Wonder about the value of High Resolution

Since the frame size is fixed at either DX or FX, the number of pixels controls the pixel size. There's a tradeoff of noise and dynamic range when pixels get smaller. Figuring out what's optimum is way above my pay grade, but marketing is trying to sell more-more-more pixels, when you can't often take advantage of them. Lenses are only so good, stabilization doesn't 100% compensate for camera shake and light levels are rarely optimum. Even if we have infinite storage space and lightening fast computers, there will be some number of pixels you probably don't want to exceed.

Right, when the frame size is fixed to 24x36mm (or whatever size), keep on increasing the pixel count is meeting diminishing returns. That is partly why I opted for a Z6 instead of Z7 in late 2018, and why I have some slight interest in the Fuji "medium format" cameras. But for the purpose of viewing on a computer screen, even 24MP on FX maybe an overkill.

 

Several years ago, I thought the pixel count war was already over. Looks like I was wrong.

You can also use a high MP camera at pretty much any res you want to....at any time...;)

Indeed.

But then people will start complaining about having to carry... :-)

I was taught to get all of the quality possible onto a negative and have obsessed on that concept over the years, dabbling into medium and large film formats, and now more and more MP. Like Shun, I don't print much. I think my images did look better when I went from 12MP to 36, but stepping "back" to a 24MP Z6 did not seem like much resolution loss (as most of you know that is about a 20% loss of resolution) My neighbor wanted my D810, so it is gone. I did stumble across a deal I could not refuse on a Z7, but don't see huge resolution differences on my screen at 100% Z6 vs scaled down to match Z7. Maybe I should look at a large print from each, but for now I would not fret over which camera is in hand.

 

What I think I do get from the Z cameras is more accurate focus consistency vs the D810. I will take perfect focus at lower resolution over many more MP any day. Still have a D3s & D3x, and D500 for shooting action but not much of that going on with Covid around.

Right, when the frame size is fixed to 24x36mm (or whatever size), keep on increasing the pixel count is meeting diminishing returns. That is partly why I opted for a Z6 instead of Z7 in late 2018, and why I have some slight interest in the Fuji "medium format" cameras. But for the purpose of viewing on a computer screen, even 24MP on FX maybe an overkill.

 

Several years ago, I thought the pixel count war was already over. Looks like I was wrong.

I never noticed much of that war, because i don't care.

 

Re sensor size: even at the same pixelcount, using different (size) sensors can make sense.

there will be some number of pixels you probably don't want to exceed.

Co-incidentally, I got around to wondering if my 60 Mp Sony a7Riv's ability to pixel-shift its way to 240 Mp was a match for 5"x4" film at equal ISO speeds. I still haven't got around to developing the sheets of film yet, but on the way, came to the conclusion that it might all be irrelevant.

 

To ensure a level playing field, I ran the numbers to get equal depth-of-field, same field of view, etc. And as a sideline, got the spreadsheet to calculate the diffraction-limited resolution.

 

It turns out that if everything else is made equal, then the diffraction limit scales exactly with format. In short, in the end it really doesn't matter what film/format size you choose. Because if all else is 'perfect', then diffraction will still getcha and limit resolution to whatever aperture is chosen, or forced upon you.

 

For example, an otherwise perfect 50mm lens @ f/5.6 on the 24x36mm format limits you to an MTF of around 135 cycles/mm. That's 3240 cycles over the 24mm frame-height. Or needing a minimum of 6480 pixels to capture it.

 

Scale up to 5"x4" (frame height ~ 95mm) and you have to use in the region of f/22 to get the same depth-of-field. That knocks your diffraction limit down to about 34 cycles/mm. Making the cycles/frame height 3230. Identical to the 35mm format for all practical purposes.

 

The take away is that; unless your glassware is incredibly exotic, and you're happy with a very thin depth-of-field, then there's very little point in having more than 6,500 x 9,750 pixels to play with. That's unless Mr Airy was a long way out with his calculations!

 

OK, that's still around 64 megapixels, but a long way from the hundreds of Mp claimed for some sheetfilm, or even medium-format, drum scans of yesteryear.

 

And yes, I agree that for most jobs, somewhere in the region of 20 to 24 megapixels is ample. But it's nice to have something in reserve.... just in case.

Edited January 28, 2021 by rodeo_joe|1

Roughly gauging the difference in prices of the old 20 megapixel models and the new 50 MP models, I'd say the value of high resolution is roughly US $3000. I can work it out more finely if you want.;)

It turns out that if everything else is made equal, then the diffraction limit scales exactly with format.

And then someone invented software based focus stacking to do away with diffraction limiting, forever.....:D

 

However, even minutely moving objects or photographers need not apply for this treatment....;)

And then someone invented software based focus stacking to do away with diffraction limiting, forever.....:D

Ah, ah!

No amount of focus-stacking does away with diffraction.

 

If you could eliminate diffraction, there'd be a lot of deleriously happy astronomers and microscopists out there.

Coming from the 10-12 Mpx era (D200-D300) that Nikon got stuck in for quite some time.. I did a LOT of landscape stitching with those cameras in 2011-2012 during an Africa trip and became addicted to high-res images.

After that I bought a D800 in 2012 and I'm still in love with it. Even stitching landscapes. High-res is simply sensational and goes beyond a simple 'need/enough' reasoning. I'm in awe and thankful for that .. and almost feel guilty that so many images I make these days are done with a mediocre smartphone..

I only have a Nikon D600 24MP and with my Epson R2880 used cos that is the only way to print economically and esp I am not in the USA so ink and paper are 2x to you guys. Anyway. With my camera printed at 13x19 it's so dang good so much better than my lab that uses a Fuji Frontier I haven't used their wide format inkjet prints yet and it cost more money. So from what I seen on my own printer I don't see anymore of a need. I guess cropping but that is largely posted online for myself I don't print with that. From the 1 or 2 wedding pros as 100% of their income I know in person they shoot with 24MP dSLRs.

Why did you buy one then - and not the Z6?

Good question. I have both. As mentioned in my initial post, this issue became troubling when I realized how good the current Topaz Gigapixel AI is. It is becoming better and better - not so much a year or two ago. I enlarged a 400x600 @72ppi jpg to 300dpi fit for print on a magazine-size page. Just incredible. So, on the few occasions that we may need a huge print, it would make sense to use this extrapolating software rather than shooting and storing thousands of 46mp files that may never see a printer. Also mentioned is the fact that I had to scale down a Z7 file to print on a church calendar. So it seems the huge files are normally just sitting there taking up space.

 

That said, I am still torn on whether to go ahead with the Z7II upgrade.

What I think I do get from the Z cameras is more accurate focus consistency vs the D810

I am not too sure about that based on a few YouTube reviews I saw. Z has other advantages of course. I have made my decision to go Z thinking to just deal with it, whatever. One cannot have everything. ;)

Ah, ah!

No amount of focus-stacking does away with diffraction.

 

It doesn't do away with diffraction, but focus stacking can allow you to use a larger aperture with shallower DOF and less diffraction.

There's a tradeoff of noise and dynamic range when pixels get smaller

That's incorrect. Noise and DR are mainly a function of exposure and sensor area, so if you have a large sensor that is well exposed it doesn't matter much if it's 12 MP or 46 MP. In fact, often a higher pixel density sensor has more DR than a lower pixel density sensor -- for instance, see:

 

https://www.photonstophotos.net/Charts/PDR.htm#Nikon D4S(DX),Nikon D5(DX),Nikon D500

 

Cropped to the same sensor area the D4s and D5 both lose out to the D500, and even using the entire FX frame the D500 has less noise and more DR at base ISO than these two cameras with their larger photosites.

Focus stacking is amazing ... It takes a bit of time and you really need to clean your subject (if it's close by) because you'll see every spec of dust and dirt. One of the first things I did when I got the D850. Haven't used it much since really.

Here are a few eagles for Shun :-) Shot with the D850 and a 70-200 VRII.

 

 

It doesn't do away with diffraction, but focus stacking can allow you to use a larger aperture with shallower DOF and less diffraction.

Indeed it does! (allow the use of the 'best' aperture regardless of DoF)

 

My normal procedure is to use Focal to determine the sharpest aperture of a macro lens at, say, 1:1 and use that. It's often very similar to the molehill shaped res:aperture graphs published by Lenstip, like this one.

 

Sigma 150 mm f/2.8 APO EX DG OS HSM Macro review - Image resolution - LensTip.com

 

Microscopists have been using Z stage slicing/imaging (and subsequent stacking) for ages to reduce diffraction and increase DoF, but yes the elimination of it is, AFAIK, impossible.

Those considerations of mtf and diffraction Rodeo presented do not make sense. MTF, for instance, says not a lot about what resolution you can achieve, but how much of the source contrast is reduced at whatever resolution you want to consider. Yes, when the outcome drops to nearly nothing, you will not see much of detail at the spatial frequency you're considering.

Et cetera.

 

Most of all, it goes against all experience to say that 35 mm or 4x5" are equal, unless you use exceptional lenses. They are only equal if you use exceptionally bad lenses.

It's just (another) load of humbug.

Edited January 29, 2021 by q.g._de_bakker

What I assume RJ is referring to is that when you have a perfect, diffraction-limited lens, or a lens at an aperture which is only limited by diffraction and not aberrations, and compare with a different format (sensor size), focal lengths set to equal field of view, and set the apertures so that you get the same depth of field, then the effects due to diffraction on sharpness are similar at the image level across the two formats. This assumes there are enough pixels in both cases so that it is not a limiting factor ie. photosite spacing is small compared to diffraction-elicited blur.

 

This means in practice that if you want deep DOF in a single exposure, e.g. MFT sensor you can get similar results as with FF assuming that both have the the same number of pixels. But to get truly equal image qualitt, we also have to assume that we are going for the same exposure time and the FF camera is set to two stops higher ISO and two stops smaller aperture (for fairness, so that blur due to wind or subject movement is similar).

 

I have done some photogrammetry using MFT cameras and Leica Panasonic 25 mm lens and indeed the detail at small apertures is very impressive and at the time Nikon DX cameras didn't produce as good results in my recollection (the MFT camera and especially the lens were newer and more expensive though). Larger formats pull ahead if they have (1) a lot more pixels, (2) if one is willing to use longer exposures or push through more light with flash to use lower ISO such as 64, or (3) if one wants shallow depth of field, in which case it is typical that aberrations kick in and give the advantage to the larger format.

 

I tend to shoot at two extremes: either wide open or stopped down quite deep and only a small number of photos in between. FX does give the advantage in both cases because I have access to higher pixel count, lower ISO, tilt/shift lenses to further expand or manipulate the plane of sharpness, and aberrations per image height are typically lower and wider aperture lenses are now widely available. But if my objective was to maximize depth of field without the use of focus stacking or tilt, and if I put a premium on portability, MFT is very appealing. For low light photos of moving subjects, by contrast, it doesn't do well.

It doesn't do away with diffraction, but focus stacking can allow you to use a larger aperture with shallower DOF and less diffraction.

OK. I see where Mike was coming from now.

 

That only works with static subjects though. Almost as limiting as pixel-shift.

 

My point was that when limited to a single shot and once in the realm of diffraction limiting, it doesn't matter what format size you use, or how many pixels you have - beyond a certain number. That's if you want the picture to look exactly the same from format to format.

 

There's also the issue of lens quality, and I suspect that not many lenses give diffraction-limited resolution at apertures much wider than f/4.

It's just (another) load of humbug.

Yes it is. From you!

But to get truly equal image qualitt, we also have to assume that we are going for the same exposure time and the FF camera is set to two stops higher ISO and two stops smaller aperture (for fairness, so that blur due to wind or subject movement is similar).

Isn't that the wrong way round? To keep the exposure times identical the ISO has to be increased in line with format size. Thereby levelling the playing field even more.

 

Edit. Oh right! MFT is micro four-thirds. I've never used a sensor that piddling. I thought MFT was referring to medium format.

Edited January 29, 2021 by rodeo_joe|1

Think 20 megapixels or thereabout should be plenty.

When it comes to who "needs" what, who makes that decision?

 

The D800 has plenty of resolution, were it not for an AA filter, which cuts it by about 1/3rd, and arriving at a time dominated by SLR lenses designed for film and 12 MP digital camera. The only Nikon lens from that era which can (almost) resolve at the pixel level is the 55/2.8 Micro. High resolution shows details in a print below what you resolve without a magnifier. These details include those of skin and fabrics in portraits, plants and insects, and textures in general which add a sense of reality.

 

Regarding printers, I think Epson has the best software, and piezoelectric heads can handle thicker inks. I switched to Canon because the heads can be easily replaces, whereas Epson printers must be returned to the factory.

When it comes to who "needs" what, who makes that decision?

 

The D800 has plenty of resolution,

The D800 has plenty of resolution for you, which is one answer to the question you posed. For me I'm going to go with 72 MP, and in the article by Thom Hogan on this topic that I link to below he writes that "...for most full frame purchasers [it] is almost certainly the 24mp cameras and a modern convenience lens (e.g. 24-105mm f/4 for Sony FE, 24-70mm f/4 for Nikon Z)."

were it not for an AA filter, which cuts it by about 1/3rd,

Where are you getting that number? At what aperture? Acuity and resolution are not the same. I've seen lots of detail that has low micro-contrast and needs additional sharpening to give it some snap. I'm sure there is a resolution number attached to the AA filter, but I doubt it ever reaches 33% at any aperture, and by the time I'm working at f/11 I suspect it amounts to zero.

 

Another consideration is color moiré, which is triggered by the lens resolving more than the sensor, and which is why Nikon put a relatively weak AA filter on the D800.

and arriving at a time dominated by SLR lenses designed for film and 12 MP digital camera. The only Nikon lens from that era which can (almost) resolve at the pixel level is the 55/2.8 Micro. High resolution shows details in a print below what you resolve without a magnifier. These details include those of skin and fabrics in portraits, plants and insects, and textures in general which add a sense of reality.

Well, I will direct you back to what I wrote above about color moiré and to this article by Thom Hogan on the subject:

 

http://www.sansmirror.com/newsviews/2019-mirrorless-camera/july-september-2019-mirrorl/more-sampling-is-always.html

 

I think Thom's article sums up this discussion rather well.

 

Also see Roger Cicala's response here to Marianna Oelund:

 

https://www.dpreview.com/forums/post/60254934

 

My take on this is in the center of the lens' image circle at wide apertures there are a few lenses made before the D800/D800E came out that can take advantage of at least some of that level of sensor resolution.

Where are you getting that number? At what aperture? Acuity and resolution are not the same. I've seen lots of detail that has low micro-contrast and needs additional sharpening to give it some snap. I'm sure there is a resolution number attached to the AA filter, but I doubt it ever reaches 33% at any aperture, and by the time I'm working at f/11 I suspect it amounts to zero.

Mostly from experience. For example, a Leica M9 with 18 MP is sharper than a Sony A7iii with 24 MP, the ratio being roughly 1:1.5 (or 0.67:1).

 

Sharpening improves acuity, but not necessarily resolution. The D800E used two AA filters, the second to somewhat counteract the first, much like sharpening in POST. Nothing really helps those legacy lenses in comparison to modern lenses designed for Sony or other MILC's.

The D800 has plenty of resolution, were it not for an AA filter, which cuts it by about 1/3rd, and arriving at a time dominated by SLR lenses designed for film and 12 MP digital camera. The only Nikon lens from that era which can (almost) resolve at the pixel level is the 55/2.8 Micro.

 

Basically all lenses show improved detail on this sensor compared to its predecessors which had lower resolutions. I have never been in a situation with my many Nikon and other lenses where I would have felt the sensor resolution to be wasted when the lens is correctly focused. With the D810 (no AA filter, 36MP) it is easy to get moire which is a sign that the sensor has insufficient resolution to properly render images formed by the lens. And I tend to stay away from the sweet spot of the lens but still I could get moire.

 

What limits the quality of D800 images the most is not the lens but the relatively shoddy viewfinder AF which I had a lot of problems with. That too was largely resolved by the succeeding D810. Another major improvement is the EFCS in the D810 and especially D850.

 

To completely get rid of aliasing and to resolve all the detail rendered by the lens, the sensor needs to be in the hundreds or megapixels, or even in the gigapixel range for the sharpest lenses.

 

The 55 micro was decades old at the time of launch of the D800. It's not from "that era". Perhaps you didn't try the camera with any contemporary Nikkor.

Edited January 29, 2021 by ilkka_nissila