:24:18 AM »
ReplyReply

Hi Bart,

To illustrate some of these points I shot your resolution chart with my 800e and performed measurements using your method with the sinusoidal star chart and Imatest. Rendering was done with ACR 7.1 usisng PV2012 and no sharpening. The 60mm f/2.8 AFS was used with a weighted tripod, live view focusing, and mirror lockup.The optimum aperture for this lens and camera system is about 5.6. The red circle is at a radius of 92 pixels, which is the Nyquist for the sensor (103 cy/mm).

Hi Bill,

Thanks for that test. I don't have a D800(E) otherwise I would have done it, but theory and practice should still be in line.

Quote
Alaising is prominent at the optimum aperture, but is reduced along with resolution to the smaller apertures.

As the theory (based on solid physics) predicts. Besides, the target is a torture test, in practice the aliasing effects can be more benign, and the Raw converters also offer some relief in allowing to suppress the false color demosaicing.

Quote
Here are the results of the resolution by your method. As per a recent discussion, this method appears to measure resolution near the Rayleigh limit. The graph shows the Rayleigh limit for each aperture, and the resolution at 50% MTF for a perfect lens.

A small remark, MTF50 is more of a subjective impression of sharpness metric than a significant resolution metric. It is not tied in with any specific physical phenomenon, just subjective ones (and therefore relevant for side to side comparisons). And yes, Lord Rayleigh specified his limit based on resolution being limited by reduced contrast between 2 signals (point-sources in his scenario), which is exactly what happens at the limiting resolution that my target visualizes. So they should give similar outcomes.

I've attached a zoomed-in screen capture of your f/5.6 image, and added (in yellow) where I would place the limiting resolution boundary, at 99 pixels (91 is too opimistic) or 95 cy/mm (assuming a 4.88 micron sensel pitch). The exact boundary diameter is slightly arbitrary because it also depends on the exact alignment of the pattern with the sensel grid, especially without Optical Low-Pass filtering (OLPF). Inside that 99 pixel diameter I already see a widening of the pattern which indicates aliasing. The aliasing does mimic the underlying pattern somewhat, but it already misrepresenting the input signal, hence 'alias'.

What is also shown in the f/5.6 example, is the reddish cast that Photoshop ACR and Lightroom produce on high resolution/contrast edges, specifically on the D800E version Raws. ACR and LR do that more than other Raw converters (it can be corrected, but still).

Quote

Yes, with the proviso of the slightly larger diameter of the blur circle I would have used, it illustrates nicely what has been stipulated earlier. From f/16 and narrower we will see actual loss of resolution to significantly below Nyquist, and diffraction serving as a pseudo AA-filter. It also reduces overall contrast, but that can be somewhat restored by deconvolution sharpening. The resolution that was lost cannot be restored.

Quote
Sharpening of the f/5.6 image with ACR using a detail of 100% for deconvolution and a radius of 0.8 and amount of 50% regains considerable MTF and resolves at the Nyquist limit, but the MTF at Nyquist is quite high, indicating considerable aliasing. The image is likely oversharpened, but no overshoot that would indicate halos is noted in the edge plot.


Yes, but even with a somewhat lower 'Detail' setting (to avoid jaggies) and lower amount, an f/5.6 image should be able to be sharpened close to perfection.

A quick analysis of the f/5.6 example with my optimal Capture sharpening analysis tool suggests a slightly larger sharpening radius (which might indicate that focus could be improved), but I would prefer a somewhat lower brightness unsharpened rendering (preferably after some defringing) to base such  an analysis on.

Your example at least shows that with sharpening one can reproduce a very high MTF output that introduces few artifacts. It just takes some precautions to remove/reduce the artifacts that were part of the capture and Raw conversion process before sharpening to get the best results.

Cheers,
Bart
Logged
NashvilleMike
Full Member
***
Offline Offline

Posts: 171


« Reply #74 on: July 03, 2012, 10:52:47 AM »
ReplyReply

I also think that paper choice, and RIP will have quite a big effect on the appearance.  Some papers show much more detail. I don't know your paper choice but it sounds like it might be fairly detailed.  The same prints on a watercolour paper might be less different.  I don't use RIPs myself as I've not really preferred the results in the past, but I know a lot of people use them.

You're absolutely right. As a note, I print on 4 media (actually mostly the first 3): Epson Exhibition Fiber (color landscape), Canon Infinity Baryta (personal B&W), Moab Slickrock Metallic (some color dependent on subject, some B&W dependent on subject, often nudes and figure work), and for rough testing, your basic Epson Premium Glossy. I don't use a RIP.

Another somewhat related note: A few years ago I saw, at an Epson booth, a 16x20" print from Douglas Dubler reportedly shot on a Leaf MFDB. A fashion shot. It was quite possibly the finest print of a fashion subject I've ever seen, with natural skintones and tone transitions, realistic yet unforced detail; simply put, I was jealous. With the D800E I'm getting a whole hell of a lot closer to that level of quality than I ever had with DSLR, and that to me is a breakthrough as the MFDB aren't financially feasible for me at this time and my shooting style is more attuned towards the smaller format cameras.

-m
Logged
theguywitha645d
Sr. Member
****
Offline Offline

Posts: 970


« Reply #75 on: July 03, 2012, 11:25:52 AM »
ReplyReply

The answer may be easier to find than you think. The D700 has a limiting on sensor resolution (at the Nyquist frequency) of 59.1 cycles/mm. That is determined by it's fat sensels with a 8.46 micron pitch. If we follow the common assumption that 5 cy/mm (254 PPI) represents good, and 8 cy/mm (406 PPI) excellent, output print quality then we can divide the on sensor resolution by the magnification factor that results in such output resolution, i.e. 11.82x or 7.39x.

Multiplying the physical sensor array dimensions (36x23.9mm) by those magnification factors would give 426mm x 282mm (approx. 16.8 x 11.1 inch) output size at good quality, and 266mm x 177mm (approx. 10.5 x 7.0 inch) output size at excellent quality.

Why are you assuming a fixed viewing distance? Basing this on a fixed frame to describe image quality does not work; photography is subjective. That is like saying an APS-C sensor has more detail than a 35mm one because it is working at higher resolving power.
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3012


« Reply #76 on: July 03, 2012, 12:28:26 PM »
ReplyReply

Why are you assuming a fixed viewing distance?

Hi,

That's not what I did. I calculated equal output size for both images, magnified to the point where further magnification would lose output resolution on the D700, while the D800 had more to spare.

That should answer Mike's question:
Quote
So the big question is: where is the line of demarcation in terms of print size, where it's not advantageous to use the D800E?

It is not advantageous below the output sizes that I calculated.

Quote
Basing this on a fixed frame to describe image quality does not work; photography is subjective. That is like saying an APS-C sensor has more detail than a 35mm one because it is working at higher resolving power.

An APS sensor does have a higher on-sensor resolution, but a smaller size. When magnified to the same output size, the APS image usually requires more magnification. Only when the number of additional pixels compensates the increased magnification will we reach the 'line of demarcation', just as in my calculation but that was for equal sensor array sizes.

Now, viewing distance is only relevant in that the visual resolution of both images will be impacted in exactly the same way.

Cheers,
Bart
« Last Edit: July 06, 2012, 02:10:24 AM by BartvanderWolf » Logged
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1273



WWW
« Reply #77 on: July 03, 2012, 01:02:19 PM »
ReplyReply

I thought it was only lenses that can be diffraction limited, Guillermo. Camera bodies could be pixel limited though.
Not interested in your semantic discussions Ray. The point is that at f/16 the D800 cannot reach its maximum effective resolution, and that means you cannot enjoy its full 36Mpx because of diffraction (among other possible factors).
Logged

free1000
Sr. Member
****
Offline Offline

Posts: 397


WWW
« Reply #78 on: July 03, 2012, 01:15:09 PM »
ReplyReply


Another somewhat related note: A few years ago I saw, at an Epson booth, a 16x20" print from Douglas Dubler reportedly shot on a Leaf MFDB. A fashion shot. It was quite possibly the finest print of a fashion subject I've ever seen, with natural skintones and tone transitions, realistic yet unforced detail; simply put, I was jealous. With the D800E I'm getting a whole hell of a lot closer to that level of quality than I ever had with DSLR, and that to me is a breakthrough as the MFDB aren't financially feasible for me at this time and my shooting style is more attuned towards the smaller format cameras.

Its rapidly getting to the point where MF at that level will be a similar price point if you get second hand gear, the convenience of the Nikon could well beat it into second place for many purposes though. The main reason I'm going down the D800E route is that my Aptus 75/Mamiya combination doesn't work with long exposures or in a myriad of other situations that can crop up. My limited testing (darn it I'm too busy at the moment to test as I'd like to) shows that in terms of resolution the D800E is very close. I don't think the colour and DR are as good, but that is very subjective on my part at this point, it might be I just need to learn how to get the best out of the D800E.

Logged

@foliobook
Foliobook professional photography folio for iPad
www.foliobook.mobi
bjanes
Sr. Member
****
Offline Offline

Posts: 2714



« Reply #79 on: July 03, 2012, 01:52:27 PM »
ReplyReply


I've attached a zoomed-in screen capture of your f/5.6 image, and added (in yellow) where I would place the limiting resolution boundary, at 99 pixels (91 is too opimistic) or 95 cy/mm (assuming a 4.88 micron sensel pitch). The exact boundary diameter is slightly arbitrary because it also depends on the exact alignment of the pattern with the sensel grid, especially without Optical Low-Pass filtering (OLPF). Inside that 99 pixel diameter I already see a widening of the pattern which indicates aliasing. The aliasing does mimic the underlying pattern somewhat, but it already misrepresenting the input signal, hence 'alias'.

What is also shown in the f/5.6 example, is the reddish cast that Photoshop ACR and Lightroom produce on high resolution/contrast edges, specifically on the D800E version Raws. ACR and LR do that more than other Raw converters (it can be corrected, but still).

Yes, with the proviso of the slightly larger diameter of the blur circle I would have used, it illustrates nicely what has been stipulated earlier. From f/16 and narrower we will see actual loss of resolution to significantly below Nyquist, and diffraction serving as a pseudo AA-filter. It also reduces overall contrast, but that can be somewhat restored by deconvolution sharpening. The resolution that was lost cannot be restored.

Yes, but even with a somewhat lower 'Detail' setting (to avoid jaggies) and lower amount, an f/5.6 image should be able to be sharpened close to perfection.

A quick analysis of the f/5.6 example with my optimal Capture sharpening analysis tool suggests a slightly larger sharpening radius (which might indicate that focus could be improved), but I would prefer a somewhat lower brightness unsharpened rendering (preferably after some defringing) to base such  an analysis on.

Bart,

Thanks for your pointers on choosing the proper radius to determine the resolution from your chart. I did notice the reddish case at f/5.6. What causes that and how would it be corrected?

I did use your tool to determine optimal sharpening radii, but did not post the results since the values for f/2.8 and f/4.0 seem too high and I intend to repeat the analysis when I get time. The tool is quite helpful and many thanks for publishing it. Adjusting the radius according to the Gaussian radius is a departure from the conventional sharpening advice according to the Jeff Schewe/Bruce Fraser model where they suggest a small radius for high frequency images such as landscapes and a larger radius for low frequency images such as portraits. There are better deconvolution tools than ACR/LR, but I prefer parametric  editing since 16 bit TIFFs need for a stand alone produce get rather large when one is using the D800. The extra effort of using a standalone tool would be worth the trouble for one's most important images that need to be printed large.



Regards,

Bill
Logged
Pages: « 1 2 3 [4] 5 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad