jwcat
Well-known
http://www.thephoblographer.com/201...omises-unprecedented-resolution/#.UzhnGzClnFL
Can any of our German members comment on this?
Can any of our German members comment on this?
A Y
Member
It's probably a re-labeled version of the Pacific Image Prime Film XA and XE, which have been out for a while.
Plustek has a useful blog post about the actual resolution of their scanners, including their 10k DPI models:
http://plustekusa.blogspot.com/2012/11/film-scanner-resolution-explained-what.html
5k DPI is overkill for most film.
Plustek has a useful blog post about the actual resolution of their scanners, including their 10k DPI models:
http://plustekusa.blogspot.com/2012/11/film-scanner-resolution-explained-what.html
5k DPI is overkill for most film.
edge100
Well-known
10,000 dpi is FAR more than is needed for 35mm film. Even with 4000dpi on my SprintScan 120, I'm resolving grain; there's just not that much left on the film.
I'd hate to scan 10,000 dpi TIFFs, just to get 5,000dpi of information.
Still, any new scanner is a good thing.
I'd hate to scan 10,000 dpi TIFFs, just to get 5,000dpi of information.
Still, any new scanner is a good thing.
Noll
Well-known
Not necessarily, at least IMO. It's overkill in the same way that 36mp is for a DSLR. You can always set it lower as needed. And of course there is always the issue of theoretical vs actual scanner resolution.
I have done a lot of camera-scanning using a macro lens and 16mp 4/3 body. At 1:1 reproduction it is about equal to a 6000 dpi "scan". I find this is just enough for Tri-X while resolving a decent amount of grain "headroom". Much of what we see as grain in sharp scans at lower dpi could just at easily be grain aliasing.
On an iso 100 or slower B&W film, I find there can be an astounding amount of detail present even on 35mm. I have little doubt that 8000 dpi or more could squeeze additional detail from some of my sharpest negatives.
There is nothing wrong with making big, beautiful scans!
(for big beautiful prints)
Joosep
Well-known
I do not understand how this is making the news. Second blogpost this week and it is tuesday.
This is not new tech. As everyone who knows something about scanners, knows that Reflecta=Pacific Image, same scanners for EU/US markets accordingly.
This scanner came out about half a year ago, under the name Pacific Image Primefilm XE. Also a version with bulk scanning capabilities and a better dMax came out, named the XA. Reflecta said they will start selling the EU version of the XA in 2014 Q4.
I have had the XA for a couple of months now.
This is not new tech. As everyone who knows something about scanners, knows that Reflecta=Pacific Image, same scanners for EU/US markets accordingly.
This scanner came out about half a year ago, under the name Pacific Image Primefilm XE. Also a version with bulk scanning capabilities and a better dMax came out, named the XA. Reflecta said they will start selling the EU version of the XA in 2014 Q4.
I have had the XA for a couple of months now.
brbo
Well-known
And...?! How good is it?
Joosep
Well-known
Oh yes. I have been meaning to do a review, but time is just so very limited.
I used to work in a lab for 5 years, so Frontiers, Noritsus and Epsons were bread and butter.
It took me about a week to test Silverfast vs Vuescan.
After that it took another week to get what I wanted from the scanner.
I have not yet done a more "scientific" test.
But I have to say, it wins a Frontier in every aspect, especially dMax. I am planning to write something up one day, I have many different scanners close to me. (I am right now finishing my masters degree and working at the same time, so time is not something I can spend carelessly
).And yes. I need a new computer...
Athiril
Established
I whole heartedly disagree with that Plustek blog.
While I do agree, that the resolving power of plustek scanners are quite good, and sufficient for most uses.
5000 dpi is not overkill. I tested a Plustek 120 scanner at work yesterday, I found about 70 lp/mm peak resolution, or about 3600 dpi resolving power. About the same filmscanner.info rates it at.
I can get vastly superior scans at the same resolution on another scanner. Because at the same resolving power, the Plustek is very very very soft, as it is the utmost peak of it's resolution of the lens, anything that hits peak resolution of a lens, is, well, soft looking.
You dont need a higher resolution sensor to get it very sharp at that resolution, just a better lens.
That said the Plusteks are excellent, and there's no (new) alternative at that price point.
"A current film resolution study (http://cool.conservation-us.org/) finds that average film grain resolution of current generation films is 3264 ppi"
I take issue with this, as it couldn't be more wrong.
I get a little more than can be stored in 3264 ppi with both old expired FP4+ and Shanghai GP3, let alone more modern generation films with tabular grains
Rollei Retro 80S, while not a common film as other films gives me ~100 lp/mm with a Rodinal 1:250 1 hour stand (which may not be the best way to get resolution out of the film, but it looks good tonally) my budget AE-1 and 28/2.8 FD lens < 5K dpi.
Let alone with a camera/lens combo that costs more than $100.
Henning Seger has published his personal tested results in numerous places with more pricey gear.
Eg: http://www.apug.org/forums/forum37/110218-research-development-film-3.html
Everything tested hit above 100 lp/mm. Many of them double or more of the suggested ~3200 dpi value.
iirc, I think I even managed to hit ~90 lp/mm on a Jupiter-9 with a Kiev-4, where the focus was out a little with Adox CMS 20.
Many lenses are capable of resolution far exceeding the suggested values, and many films also can record more than that.
Now I would prefer a scanner I can get a sharp 3200-3600 dpi out, than a soft 3200-3600 dpi (soft, but real resolution mind you), they don't exist in my price range, but lucky I do have one like that a work (and one that gives way more than that), the Plusteks are still excellent however.
While I do agree, that the resolving power of plustek scanners are quite good, and sufficient for most uses.
5000 dpi is not overkill. I tested a Plustek 120 scanner at work yesterday, I found about 70 lp/mm peak resolution, or about 3600 dpi resolving power. About the same filmscanner.info rates it at.
I can get vastly superior scans at the same resolution on another scanner. Because at the same resolving power, the Plustek is very very very soft, as it is the utmost peak of it's resolution of the lens, anything that hits peak resolution of a lens, is, well, soft looking.
You dont need a higher resolution sensor to get it very sharp at that resolution, just a better lens.
That said the Plusteks are excellent, and there's no (new) alternative at that price point.
"A current film resolution study (http://cool.conservation-us.org/) finds that average film grain resolution of current generation films is 3264 ppi"
I take issue with this, as it couldn't be more wrong.
I get a little more than can be stored in 3264 ppi with both old expired FP4+ and Shanghai GP3, let alone more modern generation films with tabular grains
Rollei Retro 80S, while not a common film as other films gives me ~100 lp/mm with a Rodinal 1:250 1 hour stand (which may not be the best way to get resolution out of the film, but it looks good tonally) my budget AE-1 and 28/2.8 FD lens < 5K dpi.
Let alone with a camera/lens combo that costs more than $100.
Henning Seger has published his personal tested results in numerous places with more pricey gear.
Eg: http://www.apug.org/forums/forum37/110218-research-development-film-3.html
Everything tested hit above 100 lp/mm. Many of them double or more of the suggested ~3200 dpi value.
iirc, I think I even managed to hit ~90 lp/mm on a Jupiter-9 with a Kiev-4, where the focus was out a little with Adox CMS 20.
Many lenses are capable of resolution far exceeding the suggested values, and many films also can record more than that.
Now I would prefer a scanner I can get a sharp 3200-3600 dpi out, than a soft 3200-3600 dpi (soft, but real resolution mind you), they don't exist in my price range, but lucky I do have one like that a work (and one that gives way more than that), the Plusteks are still excellent however.
mdarnton
Well-known
So, Athiril, what is the one at work that does such a good job?
ColSebastianMoran
( IRL Richard Karash )
k__43
Registered Film User
The leaflet says fixed focus (again). Wake me up when this changes.
Why the f..k does no one just copy the Canoscan FS4000 design and just add some modern components and makes it faster? That thing was almost 100% efficient in regards of resolution.
Also they state "LED" for the lamp .. I bet it's a white LED and not discrete RGB.
Why the f..k does no one just copy the Canoscan FS4000 design and just add some modern components and makes it faster? That thing was almost 100% efficient in regards of resolution.
Also they state "LED" for the lamp .. I bet it's a white LED and not discrete RGB.
Athiril
Established
Imacon Flextight 949.
jwcat
Well-known
Thanks for the input. I guess I will stick with my X-E1 and Nikkor 55 Micro.
A Y
Member
I've seen Henning's post on APUG, and I am very doubtful of his figures. For example, for a lens+film combination to hit 110 lp/mm (Velvia 50, using his most conservative figures) means that the lens and film individually has to resolve 220 lp/mm, if both are of equal quality. That seems quite unbelievable for either lens or film.
For example, for film, that's more than 11000 DPI! Each resolvable detail is on the order of 2.3 microns on the negative. Silver halide particles are not much smaller than that, and one needs more than a few of those particles to make anything out.
And even if all of that were technically possible, the kind of shot discipline required to consistently achieve that kind of resolution is tremendous. The Nikon D800 is about 105 lp/mm, and people have enough trouble getting sharp images from that camera. Forget PDAF autofocus: you'd have to live-view focus locked down on a good tripod.
And the common misalignment between the reflex mirror and the film plane on film SLRs would wipe all of that resolution out.
And even if your shot discipline were perfect, as the D800 has shown, there are only a small handful of lenses that can resolve at 36MP or 105 lp/mm, let alone more than 120/mm.
Remember that the system resolution is calculated by:
Rsystem = 1/(1/Rlens + 1/Rfilm)
where Rlens is the resolution of the lens, and Rfilm is the resolution of the film. This means that if film and lens are equally resolving, then they have to be twice the system resolution. If one is less than the other, than the other has to be even more resolving to make up the differences.
For example, for film, that's more than 11000 DPI! Each resolvable detail is on the order of 2.3 microns on the negative. Silver halide particles are not much smaller than that, and one needs more than a few of those particles to make anything out.
And even if all of that were technically possible, the kind of shot discipline required to consistently achieve that kind of resolution is tremendous. The Nikon D800 is about 105 lp/mm, and people have enough trouble getting sharp images from that camera. Forget PDAF autofocus: you'd have to live-view focus locked down on a good tripod.
And the common misalignment between the reflex mirror and the film plane on film SLRs would wipe all of that resolution out.
And even if your shot discipline were perfect, as the D800 has shown, there are only a small handful of lenses that can resolve at 36MP or 105 lp/mm, let alone more than 120/mm.
Remember that the system resolution is calculated by:
Rsystem = 1/(1/Rlens + 1/Rfilm)
where Rlens is the resolution of the lens, and Rfilm is the resolution of the film. This means that if film and lens are equally resolving, then they have to be twice the system resolution. If one is less than the other, than the other has to be even more resolving to make up the differences.
Noll
Well-known
Andre - that system resolution equation you cite - is that a hard and fast cnstant or more of a general rule of thumb. Could you cite a reference so I can read more about this?
Joosep
Well-known
I have the XA and it is not fixed focus.
A Y
Member
I don't have a good source for it, and I've tried to derive it from first principles with little success, but I believe it is a pretty hard and fast rule for determining the total resolution from the resolution of its individual components.
The most authoritative source I've seen for it is Fuji's own film guide:
http://www.fujifilmusa.com/shared/bin/ProfessionalFilmDataGuide.pdf
Look at the 2nd to last page (printed page number 129) under “System Resolving Power.”
I first found it on Dante Stella's excellent site: http://www.dantestella.com/technical/formats.html (paragraph a bit after “Enlargement Kills”) He's a regular poster here, so maybe he will chime in if he sees this thread.
Athiril
Established
Sorry to derail, but I need to reply to this misinformation.
No, you don't need more than 11000 dpi of resolution to get 110 lp/mm, you need approx half of that.
I've gotten about 130 lp/mm out of Flextight scans, and 150 lp/mm when I've used 2 in a SR program. The maximum sensor resolution of it is 8000 dpi, the Magnagon enlarging lens in the scanner resolves normally a little bit under that.
Detail size can be smaller than the size of a grain, that is not how film works. Film is not digital, grains are not pixels. A single grain of the average size in an emulsion (an emulsion for one has a range of grain sizes) doesn't behave like a pixel.
Film grain is not the fundamental image element in film, ie: it is not equivalent to a pixel.
The D800 hasn't shown anything. The theoretical maximum resolution is 105 lp/mm if you count it as 100% efficiency. Of course you cannot measure higher values than the limit of the recording medium. That should be obvious. So of course you can't test lenses higher with a D800.
I also have a GH2, which exceeds D800 resolution. Most of the lenses I put on it, including the kit lens and old manual lenses resolve more than 105 lp/mm.
"Only a handful of lenses can resolve D800-level resolution" is a rubbish claim. Take them off the D800, and you can demonstrate otherwise rather simply.
Do not make the assumption that you will have no loss with a lens that's only a little better resolving than 105 lp/mm on a D800, or that that particular optical formula will work well on a sensor like that.
Henning is using a lens Zeiss have stated will hit 250 lp/mm from f/2 to f/5.6 iirc.
He stated he shoots the test at f/5.6 iirc, and has confirmed 240-260 lp/mm on a very high resolution film.
To put down 110 lp/mm on a film, with a 250 lp/mm lens, reversing the formula (1/sr^2 = 1/lr^2 + 1/fr^2 - Higgins, G.C.Appl. Opt. 3, v.1, 9, Jan 1964) - assuming it fully applies on a film like this where the grains have a smaller size range, gives a film resolution of 123 lp/mm. His upper range for Velvia 50 was 125 lp/mm, which gives a film resolution of 144 lp/mm. S = system, L = lens, F = film.
You do not need 220 lp/mm on both, that would give 156 lp/mm on the film under the proper formula. What you have given is NOT a "hard and fast rule". It couldn't be any more further from the truth. It is way off Higgn's equation, and Higgin's equation is just a first order estimate, and not a rule.
My Canon FD 28mm f/2.8 @ f/5.6 hit 150 lp/mm in a scan on Adox CMS 20. I do not know if it is higher or not, because I haven't looked at it on a microscope. I don't find that pictorially useful, but I would for investigation/research purposes.
I've hit just over 130 lp/mm with a Yashica 50mm f/1.4 ML on my GH2 and many others. I do have a Nikon adapter somewhere, and a bunch of Nikon lenses at work in the rental department, as well as Nikon film SLR I can run CMS 20 through and put this to rest.
No, you don't need more than 11000 dpi of resolution to get 110 lp/mm, you need approx half of that.
I've gotten about 130 lp/mm out of Flextight scans, and 150 lp/mm when I've used 2 in a SR program. The maximum sensor resolution of it is 8000 dpi, the Magnagon enlarging lens in the scanner resolves normally a little bit under that.
Detail size can be smaller than the size of a grain, that is not how film works. Film is not digital, grains are not pixels. A single grain of the average size in an emulsion (an emulsion for one has a range of grain sizes) doesn't behave like a pixel.
Film grain is not the fundamental image element in film, ie: it is not equivalent to a pixel.
The D800 hasn't shown anything. The theoretical maximum resolution is 105 lp/mm if you count it as 100% efficiency. Of course you cannot measure higher values than the limit of the recording medium. That should be obvious. So of course you can't test lenses higher with a D800.
I also have a GH2, which exceeds D800 resolution. Most of the lenses I put on it, including the kit lens and old manual lenses resolve more than 105 lp/mm.
"Only a handful of lenses can resolve D800-level resolution" is a rubbish claim. Take them off the D800, and you can demonstrate otherwise rather simply.
Do not make the assumption that you will have no loss with a lens that's only a little better resolving than 105 lp/mm on a D800, or that that particular optical formula will work well on a sensor like that.
Henning is using a lens Zeiss have stated will hit 250 lp/mm from f/2 to f/5.6 iirc.
He stated he shoots the test at f/5.6 iirc, and has confirmed 240-260 lp/mm on a very high resolution film.
To put down 110 lp/mm on a film, with a 250 lp/mm lens, reversing the formula (1/sr^2 = 1/lr^2 + 1/fr^2 - Higgins, G.C.Appl. Opt. 3, v.1, 9, Jan 1964) - assuming it fully applies on a film like this where the grains have a smaller size range, gives a film resolution of 123 lp/mm. His upper range for Velvia 50 was 125 lp/mm, which gives a film resolution of 144 lp/mm. S = system, L = lens, F = film.
You do not need 220 lp/mm on both, that would give 156 lp/mm on the film under the proper formula. What you have given is NOT a "hard and fast rule". It couldn't be any more further from the truth. It is way off Higgn's equation, and Higgin's equation is just a first order estimate, and not a rule.
My Canon FD 28mm f/2.8 @ f/5.6 hit 150 lp/mm in a scan on Adox CMS 20. I do not know if it is higher or not, because I haven't looked at it on a microscope. I don't find that pictorially useful, but I would for investigation/research purposes.
I've hit just over 130 lp/mm with a Yashica 50mm f/1.4 ML on my GH2 and many others. I do have a Nikon adapter somewhere, and a bunch of Nikon lenses at work in the rental department, as well as Nikon film SLR I can run CMS 20 through and put this to rest.
A Y
Member
If you use my/Fuji's resolution formula, a system resolution of 110 lp/mm requires 220 lp/mm for the lens and the film, assuming both are equally resolving (as stated in my original post).
25.4 mm/in * 2 dots/lp * 220 lp/mm = 11176 dots/in
If we take your formula, whose reference I cannot find (except here, and the reference is no clearer for tracking down the original paper) and again assume equal parts resolution, we get:
110 lp/mm * sqrt(2) * 25.4 mm/in * 2 dots/lp = 7902 dots/in
If you can tell me the name of the journal that Higgins is published in, I'd appreciate it, or better, derive that formula. I don't know how to decode "G.C. Appl. Opt."
While that is not as astounding as 11000 DPI, it is still way out there, assuming both film and lens are equally resolving. Again, the numbers don't look good if either is less resolving.
And I really doubt the Zeiss 50/2 has 250 lp/mm. The Zeiss 55/1.4 Otus, which is considered the sharpest lens they have produced, measures 1335 lp/picture height at f/5.6 on a D800E, which gives it 1335/24 = 56 lp/mm. Why is that 4x lower than the 50/2?
One could argue that 56 lp/mm is the system resolution --- the combination of lens and sensor result in a lower number. OK, and given the D800's 105 lp/mm, we can work backwards to figure out the native resolution of the lens. The Otus measures 120 lp/mm for my formula, and 66 lp/mm under Higgins. Either way, it's a far cry from 250 lp/mm.
And that brings up another sticky problem, which is interpretation of the resolution charts as these large resolution numbers are often taken where there is little actual contrast (10% MTF is not uncommon), and depends a lot on the subjectivity of the observer.
OK, but no one said that. All I said is that the smallest resolvable detail implied by 220 lp/mm is very close to the size of a silver halide particle, and in order to make an actual dot one can view as a dot takes more than one silver halide particle.
Very few lenses get to the D800's nominal 105 lp/mm --- they all measure lower. By that observation, very few lenses can reach 220 lp/mm.
Sorry, but this is unbelievable. A 16 MP m43 sensor has 133 lp/mm, which means your Yashica effectively has infinite resolution.
Athiril
Established
When observations don't fit your 'rules' it is your rules that are wrong, not the observations.
Let me make this perfectly clear. Your 'formula' is very inaccurate. It is way off the Kodak equation (Higgins). Secondly, the Higgns' equation (used by Kodak), which is far more accurate than what you have given is for estimation. It is NOT a rule.
Reality is not a reflection of the equation, the equation is a reflection of reality, and not always the case.
Even within the confines of such estimation, Henning's and my own observations are within those particular confines. There is no "rule" being broken. The only thing the observations aren't obeying is gross misinformation.
Applied Optics, Volume 3, Issue 1, January 1964.
http://www.opticsinfobase.org/ao/issue.cfm?volume=3&issue=1
I'm not going to use it, because it is poor, and doesn't fit well with real observations.
No.
Stop assuming. The equation is a first order estimation. I don't know how many times I have to say this. Estimation. Estimation. Estimation. And it doesn't fit well with things like monodisperse sensors, especially uniformly arranged ones, such as a line sensor of a scanner.
8000 dpi of scanner resolution is not require to resolve 110 lp/mm. It is only Plustek scanners that get around 50% or less of their sensor resolution resolved, because their optics are very poor in comparison to previous generations of dedicated and serious film scanners (but amazing compared to flatbeds).
Go see filmscanner.info. You can see any Nikon scanner resolving the vast majority of it's sensor resolution because it has vastly superior optics.
It is funny you should post the link you did, because it quite clearly mentions Kodak as the source.
Fuji's equation estimation is way out. The first clue that it is, is that it is linear.
Have a look at this PDF
http://vitaleartconservation.com/PDF/affects_of_lens_quality_on_image_resolution_v3a_draft.pdf
That is really horribly incorrect. It is made by someone making the same argument as you more or less. Taking what they read somewhere and using that to define reality, without ever having actually done it.
It is placing Technical Pan's resolution recorded on the negative at 83 lp/mm using a 200 lp/mm lens. Observations do not obey this 'rule'.
I get more than that on my 28/2.8 Canon FD lens on T-Max, Retro 80S etc, I'm fairly sure my lens isn't 200 lp/mm, it's the best part of 200, but not 200. Retro 80S and T-Max have lower resolution than Tech Pan, and also lower contrast which again adds to the effect.
Fuji's equation is way out all over the place. It places a 100 lp/mm film and 100 lp/mm lens at 50 lp/mm. This doesn't fit observations at all. Observed resolution is more around the 70 lp/mm, which is what the Kodak equation by Higgins estimates.
Regardless of the equation, stop putting the equation before the results.
No. Stop using dpi, and stop using inches altogether unless you are referring to scanner sensor resolution.
Here is how you do it.
110 lp/mm is the result, given a ~250 lp/mm claimed lens. If the film more or less works to Kodak's equation, then, it used like this:
1/110^2 - 1/250^2. Invert it, square root it, you get a film resolution of 123 ~lp/mm, or 144 lp/mm for Velvia 50 for Henning's upper measurement for Velvia 50.
Fuji's data sheet rates Velvia 50 at 80 lp/mm at 1.6:1 contrast, and 160 lp/mm at 1:1000 contrast, so it is not difficult to imagine the film resolution at least 123 lp/mm at 4:1 contrast, the variable resolution is not a straight line (it's not linear) over contrast, it rises sharply and tapers off.
I highly doubt the highly touted, new design, expensive, manual focus Zeiss Otus lens only gives a maximum of 56 lp/mm.
In case you don't know your Zeiss history, they use to publish test results in Camera Lens News.
The sharpest lens they tested was the Zeiss Biogon 25/2.8. It's peak resolution was 400 lp/mm (if the Otus is the sharpest, as in highest resolution lens by Zeiss, then it must excede this, I doubt it does). I think the alternative SLR lens scored 320 lp/mm. And iirc, they had a Hasselblad mount medium format lens at 250 lp/mm.
Camera Lens News has gone missing off their site, but you can probably still find it on archive.org if you know the quote to search for.
http://www.fredmiranda.com/forum/topic/361737
https://web.archive.org/web/2007060...ntents-Frame/098729cea9ef2428c1257118005d1d20
So there you go, a production copy on a production camera of a Planar is capable of 250 lp/mm according to the manufacturers themselves. Henning also tested it on CMS 20 for his tests and confirmed it.
Now, given a lens that is established to be able to achieve 250 lp/mm, a film that is rated to go to a maximum of 160 lp/mm by the manufacturer, I do not doubt an upper limit of 125 lp/mm - or 140 lp/mm for that matter for some of the other ones. It is also within the estimation of Kodak's equation.
"Each resolvable detail is on the order of 2.3 microns on the negative. Silver halide particles are not much smaller than that, and one needs more than a few of those particles to make anything out."
You pulled this figure out of thin air, it seems you're trying to relate some type of observable grain size to limitation of resolution.
Resolution isn't limited to grain, it can be finer than grain.
Very few films are monodisperse (made up of only 1 grain type and size), they have a range of sizes. Films that tend towards monodispersion have ridiculous contrast. Monodisperse films also aren't even arranged in a single layer, and in a grid like CMOS/CCD sensors are.
"Although the viewer sees a granular pattern, the eye is not necessarily seeingthe individual silver particles, whichrange from about 0.002 mm down toabout a tenth of that size"
- Kodak Handbook H-1
online here - http://www.scribd.com/doc/14741249/Kodak-Films
Incidentally this is where you'll also find the resolution estimation equation.
This is exceptionally poor reasoning. This shouldn't even have to be said. You cannot measure more than the D800 by using a D800 as the benchmark on which it is tested. That is absolutely ridiculous.
Repeatedly you state many lenses are unable to achieve 105 lp/mm on the D800. And taking this as an indication that the lens is the limitation.
Yet unwilling to accept that a lens can put down close to but less than the maximum limit on another camera? Why? Because it is higher than the D800?
Let me make this perfectly clear. Your 'formula' is very inaccurate. It is way off the Kodak equation (Higgins). Secondly, the Higgns' equation (used by Kodak), which is far more accurate than what you have given is for estimation. It is NOT a rule.
Reality is not a reflection of the equation, the equation is a reflection of reality, and not always the case.
Even within the confines of such estimation, Henning's and my own observations are within those particular confines. There is no "rule" being broken. The only thing the observations aren't obeying is gross misinformation.
Applied Optics, Volume 3, Issue 1, January 1964.
http://www.opticsinfobase.org/ao/issue.cfm?volume=3&issue=1
I'm not going to use it, because it is poor, and doesn't fit well with real observations.
No.
Stop assuming. The equation is a first order estimation. I don't know how many times I have to say this. Estimation. Estimation. Estimation. And it doesn't fit well with things like monodisperse sensors, especially uniformly arranged ones, such as a line sensor of a scanner.
8000 dpi of scanner resolution is not require to resolve 110 lp/mm. It is only Plustek scanners that get around 50% or less of their sensor resolution resolved, because their optics are very poor in comparison to previous generations of dedicated and serious film scanners (but amazing compared to flatbeds).
Go see filmscanner.info. You can see any Nikon scanner resolving the vast majority of it's sensor resolution because it has vastly superior optics.
It is funny you should post the link you did, because it quite clearly mentions Kodak as the source.
Fuji's equation estimation is way out. The first clue that it is, is that it is linear.
Have a look at this PDF
http://vitaleartconservation.com/PDF/affects_of_lens_quality_on_image_resolution_v3a_draft.pdf
That is really horribly incorrect. It is made by someone making the same argument as you more or less. Taking what they read somewhere and using that to define reality, without ever having actually done it.
It is placing Technical Pan's resolution recorded on the negative at 83 lp/mm using a 200 lp/mm lens. Observations do not obey this 'rule'.
I get more than that on my 28/2.8 Canon FD lens on T-Max, Retro 80S etc, I'm fairly sure my lens isn't 200 lp/mm, it's the best part of 200, but not 200. Retro 80S and T-Max have lower resolution than Tech Pan, and also lower contrast which again adds to the effect.
Fuji's equation is way out all over the place. It places a 100 lp/mm film and 100 lp/mm lens at 50 lp/mm. This doesn't fit observations at all. Observed resolution is more around the 70 lp/mm, which is what the Kodak equation by Higgins estimates.
Regardless of the equation, stop putting the equation before the results.
No. Stop using dpi, and stop using inches altogether unless you are referring to scanner sensor resolution.
Here is how you do it.
110 lp/mm is the result, given a ~250 lp/mm claimed lens. If the film more or less works to Kodak's equation, then, it used like this:
1/110^2 - 1/250^2. Invert it, square root it, you get a film resolution of 123 ~lp/mm, or 144 lp/mm for Velvia 50 for Henning's upper measurement for Velvia 50.
Fuji's data sheet rates Velvia 50 at 80 lp/mm at 1.6:1 contrast, and 160 lp/mm at 1:1000 contrast, so it is not difficult to imagine the film resolution at least 123 lp/mm at 4:1 contrast, the variable resolution is not a straight line (it's not linear) over contrast, it rises sharply and tapers off.
I highly doubt the highly touted, new design, expensive, manual focus Zeiss Otus lens only gives a maximum of 56 lp/mm.
In case you don't know your Zeiss history, they use to publish test results in Camera Lens News.
The sharpest lens they tested was the Zeiss Biogon 25/2.8. It's peak resolution was 400 lp/mm (if the Otus is the sharpest, as in highest resolution lens by Zeiss, then it must excede this, I doubt it does). I think the alternative SLR lens scored 320 lp/mm. And iirc, they had a Hasselblad mount medium format lens at 250 lp/mm.
Camera Lens News has gone missing off their site, but you can probably still find it on archive.org if you know the quote to search for.
http://www.fredmiranda.com/forum/topic/361737
https://web.archive.org/web/2007060...ntents-Frame/098729cea9ef2428c1257118005d1d20
So there you go, a production copy on a production camera of a Planar is capable of 250 lp/mm according to the manufacturers themselves. Henning also tested it on CMS 20 for his tests and confirmed it.
Now, given a lens that is established to be able to achieve 250 lp/mm, a film that is rated to go to a maximum of 160 lp/mm by the manufacturer, I do not doubt an upper limit of 125 lp/mm - or 140 lp/mm for that matter for some of the other ones. It is also within the estimation of Kodak's equation.
"Each resolvable detail is on the order of 2.3 microns on the negative. Silver halide particles are not much smaller than that, and one needs more than a few of those particles to make anything out."
You pulled this figure out of thin air, it seems you're trying to relate some type of observable grain size to limitation of resolution.
Resolution isn't limited to grain, it can be finer than grain.
Very few films are monodisperse (made up of only 1 grain type and size), they have a range of sizes. Films that tend towards monodispersion have ridiculous contrast. Monodisperse films also aren't even arranged in a single layer, and in a grid like CMOS/CCD sensors are.
"Although the viewer sees a granular pattern, the eye is not necessarily seeingthe individual silver particles, whichrange from about 0.002 mm down toabout a tenth of that size"
- Kodak Handbook H-1
online here - http://www.scribd.com/doc/14741249/Kodak-Films
Incidentally this is where you'll also find the resolution estimation equation.
This is exceptionally poor reasoning. This shouldn't even have to be said. You cannot measure more than the D800 by using a D800 as the benchmark on which it is tested. That is absolutely ridiculous.
Repeatedly you state many lenses are unable to achieve 105 lp/mm on the D800. And taking this as an indication that the lens is the limitation.
Yet unwilling to accept that a lens can put down close to but less than the maximum limit on another camera? Why? Because it is higher than the D800?