Related Local Links:
Medium Format Home Page
Medium Format Cameras List Page
Quality Factors
MTF charts
MTF Chart Interpretation
Film Flatness Issues
Lens Coloration
Bokeh Pages
Lens Faults pages (aberrations..)
Lens and Camera Testing

Related Links:
Film Characteristics Table
(includes RMS, lpmm, latitude, and related info!)

The table of film resolution limits provided by Patrick Bartek below is very critical to understanding the major role of color film resolution as limiting our photography. I have added an extended table of film resolutions available for other current (1998/9) film emulsions.

For years, I used to be confused how so many of my third party prime lenses did so well in the real world against my OEM (Nikon, Pentax..) primes. After all, the magazine lens tests showed how much better my OEM lenses were than my lower cost third party lenses in resolution. How come I couldn't see these huge reported differences in my side by side comparisons?

Now I think I have the answer. The lens tests are typically done with very slow film, often black and white. In any case, the lens resolution targets are invariably black and white lines. The contrast of these lens test targets is very high, typically 1000:1 or better.

By contrast (pun intended), the real world isn't made of pure blacks and pure whites. It is much lower contrast, typically closer to the 1.6:1 ratio shown in Mr. Bartek's table below.

Lenses by themselves have spectacular aerial resolutions, often reaching 300+ lpmm and even 650+ lpmm for typical 35mm SLR optics. To measure such aerial lens resolution, you have to use a microscope (50X to 100X) mounted at the film plane position while focused accurately on a lens test target. Lens aerial resolution is also highly dependent on diffraction, which reduces lens resolution as we stop down towards f/16 and beyond.

When you test a modest performance lens on a test chart, the lens may noticeably reduce system performance when tested with a test chart with 1000:1 contrast and a film resolution limit of 100 lpmm (color) to 320 lpmm (tech pan) with such high contrast subjects. So modest differences in lenses show up on high contrast lens test chart based tests.

Now switch to the real world, where contrast is much lower on the average. Now the color film becomes much more limiting, typically running 40 lpmm to 63 lpmm.

Obviously, if the film can only resolve 50 lpmm, then you aren't going to get all the resolution your lenses can deliver given their aerial resolution greatly exceeds this limit. At best, more expensive and better lenses will enable the lens to do better with black and white and slower films (80-100 lpmm) and get the maximum out of typical color films (50-63 lpmm).

Mr. Bartek's table also explains why slow speed fine grain color films like (now discontinued) Ektar 25 are so popular, and why Velvia is so popular for slide use among professionals and amateurs alike. Both of these color films offer up to 80 lpmm film resolution limits. Finally, the black and white tech pan films offer up to twice or more of the resolution (at 100 lpmm to 125 lpmm) with average contrast ratio (1.6:1) real-world subjects. Using these black and white tech pan films, you can finally get back to the point where your lenses are limiting.

I am not saying that other lens performance criteria are equally impacted as lens resolution by film choice and subject contrast. More costly lenses may have better edge performance, less falloff, lower aberrations, better mechanical build, and lower barrel or pincushion distortion. Issues like lens color and bokeh are important to many professionals and amateur photographers too. Finally, there is a tradeoff between contrast and resolution, as can be seen from MTF test charts.

Now I understand why so many lenses that test so well on black and white high contrast test charts don't do equally better than my less costly third party lenses in real world shots. It also explains why so many zoom users can be happy with less resolution than prime or fixed lens users. If they are using color print film, as most do, the loss of resolution in the lenses is less important when the film is the limiting factor.

These values help explain some comments by noted author Roger Hicks (The Lens Book, The Film Book) that the resolution of color film is more of a limit than lens resolution in most photography. You can also see why professionals recommend slower fine grain color films such as Ektar 25. Film really makes a difference, while lenses are less critical. Surprise!

First, stop worrying about small differences in lens resolution if you are using typical color films (e.g., ASA 100 and up). For real-world shooting with color film, your technique is likely to be much more limiting than your lenses.

By technique, I mean things like using a lens hood or tripod are much more likely to have a positive impact on your final image than some minor resolution differences based on a high contrast lens chart that your film can't record anyway. We have a page of tips on how to get the most out of your lens and camera. I recently posted a page on mirror lockup benefits where using MLU was shown to improve resolution by up to 172% and over 50% on average at slower shutter speeds. Compare that to the 10-25% difference in lens resolution between pricey OEM and lower cost third party lenses. If you are shooting hand-held with typical color print film, your technique is going to be much more limiting than your lens resolution!

Second, look at Mr. Bartek's table again. Is it possible that you are using the wrong film? Just changing the color film you are using can have a lot more impact on final on-film resolution than switching lens brands or paying a lot more for a bit more lens resolution. Isn't it a lot cheaper to push your resolution results up by switching film boxes than paying kilo-bucks for slightly better lenses?

In short, this new understanding should free you from obsessing over your lens quality and resolution values. Your technique and choice of film are much more important and more limiting than slight differences in the resolution of your lenses. So enjoy them, and good shooting!

system resolution =                     1
 (lpmm)             -------------------------------------------
                    (  1/lens lpmm) + (1/film lpmm)...

In other words, if the film resolution is 100 lpmm and the lens resolution is 200 lpmm at a given f/stop, the lens-film system resolution is 67 lpmm (!) and not the average (150 lpmm) or the worst (100 lpmm). If you add an enlarger lens into the above equation, it will further reduce the overall system resolution typically by 15-25% when making a print (rather than a slide as assumed here).

What if we take a typical range of lens lpmm limits (which varys with the lens quality and f/stop and diffraction factors etc.), say from 100 lpmm to 600 lpmm aerial resolution values. Now let us calculate the overall system resolution using films (and test targets) which correspond to film resolution limits of 100 lpmm, 80 lpmm, and 50 lpmm - perhaps a favorite black and white film, a color slide film such as Velvia, and a color print film.

Here is the resulting simple table that contains some surprises:

lens    film   system  film   system   film   system
lpmm    lpmm   lpmm    lpmm   lpmm     lpmm   lpmm
100	  100	   50	      80	44	    50	33
200	  100	   67	      80	57	    50	40
300	  100	   75	      80	63	    50	43
400	  100	   80	      80	67	    50	44
500	  100	   83	      80	69	    50	45
600	  100	   86	      80	71	    50	46

Let us say you buy a pricey superior lens that delivers 600 lpmm aerial resolution (vs. 300 lpmm cheapie). With film resolution of 100 lpmm, the resulting system resolution should be 86 lpmm. This resolution modestly beats the cheapy lens, whose system resolution will be circa 75 lpmm.

What if you switch to typical color print film providing only 50 lpmm film resolution (third section of table above)? Now your pricey lens is only able to deliver 46 lpmm on the film, while the cheapy lens is uncomfortably close at 43 lpmm! A 3 lpmm resolution improvement is going to be awfully hard to see on most minilab prints, even at 8x10". That is why it doesn't matter much which lenses you use with low resolution limit color print films.

The big surprise on the above chart is how major improvements in lens quality and performance does NOT result in equally major improvements in the on-film overall system results. The film is the limiting factor here, not the lenses. Doubling the lens performance with 100 lpmm film resolution limits only resulted in a 15% overall system improvement. Using 50 lpmm film limits, the corresponding system improvement is only about 6%.

Again, I emphasize that this discussion only looks at lens resolution issues. Other factors like light falloff and distortion and aberration corrections should go into lens selection and evaluation criteria. But I find this understanding very helpful in explaining why lenses don't seem to perform very noticeably better in the real world of low contrast subjects (vs. high contrast test charts), given the huge differences in their costs and aerial performance.

[see related posting on formula to use...]

Unfortunately for those of us who want to get the maximum performance possible out of our lenses, the trend is to discontinue slow films (see table below). Slow films such as Ektar 25 were capable of providing up to 200 lpmm in real-world situations (sunlight, focusing the zeiss lenses using the camera's focusing aids, and standard photofinishing services. These results (which are +/- 10 line pairs per mm) may differ from film manufacturers specifications as those values are often determined with specialty lenses (e.g., microscope objectives) rather than actual cameras and lenses. The high performance of the zeiss lenses will probably come as no surprise to those of us who are using them!

Adapted from Zeiss Camera Lens News No. 19 :

Sweet Spots of Lenses Explained

The tradeoff is between diffraction, which lowers lens resolution as you stop down from f/2 towards f/22, and film resolution, which rises somewhat as you stop down. Stopping down a bit lowers the lens' diffraction limited resolution, but it also improves many aberrations. Lowering certain lens aberrations reduces the spreading of light out of the ideal point of light desired at the film plane, so contrast and resolution can go up. Naturally, any given lens may have construction or design criteria which make it vary from this ideal diffraction limited lens, so you have to test each lens for its optimum performance. Still, you can understand from the above chart why so many photographers just say:

Update - Nov. 2000

Similarly, aspheric surfaces and ultra-low dispersion glasses are making improved lens designs possible, if expensive. Unlike film improvements, the costs of such improvements are likely to be large for even modest improvements in overall lens aerial resolutions.

My personal prediction is that the limiting factor in optical system performance may shift to our inability to precisely control the flatness of film. With both 35mm and 120/220 rollfilms, film flatness is a potential problem with faster lenses used wide open. Such faster lenses have higher inherent resolution used wide open due to diffraction effects. But the depth of focus at the film plane gets to be correspondingly shallow. In such conditions, the curling (from the curved lip of a 35mm film cartridge) or the problems with film buckling on 120 rollfilm can cause random parts of the image to be out of the plane of precise focus. Some solutions to this problem include expensive vacuum backs in 35mm and 220 rollfilm (which lacks the blocking paper backing of 120 rollfilm).

For most users, this isn't something to worry about until you control all the other intervening variables. Our tips on improving performance highlight the need to use tripods, clean optics at their maximum performance "sweet spot", and the right high performance films (in lpmm terms). Unfortunately, most autofocus designs limit you to circa 50 lpmm resolution limits due to techical issues (e.g., sensor minimum sizes). If you are using an autofocus camera with consumer grade zoom lenses, hand-holding it, with color print film developed at a local quickie minilab, you probably have at least three or four factors conspiring against you beating our 50 lpmm quality barrier!

So even if new technologies improve one or the other limiting factors such as film or lenses, the overall impact on image quality is likely to be modest. Currently, the resolution limits of typical color print film are such that the film is likely to be the limiting factor rather than the lenses or camera (alignment, flatness, focusing accuracy..).

What of the future? In digital photography, the current resolutions are limited by the distance between sensor patterns on the sensor chip (CMOS or CCD). Most current chips are limited to circa 55 lpmm equivalent, although this could improve with newer chip making technologies (xray synch. sources..). Here again, the highest optical resolution possible may be traded off in digital camera lens design to other factors such as lower cost, smaller size, wider zoom ranges, and so on. The greater light sensitivity of digital devices versus today's films give digital cameras another boost, but can also translate into slower and smaller zoom lenses (f/8 or even f/11). The smaller chip sizes means that lens coverage can be minimized, meaning lens resolution can be maximized with less tradeoffs due to smaller coverage requirements. A typical point and shoot digital camera may have a 6mm f/4 lens not much bigger than this letter "O"!

My guess is that we will end up with a default "50 lpmm" quality standard, based on film and digital resolution barriers cited here. A small fraction of serious amateur photographers will make the effort to get the most out of their cameras and lenses, potentially achieving resolutions several times that of the average photography user. Will there be enough of us so manufacturers will maintain production of the needed films and hardware? Will digital cameras be made simply with bigger chips and higher resolutions, or will the efforts and costs be expended to raise their resolution out of the current 55 lpmm resolution limits? Based on the sales and acceptance of color print film (now 96% of USA consumer sales), I would be surprised to see higher resolution photography be anything but a professional and niche serious amateur market in the future. But I hope I'm wrong! ;-)

n.b. Kodak has stopped providing current film resolution values in the lpmm format, so those shown above are from previous values and tests by Petersen Photographic magazine and other resources. Where only the high contrast value was provided, it is shown above with not/available (n/a) where no data was provided.

For more support for formula: 1/net resolving power= 1/lens resolution + 1/film resolution see R. Kingslake, Lenses in Photography, 1963, p. 77

Related Postings:

From [email protected] Wed Dec 17 1997
From: Patrick Bartek [email protected]
Newsgroups: rec.photo.equipment.35mm
Subject: Re: film resolution in lines/mm?
Date: 16 Dec 97 19:42:21

Regarding film resolution in lines/mm?, B Yen wrote:

Hi,

I've heard that Kodak Technical Pan 2415 (ultra-fine grained B&W) is rated at 200-300 lines/mm. Making a "quantum leap" to a pixel-size, this translates to a "pixel size" of 3-6 microns. Is this the finest grain film available?

I couldn't find any lines/mm rating for KOdak's color films, negative or reversal. I need a numerical rating for Ektar 25, Gold 100, PMC 400, Kodachrome 25, Kodachrome 64, Elite 100, Elite 50. Also need numbers for Fuji's Velvia 50, Fujichrome 100. Someone claimed 13 microns for Fuji's 800 negative film. Does Fuji have any negative film which is competitive with Ektar 25?

I'm shooting a target, which I've figured has a resolution of 5-7 microns. Since the color films have less resolution than TechPan 2415 (~3-6 micron resolution), I figure color film grain will be the limiting factor. But, I need to know some numbers to figure out a ''strategy''.

Regardless of the maximum resolution of the film, the main limiting factors are actual lens resolution and image contrast. With average image contrast, most 35mm optics (even top of the line) resolve only 50 to 55 lines/mm max. There are some exceptional lenses, however. I remember seeing a report of a 90mm one for the Leica resolving 75 lines/mm. Also the Hasselblad 100 f3.5 at f11 to f16 resolving the same.

Now for higher contrast, like a b&w test target, any lens will resolve better, but if you're going to shoot things in the real world and not test targets, the best you'll get regardless of film is what the lens will resolve.

But just to help, here's a few numbers for your tests. There are two resolutions for each film. The first is for a contrast of 1.6:1 (average) and 1000:1 (very high, like a test target).

Good Luck...

--
Patrick Bartek
NoLife Polymath Group
[email protected]

rec.photo.equipment.medium-format
From: "Charles Petzold" [email protected]
[1] Re: MF improvement vs 35
Date: Sun May 03 21:18:03 CDT 1998

According to the "Amphoto Black and White Data Guide" by George Schaub, the resolving power of Pan F Plus (ASA 50) is 170 lines per mm. The resolving power of FP4 (125) is 145 lines per mm. Delta 400 is given as 125 lines/mm and HP5 Plus (400) is 100-125 lines/mm.

So, it appears that for equal enlargements, using ASA 400 film in medium format gives better resolution than ASA 50 film in 35mm cameras.

-- Charles

From: Kerry Thalmann [email protected]
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Lens resolution and different film formats
Date: 20 Aug 1998

Hi John,

Over the past few months, Chris Perez and I have tested a large number of large format lenses of various vintages, brands and focal lengths. Our results can be found on Chris' web site at:

http://www.hevanet.com/cperez/testing.html

Chris has also started testing a few older medium format roll film cameras. Next week we will test a couple more classic medium format cameras along with a couple newer models that are reputed to have very sharp lenses (Mamiya 6MF and Fuji GW690 III). You may find the results interesting.

John wrote:

> Made some comparing tests between the Hasselblad 100mm lens and my Arca
> Swiss with the 90 mm Rodenstock lens.

This is not really an apples:apples comparison. 100mm is slightly longer than normal on the 6x6 Hassy format and 90mm is a fairly wide angle on the 4x5. In our tests, we have found that most wide angle designs (with a few notable exceptions) do not perform as well in the corners as lenses of normal or longer focal length. Because of the much larger coverage angles required, it is harder to design a wide angle to cover a given format (and keep the cost, size and weight raesonable) than to design a normal or long lens for the same format. A better comparison would be the 100mm Hassy against something like a 180mm APO Symmar or APO Sironar-S. Or, if you want to compare wide angles, the 90mm Grandagon against something in the 50mm range for the 6x6 format.

> Speaking about sharpness, I wonder, in what way is there an interaction
> between lens sharpness (MTF) and film format. If we - just for this
> thinking experiment - assume that the film is infinitely high res and there
> are no grains at all, would a sharper lens and shoting on 6x7 cm and
> blowing up to 8x10 prints be less or more sharp than a slighly less sharp
> lens use to shoot on 4x5" and then blown up to an 8x10 print.

Using today's technology, you can't ingnore the effects of film sharpness and grain. They are very real and significant.

> I would think that chromatic effects such as colorized halos etc. and such
> "side effects" would be blown up more shooting on 6x7, but what would
> happen to the resolution? Will the large format accept a proportionally
> less sharp lens and produce the same result, etc etc

Assuming you can ignore the limitations of the film (which you can't, but we'll pretend you can), the sharpness required for indentical prints from the two formats is simply the ratio of the enlargement factors (e). For example, to get an 8x10 print for a 6x6 (2.25"x2.25") negative requires an e of 4.44 and an 8x10 fron 4x5 has an e of 2. So, assuming "ideal" film, the resolution of the medium format lens would need to be 2.22 times greater than the 4x5 lens. This makes a lot of assumptions about film flatness being equal, comparable focal lengths, a perfect enlarging system, etc. So yes (given the stated assumptions), in the larger format, a lens of proportionally lower resolution will be acceptable for the same degree of sharpness in prints of like size.

In addition to just testing the resolution of the lenses, it is a good idea in the real world to test the entire system (as you would use it for making real photographs). There are so many variables other than lens sharpness (which is very important, but just one factor), that to really see the differences, I recommend shooting identical photographs of your preferred subject matter and doing a double blind comparison of side by side like sized prints. If you can tell the difference, go with the format that gives superior results. If you can't tell the difference, go with the format that you feel most comfortable using for your style of photography.

Kerry

--
Kerry L. Thalmann Large Format Images of Nature
A Few of My Images Online at: http://home.att.net/~k.thalmann/

From: "Peter H. Groepper" [email protected]
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Lens Resolution Test Chart???
Date: Wed, 16 Sep 1998

here is a table, which gives theoretical resolution in relation to aperture. (Most of you will know, but it's interesting for those, who don't).

Aperture                Resolution (Pairs of lines / mm)
45                      35
32                      50
22                      70
16                      100
11                      140
8                       200
5.6                     280
4                       400
2.8                     560

Peter

From: Bob Wheeler [email protected]
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Lens Resolution Test Chart???
Date: Wed, 16 Sep 1998

Yes, but wouldn't it be interesting to cite resolutions allowing for a little slop in focusing. For a 150 mm lens at 1500 mm with a 1 mm error in depth of focus (about 80 mm DOF), one has in l/mm

f-stop        diffraction*    depth of focus    combination
64                21            71                21
45                30            50                29
32                42            35                36
22                62            24                38
16                84            13                32
11               122            12                24
 8               169             9                18
 5.6             241             6                12
 4               338             4                 9
 2.8             482             3                 6

One may quibble abut the combination column, but clearly the actual resolution will lie between the first two columns. Not too many pictures are taken with all parts in exact focus, and even where they are, very slight focusing errors change things dramatically. (For those who might be interested, the shape of the combination curve is what many questionably ascribe to lens behavior, saying that the best stop is 2 or three down from the maximum.)

*This differs from Groepper's table simply because we chose slightly different constants -- there is no right choice.

--
Bob Wheeler --- (Reply to: [email protected]) ECHIP, Inc.

From: [email protected] (Richard Knoppow)
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Lens Resolution Test Chart???
Date: Wed, 16 Sep 1998 19:23:16 GMT

There are a couple of things to consider here. First, the usual "theoretical diffraction limit" is a calculated value given for monochromatic light usually at some wavelength in the green or yellow. Actual lens performance with white light will be significantly less.

Also, the diffraction limit is often quoted at the center of the image only. Actually, it varies with the angle from the image axis, and, off axis has different values for radial and tangential lines. The reason for this is that the aperture is no longer symmetrical off axis.

The rule of thumb which states that lens performance is optimum at a stop or two down from maximum has nothing to do with depth of focus, rather it depends on reducing the angles which light rays take through the lens (actually this is also what controlls depth of focus and depth of field) reducing the effects of certain aberrations. Stopping down reduces spherical aberration in all parts of the image and reduces coma away from the center (which is where it exists). Stopping down also reduces the effect of some other aberrations although it doesn't get rid of them.

These effects can be seen when examining an aerial image or on a ground glass, where the focus can be set exactly.

There is certainly a practical effect of reducing lens performance caused by inacurate location of the film in the focal plane whether caused by an incorrectly adjusted camera or by film sag but the above effects are independant of this.

1 mm seems to me a very large error for the focal plane of a medium format camera.

It should also be understood that there are many other factors which control actual lens resolution. The lens can not have better performance than the diffraction limit but it can be worse. Also, the "sharpness" of a lens is affected by the shape of its MTF curve. Its possible and often the case that a lower resolution lens may give sharper looking images than one with higher resolution because it has better edge contrast. Typically, such a lens will have an MTF curve which stays at relatively high values up to medium resolutions before falling off. A lens with a more gradual slope will have a higher resolution limit but may have lower contrast along edges. The eye tends to interpret edge contrast as sharpness rather than resolution.

For medium format cameras an actual _system_ resolution of 50 l/mm, i.e. resolution on the film, is doing very well. Large format systems do not, in general, need this much resolution.

A last comment. I certainly agree that small errors in focus, whether caused by the camera or the photographer, will have a seriously deleterious effect of the sharpness of medium and small format cameras. But this is also likely to have a much less noticeable effect on three dimentional pictorial subjects than on a flat test chart.

---
Richard Knoppow
Los Angeles, Ca.
[email protected]

From: steven T koontz [email protected]
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Lens Resolution Test Chart???
Date: Wed, 16 Sep 1998

Christopher M. Perez wrote:

> Allow me to digress just a moment and say that testing has allowed me to learn
> the limits of the various formats.  For example, I've recently learned that
> lenses constructed for 120 camera systems test at best only marginally  better
> than the average large format lens.  So what does this mean _to me_?  It means
> that I've learned how large I can make a 120-generated image and still make it
> look sharp.  I've learned that LF lenses are quite outstanding and  shouldn't be
> considered in any way less sharp than it's smaller formats brethren. After all,
> they are of similar design manufactured by the same players in the  industry.

Wow! Just looked at your tests and you are testing OLD mostly 3 element folding camera's to modern LF lenses and say "LF lenses are quite outstanding and shouldn't be considered in any way less sharp than it's smaller formats brethren."?

Why don't you compare apples to apples and try something like a hassy 80mm planar or a fuji 65mm 6X9 camera instead of a balda with an enagon lens.. I have a camera with one of those enagons and it's sure not the sharpest med format lens I have. I bet you'll not say this after you test a "real" med format lens, especially at the apertures they get used at (f8-f11) Either one of these will resolve close to if not more that 100 LPMM which is WAY more than a LF lens will resolve.. Don't make such blanket statements from a few tests on some old folding camera's that were never even real good ones when new and most had a very poor film transport/pressure plate system to boot.... --

steve's photography & Z car stuff @ http://www.mindspring.com/~skoontz [email protected]

From: steven T koontz [email protected]
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Lens Resolution Test Chart???
Date: Thu, 17 Sep 1998

Jean-David Beyer wrote:

>
> steven T koontz wrote:
>
> > Christopher M. Perez wrote:
> > >
> > >
> > > Allow me to digress just a moment and say that testing has allowed  me to learn
> > > the limits of the various formats.  For example, I've recently learned that
> > > lenses constructed for 120 camera systems test at best only marginally better
> > > than the average large format lens.

> > Why don't you compare apples to apples and try something like a hassy
> > 80mm planar or a fuji 65mm 6X9 camera instead of a balda with an
> > enagon lens.. [...]
>
> Perhaps because lenses cost money? Perhaps you would send the lenses  you wish tested
> and Christopher M. Perez might find the time to run the tests.

So you agree that med format should be considered limited because a 3 element lens on a folding camera wasn't sharp? Should it be stated that "lenses constructed for 120 camera systems test at best only marginally better than the average large format lens." when they aren't even testing a "system" lens but a few OLD folders and one TLR? Sorry but I don't see making a blanket statement like this using the testing he did as correct. I can't believe after seeing the huge list of LF lenses he tested, that a enagon was the best MF lens he could put his hands on..

This just seemed a way for him to "prove" that the camera (LF) he uses is "The best".. From this type of testing and reasoning someone could also "prove" that 35mm is better in final print than MF or MF was better than 4X5.. That was my point.. If you're going to make a blanket statement compare apples to apples. I bet if I compared prints and resolution of a 90mm optar on 4X5 and a 65mm fuji MF and posted that the MF ones look better, the people here would go ballistic.. --

steve's photography & Z car stuff @ http://www.mindspring.com/~skoontz [email protected]

From: [email protected] (H.Gunnarsson)
Newsgroups: rec.photo.equipment.misc
Subject: Re: Could you tell me some homepages about lens test ?
Date: Thu, 8 Oct 1998

Check out http://www.photodo.com/lens , choose "products & tests"

--
H�kan Gunnarsson
G�teborg/Gothenburg, Sweden

rec.photo.equipment.35mm
From: [email protected]
[1] MTF of Film and Lenses
Date: Sat May 08 07:43:51 CDT 1999

So, I'm checking out the MTF of even good films like the fuji superia 100 and 200, and I see that these guys are falling off pretty quickly, even faster than the MTF of moderate lenses.

So, what's the deal. Seems that even halfway decent lenses are already plumbing the depths of good film.

Date: Tue, 11 May 1999
From: Bert Otten [email protected]
To: [email protected]
Subject: Re: Lens sharpness

Well Bernard,

The present state of Zeiss lenses for Hasselblad is comparable to what Rolls Royce wrote in the early days about their engine power: "sufficient".

To get more specific: If you use an Ektachrome E100S film and shoot a testwall with grids at increasing spatial frequency, up to 100 linepairs/mm on film, you'll see that the film is the limiting factor when the lenses are used at f8 throughout the surface of the image. You'll need an excellent projector (for instance the PCP80 with planar 150 mm) to actually reach the limit of the Ektachrome film over the whole surface. Moreover, you need excellent eyes at a distance of 3.5 m from a projected image of 2 x 2 m to be able to reach that limit.

To give some more details:

1) An Ektachrome E100S reaches a modulation of 0.5 at 40 linepairs/mm, that's 1.1 linepairs/mm on a projected image of 2 x 2 m

2) An 80mm CF Zeiss lens has an average modulation of 0.5 at 40 linepairs/mm at f5.6, that's 1.1 linepairs/mm on a projected image of 2 x 2 m

3) The human eye resolves 60 linepairs/degree, that's 1 linepairs/mm at a distance of 3.5 m from a screen

Conclusion concerning sharpness:

Sufficient

Concerning strategy:

Buy some modern lenses and forget about their quality: just enjoy it and start photographing.

The purpose is not the camera (including the lenses), but the photographs. Another enthusiastic photographer,

Bert Otten

From Nikon Mailing List:
Date: Wed, 12 May 1999
From: "Tyler Harris" [email protected]
Subject: [NIKON] Canon TS-E lens on Nikon Body / 100-400 / Fast Film?

....

Fast Film? Can anyone help me fill out or expand the Chart?

I am looking for a good 800-1600 ISO negative film. I would like to stick with one of the two giants (Fuji or Kodak) and would prefer not to push process. I have read lots of good things about the Royal Gold films but I cannot compare the Kodak Royal Gold 1000 to any Fuji film because Kodak is now using some 'print grain index' thing they made up while Fuji uses the (standard?) Diffuse RMS Granularity thing. Does anyone know of an independent test that uses a standard method regarding granularity and resolving power? I have searched the web, but found nothing. This is what I know of Fuji (from http://www.fujifilm.co.uk/technical/index.html):

                              Resolving Power (lines/mm)
                    RMS Gran.     @1.6:1     @1000:1
*Superia 800 [CZ]      5!          50          100
Super HG 1600 [CU]     10          ??          ???
Royal Gold 1000        ??          ??          ???
**Velvia 50 [RVP]      ?           ??          ???

Based on this information I would guess that the [CZ] has much finer grain than the Royal 1000, but what about sharpness and other characteristics?

*Same as [NGHII] (spec. wise).

**I know Velvia is not fast (nor is it negative) but I would like to know the specs.

Forgive the long post.
Thank You Alexander!!!

Tyler Harris

From Hasselblad Mailing List:
Date: Tue, 11 May 1999
From: [email protected]
To: [email protected]
Subject: Re: Lens sharpness

Last time I checked sharpness using the old Modern Photography test kit, the Hasselblad 100 mm lens provided 100 lpmm on the film. This is pretty darn good even for Leica lenses.

In side-by-side comparisons using Tech Pan film in both the Leica and Hasselblad, the prints are darn hard to distinguish and this is printing with a Durst L 1200 with point light source and the 50 mm and 80 mm special condensers with the Rodagon 50 mm and 80 mm lenses.

I am constantly amazed at the print quality afforded by the Leica. I am getting set to start running tests with the new M series APO's versus the 903 SWC, 100 mm, and 250 mm SA Hasselblad lenses.

Had a setback today as I was supposed to have received a 135 mm APO today but got word it was backordered.

I just set up the cameras on a good tripod and take pictures. Nothing fancy.

I have to agree with you that the Hasselblad lenses are "sharp enough" but so are the Leica lenses.

Thanks for your post.

Years ago Bank Langmore suggested the reason why Hasselblad did not blow away the Leica was the Hasselblad focusing with the mirror, and the various other points at which tolerance mischief could sneak in such as the interchangeable magazines.

Date: Wed, 12 May 1999
From: Indranath Neogy [email protected]
Subject: Re: Lens sharpness

Some points:
(more anecdotal than scientific)

1) Lenses for different formats are biased differently. Leica and Hassy demonstrate this the best b/c they are both Zeiss and so "quality" may not be a big difference. 35mm lenses tend to be slightly more "constrasty" than MF lenses. This can result in a bare contact print (or the neg itself) looking "sharper." Likewise the MF image should have slightly better (finer) tonal gradation. OF course this is comparing full frame vs. full frame. If you compare 35mm portions of each neg, I doubt Hassy will look better.

2) A while back a friend and I compared an 80C vs an 80CF and some of the difference was noticeable, just, but it actually highlighted little alignment problems with the focusing screen more than anything.

3) A test at photodo (an ok website I found during the XPan release fuss) http://www.photodo.com/templates/display.lasso?show=489 points to the fact that the "sharpness advantage" of larger formats is not what we may have thought it was. Of course there are questions about the test methodology, but it is good food for thought.

4) A number of Rollei owners have always claimed that the Schneider optics were sharper than the Zeiss ones (they get both), so if sharpness is the issue, maybe that's the place to go.

5) In the end MF is a compromise between the serious size advantage of large format cameras and the portability (and wide apertures) of 35mm systems. However, much more thought and effort has been put (through the logic of the free market) into improving both 35mm film and lenses than the MF ones, so they are bound to have caught up some. If there was a 35mm system with interchangeable backs and leaf shutter lenses I think it is possible that then the professional use of a Hasselblad might well drop off. (although many of us wouldn't change b/c we're set in our ways)

Indy.

From Hasselblad Mailing List:
Date: Wed, 12 May 1999
From: Russ Rosener [email protected]
Subject: Re: Lens sharpness

You won't see a noticeable image difference between the Hasselblad optics and ANY quality 35mm camera (including Nikon, Canon, and Pentax) until you enlarge the negatives to 16 x 20 inches and beyond. You also need to make sure that the lens you use for medium format enlarging is as good as your lens for enlarging 35mm. I read in Ctein's book "Post Exposure" that the 90mm focal length and longer is ideal for 6x6 cm negatives. The standard 80mm only gives adequate negative coverage and edge sharpness. Face it; the taking optic and system is only half the equation in image quality.

Russ Rosener

From: [email protected]. (G Brown)
Newsgroups: rec.photo.equipment.35mm
Subject: Re: Lens Sharpness!!! [was Re: Nikkor AF 24-120/3.5 not professional?]
Date: Fri, 23 Apr 1999

[email protected] wrote:

> lenses from major mfrs. are uniformly pretty good.  This sounds enough like
> Pop Photography's opinions to cause a meltdown on this newsgroup.

The info a lens can give is more info than the film can handle. Goto

http://www.photodo.com/

and look for the article "35 mm, medium format, or large format?"

It says Tri-X is sharper than Fuji Velvia, and:

"A low-grade lens will give decent pictures when combined with a sharp film, while the best lens will never do itself justice if there is an inferior film in the camera. Read the MTF diagrams and choose the right film for the best results!
<>P ******

Tri-X's corresponding values are approximately 18 lp/mm at 100% and 30% at 100 lp/mm. T-Max 100 is even better than T-Max 400. It has an MTF of 100% at 50 lp/mm (which is impressive) and at 100 lp/mm the MTF is still 65%. Another interesting comparison is slide film versus black and white. The super sharp ISO 50 film Fuji Velvia passes the 100% limit as early as 25 lp/mm and at 100 lp/mm it is down as far as 15% MTF. This film does not compare with T-Max when it comes to sharpness. With the guidance of the curves it is easy to understand that the choice of film is a very important factor in the pursuit of sharpness. A low-grade lens will give decent pictures when combined with a sharp film, while the best lens will never do itself justice if there is an inferior film in the camera. Read the MTF diagrams and choose the right film for the best results!

--
[email protected] -- real address

From: [email protected] (HRphoto)
Newsgroups: rec.photo.equipment.35mm
Subject: Re: Lens Sharpness!!! [was Re: Nikkor AF 24-120/3.5 not professional?]
Date: 23 Apr 1999

>the best lens will never do itself justice if there is an
>inferior film in the camera.

But the best lenses will still outperform inferior lenses, even with low resolution films. See:

http://www.f32.com

under Articles: Manufacture and Performance of Photographic Lenses.

Heinz

From Nikon Mailing List:
Date: Thu, 3 Jun 1999
From: Daniel Berlinger [email protected]
Subject: [NIKON] How Sharp is Sharp (resolution) Part 1

> In short, there's more to photography than lens sharpness, contrast, color
> saturation... much much more.

I think it is difficult to talk about "sharpness" because the word itself is subjective. So let's talk "resolution". Secondly a bunch of this is quoted from an essay by Bryan Geyer of Really Right Stuff.

http://www.reallyrightstuff.com (They manufactuer Arca Swiss style plates that are second to none.)

I don't have the article in front of me, just some notes I made a while back, assume any mistakes are mine.

Daniel

***Lens

Charles Sleicher tested current 35mm format lenses (I have no access to specifics) and found that performance is approximately as follows. It should be noted that line pairs-per-mm resolution being discussed here is not taken off of film but through some sort of microscope arrangement.

resolution at f4 360 (+/- 15%) lpm (diffraction limit approx 400 lpm)
resolution at f22 065 (+/- 10%) lpm (diffraction limit approx 75 lpm)

Some of you will note that this contradicts the long held understanding that "optimum aperture" is a few stops down from "wide open". Don't worry, keep reading.

Note that resolution at f22 is so miserable that it approaches the accepted 0.03mm circle of confusion value (Airy disc spread radius of 0.015mm or 15 microns).

***2x Teleconverters

Now you know why these things are such killers. At best you halve the resolving power of the lens. More on this below.

***Film

Film resolution has been shown to vary appreciably with aperture and contrast. Resolving power is always highest when the light source area is minimized to retard dispersion.

*Big Problem* Highest lens resolution is wide open, highest film resolution is stopped down. This is why published "on film" resolution tests indicate that f8 is optimum with 5.6 and 11 not far behind.

At f8, assuming a lens resolution of approx 200 lpm (a common quality 35mm format lens), a film like Velvia, should provide a system resolution of 100 lpm assuming that resolution was limited solely by diffraction. (30:1 contrast ratio, typical for test charts.)

It is the combination not the "fault" of either element that maximizes resolution at f8. The book "Image Clarity" by John B Williams provides formulas, charts etc.

***What does this mean?

Again from Image Clarity, a resolution of 8 lpm is considered excellent and a standard of 6 lpm is "judged by some viewers as excellent and of exhibition quality".

System resolution consistent with 6 lpm x E, where E is the enlargement factor for the final image when viwed at a normal (15 - 20 inch) distance is within the grasp of all serious 35mm format photographers.

(For those cynics in the audience, I accept this as a reduction of a much more complex set of interactions. The real world is too complex for e-mail :)

***In the real world (sort of :)

Sure, rarely are lenses limited only by diffraction, and you can make case about other aspects... but consider that you can get 100 lpm, and you enlarge your frame 17x, the resulting 16x20 inch print will approach 6 lpm resolution.

Smaller print sizes provide greater flexibility in aperture choice since for example an 8x10 inch print is only an 8.5x enlargement and resolution need only be 51 lpm to match that magical 6 lpm in the final image.

Getting to this point is still going to take some work. On the issue of 2x teleconverters, even with a "perfect" teleconverter you experience a 24% on film loss at the *center* of the image. Adding the decreased contrast and the even greater loss at the sides and corners has led me to take photographs both with and without just to be careful.

More in Part 2...

------------------------------

Date: Thu, 3 Jun 1999
From: Daniel Berlinger [email protected]
Subject: [NIKON] How Sharp is Sharp (resolution) Part 2

....continued from part 1

***Movement and mirror bounce

Movement is a big problem for resolution. Common photographic wisdom dictates that faster shutter speeds eliminate movement issues.

If you move 1/50 of an inch and it takes you 1/10 of a second to do so, the effect is reduced by a shutter speed of say 1/250 since it only "sees" a tiny slice of the total movement. The displacement in that 1/250 of a second would be less than 1/1000 (20 microns).

Back to our theoretical system resolution (lens plus film) of 100 lpm. Adding in our 20 micron displacement, the final system resolution is the reciprocal of the system spread radius or approx 45 lpm. A whopping 55% loss in what is resolved on film! You've tossed away half the resolution your expensive lens and excellent film. At 1/500 res increases to approx 70 lpm, but still... and that's assuming that we started at 100 lpm.

This explains (in part) why mirror locking can provide important resolution increases in the critical zone between about 1/4 of a second and a 1/30. Longer exposures "consume the bounce" as a part of the entire exposure time. Shorter exposures reduce the application of bounce in the exposure. Mirror lock-up is important if you must work in the critical zone with a telephoto lens assuming that you don't have other more pressing issues.

A good tripod or other stable mount is enormously helpful in maximizing system resolution. Like any good foundation you want to sink the weight of the camera and lens (etc.) into the earth.

Unfortunately this requires two things -- balance and weight. Balance is more easily found by ensuring that you keep the camera over the three legs of your tripod and not hanging out on a limb or to one side, and to begin with, by carefully setting your tripod.

Weight is a bigger problem mostly because nobody likes carrying things that are heavy. Especially on top of some hefty camera gear.

Galen Rowel recommends carrying a small stuff stack to hang from a hook off the center post (or centered underneath the top plate) that can be filled with rocks or sand on the spot. This has allowed him to make use of fairly skimpy tripods.

There are many techniques. Pay attention to the wind. Pay attention to your heart beat. Use a shutter release with your camera carefully stabilized on a heavy tripod when the wind isn't blowing.

The point is you have two factors to contend with, mass and mass-coupling. You want a rigid system and you want it to be heavy. Do what you can.

***How Sharp is Sharp?

In the final analysis, and regardless of whether you accept the hypothesis outlined above, "sharpness" is a subjective quality. There are so many more factors that were not included here, or even touched upon. For example, edge contrast can be exploited to increase perceived sharpness (unsharp masking).

In the end you must seek out excellent prints by top image makers and look at them yourself. You must decide what you are reaching for in terms of technical excellence and use that as you would vocabulary. Knowing how to create truly sharp images will only increase your ability to "say" what you want. Another tool in your bag.

Date: Sun, 31 Oct 99
From: Charles Sleicher [email protected]
To: Robert Monaghan [email protected]
Subject: your web site on lens testing

Dear Robert,

I have discovered your useful and informative web site belatedly, but I have a number of things to say about it. First of all, let me tell you that I am the designer of the Sleicher chart that David Jacobsen and Bob Atkins have refered to. By trade I am a chemical engineer, and I am retired from the University of Washington, where I taught for over 30 years. Since retirement I have been able to devote time to photography. I do mostly nature photography - wildlife and scenics. I have had modest commercial success as a stock photographer and seller of prints. My credits include the National Geographic, the Sierra Club Wildlife Calendar, Sunset Magazine, an Audubon calendar, winner of the 1998 Annual Photography contest in the wildlife division of Nature's Best magazine, local pulications, and several exhibits.

About the table you attributed to David Swager - if I click on the link to Swager's resolution page, I get a page from a pamphlet by Brian Geyer of Really Right Stuff. The numbers in the table come from me. They appear in the instructions that I send out with my chart, and were used by Brian with my knowledge and support. I think it would be a good idea if you noted that this table is strictly empirical and subject to change if more or better data become available. Even better, you may wish to present my original table, which has an additional column for Technical Pan.

... I can send you one of my charts. The chart comes with (1) detailed instructions, (2) a set of targets designed for wide angle lenses, and (3) a slide of the chart that has targets that are resolved at 100 l/mm, which helps users to evaluate their microscope or other measuring equipment. I have received many complements on the clarity and completeness of the instructions.

---------

[Ed. note: the following is quoted from Terry's posting:

From: terry roth [email protected]
Subject: lens test chart
Date: 1998-05-18

I was entering a posting on the Sleicher test chart, but my carrier was dropped, so am reposting. The title should actually convey the idea that the chart measures resolution to 160 lines/mm, but it is the resolution of the entire camera system, including camera, camera support, lens, film, technique, etc.

There are 84 targets with 14 pairs of grids in 4 colors (including black) on this 2X3 foot chart. I have attained resolution of 120 l/mm with my Mamiya 7, on Kodak tech pan developed in TD-3. There is a hint of resolution at 140 l/mm, but there is some residual astigmatism and I am not sure the vertical group is resolved (at 100X in a Nikon lab microscope.) Needless to say, this is very nearly at the limits of even Tech pan--the grain, nearly imperceptible in normal size enlargements, in beginning to overwhelm the grid at this magnification. The grid lines (there are 10 horizontal and 10 vertical lines in each of the 14 groups (20 to 160 lines/mm), and the lines of the 160 l/mm grid are less than 0.00015 inches apart on the film.

I got the target at a local pro shop, it is also available for $28 from Mr. Sleicher , a 10-page instruction manual with helpful info on achieving highest resolution is included.

His address is

Charles Sleicher
5002 Harold PL NE
Seattle Wa 98105.

I am not affiliated in any way, other than being a very satisfied user. He suggests using a high quality microscope to view the negatives, but I have found that the enlarger using a good quality enlarging lens, (Nikon, Componon, or Rodagon) and a good focusing magnifer with the enlarger all the way at the top of the column gives comparable results, but a little more difficult to center the various targets at the edges.

[end quote]

From: [email protected] (Zane)
Newsgroups: rec.photo.equipment.medium-format
Subject: Re: MF lenses vs 35mm lenses
Date: Tue, 01 Jun 1999

[email protected] (NYCFoto) wrote:

>>Also, it seems not everyone realizes that, _in
>>principle_, a good lens for 35 mm can approach twice the resolution of a
>>same-field-of-view MF lens (or have a higher MTF value at a given spatial
>>frequency).
>
>       Twice the resolution of a MF lens??  Not true! Even though MF lenses are
>handicapped by having to cover a much larger film area they perform  nearly as
>well as 35mm lenses.

Notice that I tried to emphasize _in principle_ and for the same field coverage. A 100 mm focal length MF lens design with every dimension scaled down by a factor of two to make a _perfectly fabricated_ 50 mm 35 mm format lens would give a cutoff frequency twice the number of lp/mm as the MF lens. (For f/nos small enough that diffraction is not significant---say f/8 and faster). It would be like changing the x axis values on the MTF curve of the MF lens to be double for the 35 mm curve. For example, the contrast at 40 lp/mm would become the contrast at 80 lp/mm for the 35 mm lens. (I'm ignoring film effects for the sake of illustration.)

The reasons this is not usually realized in practice include:

---the aforementioned diffraction.

---the tolerances on the 50 mm lens would have to be approximately twice as good---usually not economically feasible.

---the tolerances on film plane position and flatness would have to be twice as good.

---camera shake has to be twice as small.

---the film MTF has to be convolved with the lens MTF, having a bigger effect on the 50 mm lens.

---the 35 mm lenses are usually not designed to have as large a minimum f/no, causing more aberrations. That is, a typical 50 mm lens is designed to work at f/1.8 or so versus f/2.8 or so for the MF 100 mm lens. This hurts the MTF at slower f/nos.

>If you check photodo.com you'll see that the Rolleiflex MF Schneider lenses
>perform as well, and in many cases better then their 35mm counter parts.

It's kind of hard to find complete MTF curves for lenses. One can do some comparisons with data from the Hassleblad lens book and the Canon lens book. I haven't looked at the Schneider data, but take your word for it. (Are you comparing 2x focal lengths for the MF lenses? That is, same FOV?)

>       I have a rather extensive Nikon F5 system, and I primarily use it for shoots
>when high speed is necessary, like cars at high speed, or on vacation. When
>it's a job, or I want the best possible quality I'll go for the  Rolleiflex for
>portability or people photography, or the 45 or 810 Sinar for still life. 35mm
>just doesn't have the image quality that I'm used to.  Once you get used to the
>image quality of larger formats it's hard to go back to 35mm.

You have better gear than I do. I use a Pentax 67 and Canon 35 mm. But I _do_ agree with your basic points. I also use the 6x7 when I want a really good print and for _all_ portrait work. I've run pretty extensive comparative tests with a bunch of lenses with my gear using high resolution film, and the 6x7 wins out on all counts.

The only point I'm trying to make is that 35 mm has _potential_ to come close to MF performance. Current lens designs (especially the usually fast 35 mm designs), economical fabrication tolerances, and film MTFs keep it from quite getting there. But it's for sure not twice as bad.

Zane

From: "Kerry L. Thalmann" [email protected]
Newsgroups: rec.photo.equipment.medium-format
Subject: Re: MF lenses vs 35mm lenses
Date: Tue, 01 Jun 1999

Zane wrote:

> Again, are these for the same FOV?

Some yes, and some no. The 10 - 20% numbers I was quoting were for the best lenses I have tested in each of the formats regardless of focal length. I was just speaking in general terms for the highest resolution numbers measured in each format for a variety of focal lengths (hence, the rather vague 10 - 20% range rather thna a fixed number). I was also speaking in terms of both on-axis (center) and off-axis (corner) resolution and over a range of typical apertures (all these factors are considered in the 10 - 20% range I mentioned).

>   I ran some tests using Ektar 25 and
> got at least 50% in _cutoff_ resolution using my best fixed focal  length FD
> lenses versus my Pentax 67 lenses. (Based on examining the negatives  with a
> microscope.)  I certainly wouldn't argue that the Pentax lenses are the
> best, but then neither are the FDs.  Just a data point for your
> consideration.

What do you mean by the term "cutoff" resolution? What test target did you use? Do you have the results posted anywhere on the web where I can see the actual numbers. I am always curious to see the results of others. There are always some variables involved in any such test set-up (film type, test target, lighting, the eyes of the person doing the focusing and reading the results, lens to subject distance - or magnification ratio, etc.), but it is still interesting to see the results. All of our testing was done at 20:1 using T-Max 100 and the USAF test chart. The LF results are at:

http://www.hevanet.com/cperez/testing.html

and the MF results are at:

http://www.hevanet.com/cperez/MF_testing.html

It's a rather odd hodge-podge of lenses we've tested. In LF, everything from 60+ year old Zeiss Dagors to a Super Symmar XL Aspheric. In MF, everything from cheap old folders through a Mamiya 6MF (unfortunately no modern Zeiss or Schneider lenses on a Hasselblad or a Rollie). Basically, whatever we could get our hands on. Although they may be slightly inferior in other ways (contrast, resistance to flare, color fidelity, etc.), it's really quite amazing how sharp some of the older lenses are. Still, give me my 110 Super Symmar XL on the 4x5 and a box of Velvia Quickloads and I'm a happy puppy. It's a tough combination to beat (unless you go with a Super Symmar or APO Sironar-S on an 8x10).

Kerry
--
Kerry L. Thalmann Large Format Images of Nature
A Few of My Images Online at: http://www.thalmann.com/

From: [email protected] (Zane)
Newsgroups: rec.photo.equipment.medium-format
Subject: Re: MF lenses vs 35mm lenses
Date: Tue, 01 Jun 1999

"Kerry L. Thalmann" [email protected] wrote:

(snip)

>What do you mean by the term "cutoff" resolution?

The point at which the MTF goes to zero. Or the spatial frequency at which no modulation (contrast variation) shows on the film at all.

> What test target did you use?

I started to just test my Pentax 67 lenses (6) against each other, including getting a more detailed idea of what my 120 mm soft lens was doing. So I wanted more information than just test target data, like contrast at lower frequency, flare, etc. I set up a pretty complex scene in a bookcase with various objects not all at the same distance (to make sure I didn't get a focus error). I included several paper charts I printed up on an inkjet printer, including some that were basically equivalent to an AF chart. Also, some shiny things so I could get some idea about flare and some lines for distortion check.

I decided that while I was at it I'd test some 35 mm lenses the same way, so I tested about thirteen lenses in all. I limited the number of f/nos to four per lens in most cases and bracketed the exposure 1/2 stop with the flash.

> Do you have the results posted anywhere on the web where I can
>see the actual numbers.  I am always curious to see the results of
>others.

No, sorry. I don't have a site. I guess I need to write up a neat summary to email to interested parties. I wasn't trying to make a formal review at the time. As you probably know, a high cutoff resolution doesn't necessarily make the best picture. This was very evident in comparing the 35 mm lenses to each other and themselves at different f/nos, and one almost needs to visually compare the whole negatives to make an assessment. (Sometimes you could see very fine bars on the negative, but the overall picture looked lousy due to lack of middle frequency contrast and edge sharpness---I don't know for sure how to quantify this with the equipment I have.) I will keep your address for when/if I summarize results.

>  There are always some variables involved in any such test
>set-up (film type, test target, lighting, the eyes of the person doing
>the focusing and reading the results, lens to subject distance - or
>magnification ratio, etc.), but it is still interesting to see the
>results.

Film was Ektar 25 for both cameras. I used four studio flashes set at symmetric angles. I focused using a 2x magnifier and, as I said, had things staggered some in distance (mostly within the depth of field range) to identify any focus errors. Although not exactly kosher, I matched the horizontal FOV for each lens by moving the camera back and forth. (All the images looked pretty much alike except for perspective.) I corrected the exposure at the scene center to the 0.1 stop reading of a Gossen Ultraspot for each f/no. The width of the scene was about 6 feet. MF lens focal lengths ranged from 55 mm to 200 mm. 28 to 135 for 35 mm.

The main test I had in mind was seeing what alternating bar frequency I could resolve at 30x. Interestingly to me, the best observation for predicting how good the print would look was how sharp and contrasty some medium size letters on some book spines were -- a purely subjective thing without a scanning densitometer.

I printed a good number of the negatives using 100 and 50 mm Schneider Componon S enlarging lenses on medium contrast paper. I took one of the best 35 mm 8x10s (50 mm lens at f/8 or f/11) and a typical MF 8x10 to my wife (who knows nothing about photography and didn't know what I was doing) and asked her which looked better. She allowed that there didn't appear to be a lot of difference, but identified the MF as somewhat better ("smoother"). (After I showed her what to look for, she could see the differences easily, but I wanted to know if the casual observer would notice much difference. Without two to compare against each other, she had to have some practice to tell whether a print was 35 mm or MF at 8x10. That Ektar is (was) really good.)

BTW, excluding the 120 Soft, the only way I could tell one MF lens from the other was by the perspective.

>  The LF results are at:
>http://www.hevanet.com/cperez/testing.html
>and the MF results are at:
>http://www.hevanet.com/cperez/MF_testing.html

Thanks a lot. I'm going there now.

(snip)

Zane

Date: Sun, 6 Jun 1999
To: [email protected]
Newsgroups: rec.photo.equipment.medium-format
Followup-To: rec.photo.equipment.medium-format
Subject: film as limiting factor Re: MF lenses vs 35mm lenses

You are both right.

Heinz is right in the sense that if you can find the right specialty film and use it in a limited test environment with high contrast situations and special processing, even use filters to create a near-monochromatic red light to minimize phase shifting etc., then you can see major impacts from higher resolution lenses that have aerial resolutions of 300 lpmm and up.

Unfortunately, I use mostly kodak and fuji color films in average contrast situations. At this point, the film really is the limiting factor; see http://www.smu.edu/~rmonagha/mf/lenslpm.html table on film limitations

most color film in real world (not test target) 1.6:1 contrast situations is lucky to get to 50 or 60+ lpmm. Even with high contrast 1000:1 test target style situations, regular films max out around 150 l/mm, vs. 300+ for typical 35mm aerial lens resolutions of most 35mm lenses, even the not so great ones...

If you pay megabucks for a Leica or Nikon or Canon lens that has 600 l/mm versus lesser bucks for a 300 l/mm aerial resolution lens by Cosina, the difference with a 50 l/mm film limiting the overall system resolution will only be 46 l/mm vs. 43 l/mm.

A very expensive 100% improvement in aerial resolution of your lenses has only resulted in a mere 6% or so improvement on film results. Big deal!

In a way, though, it is a big deal, as it explains why so many lenses that have much higher aerial resolutions don't show any really dramatic improvements on the film or final image. The film is limiting us, period!

This also explains in part why Bigger Is Really Better - if you switch to medium format, you get a larger chunk of film, which being the limiting factor, really gives you big benefits (not to mention tonality etc.).

If you shoot black and white film, with its thinner single emulsion layer, your mileage will vary from the above color film analysis. With a 125 lpmm black and white film limit for tech pan film and average subjects, a shift from a 300 lpmm cheapy lens to a 600 lpmm pricey lens may yield up to a 25% or so improvement in on film results. But if you don't use special developing tricks, that drops to only about 12-13% or so. Ooops!

Let's look at this from another perspective. Suppose you switch from Kodak ektachrome 100 (50 lpmm) to Fuji Velvia Pro (80 lpmm). How does your overall system resolution change?

Surprise! Switching films has increased the on-film results by an average of 50% (47% for 300 lpmm lens, 53% for 600 lpmm lens). Upgrading the lens resolution ten-fold from 300 lpmm to 3,000 lpmm only yields a 15% improvement when the color film is the limiting factor (e.g., at 50 lpmm)

So a small shift in what film you use has 3 times the effect of switching from a so-so to the best lenses available. Again, the film is the limit.

This also explains why some of us are so happy with Fuji Velvia Pro film; we can see a 50% improvement from the film swapout, while a 5-7% improvement due to sharper lenses is much harder to see and verify.

Conclusions:

if resolution is a problem, switch to a slower and higher resolution limit color film, or better yet, black and white if possible.

Even better, switch formats, as bigger is better when the film is the limiting factor.

money spent improving lens resolution beyond a nominal point is largely wasted when using today's color film on average contrast real-world subjects

in our example, a 100% improvement in aerial lens resolution resulted in only a 5-7% improvement in on-film results when used with typical amateur color films in typical real-world contrast situations (vs. test targets)

instead of worrying about how much better built and designed our Leica or Nikon or Canon lenses are than their competitors, we should be up in arms demanding that film makers get with it and deliver better films

grins bobm

Date: 04 Aug 1999
From: [email protected] (Hemi4268)
Newsgroups: rec.photo.misc
Subject: Re: effective resolution ?

Hi

Silver image resolution is rated in Lines Per Millimeter. Total system resolution using a ASA100 to 200 film would be somewhere between 75 and 100 l/mm.

Lens resolution a f-11 would be around 200l/mm and film resolution would be somewhere in the 200l/mm range.

System resolution can be calculated

lens reso x film reso
------------------------------   =  system reso
lens reso + film reso

200x200
------------ +  100 l/mm
200+200

Add a little camera shake and your down to 75l/mm

Larry

[Ed. note: Mr. Erwin Putts is a noted Leica lens tester, reviewer, and author...]
From Leica Mailing List:
Date: Mon, 14 Aug 2000
From: "Erwin Puts" [email protected]
Subject: [Leica] BW developments

My book will be published by Hovebooks: www.hovebooks.co.uk, wher eyou will find all info: price, ISDN etc. Signed copies and all these topics have to be discussed with them and I will keep you informed. The history and development of the Leica system, optically and mechanically has been analysed and documented, in my view at least, only partially and strongly biased. The recent discussion about the Summicron 50 versions is one small example of the scant factual evidence most Leica topics are based upon.

In the next issue of LFI I will present the first part of a two part article about Leica lens manufacture, which might be interesting to some. Now on to a different topic. I did a study on current BW films and image quality as related to current Leica lens performance. (it will be published in 'SchwarzWeiss', a German magazine devoted to BW photography). It is a part of an ongoing research project into the optimal combination of Leica lens quality and current BW-emulsions. It is still the case that BW is the best medium for Leica quality, followed by slides and at at a large distance colour neg film.

Focus was the queation if a ISO100 or slower film shows significantly more image quality than the higher speed fims.

Another myth went down. I used Ilford films PanF+, D100, D400 (to be replaced at Photokina by a newer emulsion called 400D!) and D3200. Also in the line was Kodak PlusX, a favourite of mine because of its high acutance and very fine tonal scale. All films were developed in new Ilford DD-X. I had not used this developer before and some may think it rash to use an unknown developer for important work.

I exposed the films as indicated with the exception of D400 (as E320) and D3200 (as EI1600) and according to the times in the leaflet. Result? Excellently exposed and developed negatives, so the old advice to do a long series of trials to find the correct time may be biting the dust. While it is true that some of the negatives would benefit by some fine tuning of develoment time, all negatives, even the one that were flashed with an old Metz CT-1, could be used without reservation in the darkroom. The fine quality of the current VC papers and the Heiland Splitgrade did the rest and I ended the day with a series of prints of very good quality.

What did it show. At 12 times enlargement, the PanF, the D100 are in a class of their own, with the PanF having slightly finer grain. Here we see outstandingly good reproduction of details and tone values, the resolution of the system surpassing 100lp/mm on the negative. The often discussed three dimensional close-to-life- quality is very visible. At ISO400, the image quality drops significantly, but the sharpness impression is very high, but deceptively so. The life-like impression is gone and the image is flat. lacking depth and detail. But tonal quality is very good. At ISO1600 this impression is even stronger, no detail rendition, and again a flatter image, with good tonal range.

It is evident, that the finer points of the quality of Leica lenses are mostly lost when using ISO speeds at 400 and over. I am not discussing the use of the Leica camera is situations where a 400ISO or higher film is required. The M and R in these situations are redoubtable instruments, but image quality as it stands has dropped significantly and I wonder if in these situtions a non-Leica camera would deliver comparable results, technically speaking. The picture taking qualities of the Leica camera and the confidence and inspiration of use are a different topic.

I would always use a Leica in adverse lighting conditions, but I am aware that the performance gap with the nearest competitor is narrow and can only be widened by expert use of all components of the imaging chain.

The potential quality of the combo Leica and 100ISO BW will indeed intrude into H'blad territory.

Erwin

Date: Wed, 30 Aug 2000
From: [email protected] (Richard Knoppow)
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Final Resolution - the real answer?

"jjs" [email protected] wrote:

>Michael Gudzinowicz [email protected] wrote 
>> [ BLAH BLAH BLAH]
>
>Would you, or can you submt a figure as the others have, or are you just
>going say it is impossible? Or is your metric of impossible merely the
>difficult?

He did. In fact a couple. One is the simple 1/T = 1/A + 1/B + 1/n Where T = overall resolution and A,B, etc., are the resolutions of the components.

This works for many systems.

For others 1/T^1/2 = 1/A^2 + 1/B^2 + 1/n^2 may be closer.

Such approximations are OK provided you understand that they _are_ approximations and may be misleading.

The reason for more than one approximate equasion is because of the inacuracies.

If you know where the thin ice is you can avoid falling through. The thin ice warning is the part you labled blah-blah.

---
Richard Knoppow
Los Angeles, Ca.
[email protected]

Date: 30 Aug 2000
From: [email protected] (Michael Gudzinowicz)
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Final Resolution - the real answer?

.... (quoting from above post)

A couple of references for the estimation formulas are:

1/R = 1/R1 + 1/R2 : Katz, A.H., J. Opt. Sci. Am. 38, v.1, 604-610, July 1948

1/R^2 = 1/R1^2 + 1/R2^2: Higgins, G.C.Appl. Opt. 3, v.1, 9, Jan 1964

The way I approached the problem was to do non-linear curve fits to actual data at apertures where the lens was in the diffraction limited area. Below,

I pasted in some data from Dr. Ron Harris' tests of a Zeiss 80 mm lens on a Hassy. The films examined were Tech Pan (TP; Technidol), T-Max 100 (TMX; D76), Agfa APX 100 (APX100; Rodinal), T-Max 400 (TMY; D76) and Tri-X (TX; HC110).

Rather than assume a resolution value for the film, which will usually are inappropriate, since the published values do not match actual target contrasts, etc., the film resolution is determined from a curve fit to actual data. The lens is assumed to be nearly "perfect" on-axis at f/11 to f22, which isn't unreasonable, and the lens resolution limit used was 1600/f#.

The equations were:

R_observed = 1/( f#/1600 + 1/R_film_unknown)

R_observed = 1/( (f#/1600)^2 + 1/(R_film_unknown)^2)^.5


                    Resolutions from curve fit (lpmm)
                   TP      TMX     APX100  TMY     TX

1/R                141.6   117.8   79.5    77.0    51.3
1/R^2              76.2    67.2    50.9    49.8    36.4

I've left out the statistics, and used the film resolutions derived for the two equations to calculate resolutions and compare them to the measured values.

                   TP      TMX     APX100  TMY     TX

11.3  Measured     72      64      49      47      38
11.3  1/R Calc     70.8    64.3    50.9    49.9    37.6
11.3  1/R^2 Calc   67.1    60.7    47.9    46.9    35.2

16.0  Measured     57      54      45      45      34
16.0  1/R Calc     58.6    54.1    44.3    43.5    33.9
16.0  1/R^2 Calc   60.6    55.8    45.3    44.5    34.2

22.6  Measured     47      45      40      40      29
22.6  1/R Calc     47.2    44.2    37.4    36.9    29.7
22.6  1/R^2 Calc   51.9    48.7    41.3    40.7    32.4

It might appear that either equation gives a good fit to the data, and could be used to "predict" resolution based on a given film, developer and conditions under which the lens is "perfect" on axis.

A problem should be evident if we do so using a good 35 mm camera and lens, and the same film developer and target combination at an aperture of f/5.6. The on film resolutions will all be much better than the resolutions predicted by the curve fit using 1/R^2. Actually, typical values for TP and TMX are in the 90-105 lpmm range with good lenses and technique.

In other words, the resolution limit of the film derived by the 1/R^2 curve fit must be in error, though it might look good over a very narrow range of values. One could arbitrarily use larger values for the film resolution in the 1/R^2 equation, but if that is done, the calculated resolutions aren't even close to the actual measured values.

That's the primary reason I dismissed the 1/R^2 approximation when used with pictorial films. I've run a a lot of data through different models, and 1/R apears to be far more reliable. One can also tack on equations for film thickness (spread at wide apertures, etc.)

One can also let the exponents float, and the values derived are usually very close to 1.0 (+/-0.1). If one lets the Rayleigh value (wavelength) vary, it's possible to get a "best" fit with a reasonable number (1200 for example), however, the film resolution might turn out to be -90 lpmm (minus 90), another absurdity.

From: [email protected] (Michael Gudzinowicz) Newsgroups: rec.photo.equipment.large-format
Subject: Re: Final Resolution - the real answer?
Date: 31 Aug 2000

Struan Gray [email protected] wrote:

>Michael Gudzinowicz, [email protected] writes:
>
>> R_observed = 1/( f#/1600 + 1/R_film_unknown)
>
>        Thanks for the information, and the references.  My library
>doesn't have those papers, so could you give a quick explanation of
>what it is you are measuring in your curve fit to get a value for
>R_observed?

I'm sorry if that wasn't clear. I used the values determined by Harris on film using the Hassy. These are the "measured" values in the following table. I geuss I should have labeled them as R_observed. Below each observed or measured value in the following table, are values calculated using the film resolution derived from the curve fits for the experimental conditions.

                   TP      TMX     APX100  TMY     TX

11.3  Measured     72      64      49      47      38
11.3  1/R Calc     70.8    64.3    50.9    49.9    37.6
11.3  1/R^2 Calc   67.1    60.7    47.9    46.9    35.2

16.0  Measured     57      54      45      45      34
16.0  1/R Calc     58.6    54.1    44.3    43.5    33.9
16.0  1/R^2 Calc   60.6    55.8    45.3    44.5    34.2

22.6  Measured     47      45      40      40      29
22.6  1/R Calc     47.2    44.2    37.4    36.9    29.7
22.6  1/R^2 Calc   51.9    48.7    41.3    40.7    32.4

Date: 31 Aug 2000
From: Struan Gray [email protected]
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Final Resolution - the real answer?

Michael Gudzinowicz, [email protected] writes:

> I wrote:
>>
>>> R_observed = 1/( f#/1600 + 1/R_film_unknown)
>>
>> could you give a quick explanation of what it is you
>> are measuring in your curve fit to get a value for
>> R_observed?
>
> I used the values determined by Harris on film using the Hassy.

So I guess I'm asking how Harris defined resolution.

I'm not being argumentative. I am more than happy to accept that the 1/R formula works better than the 1/R^2 one for practical working definitions of resolution, I'm just wondering why.

If we ignore film's nonlinearities we have to be convolving error functions of some sort. The 1/R^2 form assumes that you are combining two independent Gaussian error functions. It's easy to understand - and hence crops up all over basic textbooks - because the convolution of two Gaussians is another Gaussian. If the 'real' formula is 1/R, it can either be because the error functions are more sharply peaked than a Gaussian (which is interesting), or it may be an artifact of how you define their width (which isn't).

0n-axis, a properly focussed, diffraction limited lens will have an Airy error function whose 'width' is typically taken to be the position of the first zero. If you define a Gaussian with the same 'width' it looks much fatter, because Gaussian widths are defined a relatively long way up the peak. If you just plug the Airy width into the 1/R^2 formula you will automatically over-estimate the error (and hence underestimate the resolution) because the implicit model Gaussian you are using is simply too fat.

Imagine trying to predict a system resolution by actually convolving the functions. If you convolve a Gaussian and an Airy function of the same width (defined conventionally) you get a curve which is almost exactly like the original Gaussian. Naively, this implies that if you take a lens and film with the same resolution, the combined resolution of the system is almost entirely set by the film - the lens has almost no effect. The paradox is solved by taking a close look at what is meant by 'width' - the lens and film didn't have the same resolution at all because the Airy function is taller an narrower than the Gaussian.

So I wonder if at least part of the reason for the 1/R behaviour may be that it compensates for inaccurate modelling of the real response functions. I've already mentioned problems with lenses, but I would also expect film's response to be more strongly peaked than a Gaussian because grain has hard edges and tend to clump together.

If all this is covered in the references you gave, say the word and I'll shut up and wait for inter-library loan to do its stuff. if not, I'd be interested in your comments.

Struan

Date: 2 Sep 2000
From: [email protected] (Michael Gudzinowicz)
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Final Resolution - the real answer?

Struan Gray [email protected] wrote:

>Michael Gudzinowicz, [email protected] writes:
>>
>> I wrote:
>>>
>>>> R_observed = 1/( f#/1600 + 1/R_film_unknown)
>>>
>>> could you give a quick explanation of what it is you
>>> are measuring in your curve fit to get a value for
>>> R_observed?
>>
>> I used the values determined by Harris on film using the Hassy.
>
>    So I guess I'm asking how Harris defined resolution.

Harris used an Air Force type resolution target. The reason I chose his data was out of convenience. He measured resolution with different formats, with a variety of films using manufacturer recommended developers, with different apertures, and with placement across the lens field. His purpose was to compare formats with respect to granularity, and resultant print size with sharpness determined by eye and under a loupe. Since he was not concerned with lens sharpness per se, there does not appear to be any evident bias in his data. The article was published in D&CCT.

>    I'm not being argumentative.  I am more than happy to accept that
>the 1/R formula works better than the 1/R^2 one for practical working
>definitions of resolution, I'm just wondering why.
>
>    If we ignore film's nonlinearities we have to be convolving error
>functions of some sort.  The 1/R^2 form assumes that you are combining
>two independent Gaussian error functions.  It's easy to understand -
>and hence crops up all over basic textbooks - because the convolution
>of two Gaussians is another Gaussian.  If the 'real' formula is 1/R,
>it can either be because the error functions are more sharply peaked
>than a Gaussian (which is interesting), or it may be an artifact of
>how you define their width (which isn't).

In my first post, I suggested that the behavior of film "grain" when a single point is considered, is non-Gaussian. I unsucessfully tried to find my copy of Dainty & Shaw's " Image science : principles, analysis and evaluation of photographic-type imaging processes" (Academic Press; 1974) which is a fairly thorough treatment of the subject.

Instead, I clipped out two passages from Niels Jensen's "Optical and Photographic Reconnaissance Systems" (Wiley; 1968). Neither address your question directly, but recognize the problem.

The first on granularity and aperture size suggests what happens when the detector size approaches that of grain -> non-Gaussian behavior results. It isn't much of a leap to assume that when the image point is on the same order or smaller, similar behavior results. The second qoutation on point spread functions is at the end of this post. At the end of that quotation, the author comments on the suitability of the model.

p.175 (minor editing required)

"Granularity has been defined in several different ways Selwyn proposed the following formula: G =(2a sigma_d)^.5 where G is called the Selwyn granularity coefficient, a is the area of the scanning spot in square microns, and sigma_d, is the standard deviation in the density or also the RMS granularity. Table 10.3 lists values of RMS granularity for several common aerial films. These values were measured for a net film density of 1.0 and a 24-micron-diameter scanning spot. They will be higher when the density is higher and lower when the density is lower. Moreover, because the Selwyn granularity coefficient is presumably constant, the formula can be used to determine the RMS granularity of the films for different-size scanning spots.

Unfortunately, most scanners of interest have spot diameters smaller than 24 microns. A 24 micron spot would resolve about 1/[2((2)^.5)(0.024)] = 15 lpmm To resolve on the order of 100 1pmm, we need a spot diameter of about 3.5 microns. This is close to the size of the grains, and the fluctuations would no longer be Gaussian."

>    0n-axis, a properly focussed, diffraction limited lens will have
>an Airy error function whose 'width' is typically taken to be the
>position of the first zero.  If you define a Gaussian with the same
>'width' it looks much fatter, because Gaussian widths are defined a
>relatively long way up the peak.  If you just plug the Airy width into
>the 1/R^2 formula you will automatically over-estimate the error (and
>hence underestimate the resolution) because the implicit model
>Gaussian you are using is simply too fat.
>
>    Imagine trying to predict a system resolution by actually
>convolving the functions.  If you convolve a Gaussian and an Airy
>function of the same width (defined conventionally) you get a curve
>which is almost exactly like the original Gaussian. Naively, this
>implies that if you take a lens and film with the same resolution, the
>combined resolution of the system is almost entirely set by the film -
>the lens has almost no effect.  The paradox is solved by taking a
>close look at what is meant by 'width' - the lens and film didn't have
>the same resolution at all because the Airy function is taller an
>narrower than the Gaussian.
>
>    So I wonder if at least part of the reason for the 1/R behaviour
>may be that it compensates for inaccurate modelling of the real
>response functions.  I've already mentioned problems with lenses, but
>I would also expect film's response to be more strongly peaked than a
>Gaussian because grain has hard edges and tend to clump together.
>
>    If all this is covered in the references you gave, say the word
>and I'll shut up and wait for inter-library loan to do its stuff.  if
>not, I'd be interested in your comments.

This is the other quotation, where there quite a bit of equivocation at the end.

I'm sure there are better references, but I don't have them at hand. But there are real problems in appling the mathematical treatment to film near the resolution limit.

p. 22-24 (minor editing required)

"Figure 4.1 shows some typical spread functions for a lens, film, and an image tube. The spread function is a plot of the illuminance of the image as a function of distance in the image plane. Actually each element of the sensor system will have a series of spread functions tor different positions off the optical axis. And, in general, only a spread function for a point on the optical axis will be symmetrical.

If, instead of a point, our optical system images a line, then we obtain the line spread function. We can consider the line spread function as a series of point spread functions added together to form the line. Knowing the point spread functions for a sensor system, we can plot the illuminance of any image by adding together the point spread functions that correspond to the points in the object or target. This summation process can be represented by what is called a convolution integral. To illustrate, if we have an object defined by a function o(x1) and a spread function s(x2), then the illuminance of the image i(xi) is given by the convolution for these two functions:

i(x1 = integral from -inf to + inf of s(x2)o(x1-x2)dx2

For simplicity, there is unit magnification between object and image.

Figure 4.2 illustrates the convolution integral. The substitution of the argument x2) for (x1) means that the function is translated or scanned along the other function. That is, each point in the object is traced through to the image, then the spread function is scanned over the image point. If the object is a line or one-dimensional then the output is as shown in Figure 4.2. If the object is two-dimensional, then a three- dimensional spread function is scanned over each of the two directions to give a three-dimensional output.

In reality, the spread function is likely to vary for each image point and is
probably non-symmetrical to complicate the mathematics hopelessly. However, it we accept these arbitrary assumptions and simplifications they provide a relatively simple but powerful analytical technique."

I haven't gone through the book in detail recently. Near the end, the author devotes a few pages to describe problems when including film MTF curves with other data to "model" a system. This isn't nearly as straightforward as the typical photography text or article would lead one to believe.

Whenever I use formulas to make system "predictions" regarding depth of field, enlargment size, etc., I use a simple empirical model which gives the best fit to a collection of carefully acquired data. I think that far more reliable than using numbers supplied by different manufacturers derived under conditions different from practical usage, and formulas which may or may not give accurate representations of reality.

Date: Fri, 08 Sep 2000
From: "Nicholas O. Lindan" [email protected]
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Final Resolution - the real answer?

Struan Gray wrote:

>     0n-axis, a properly focussed, diffraction limited lens will have
> an Airy error function whose 'width' is typically taken to be the
> position of the first zero.  If you define a Gaussian with the same
> 'width' it looks much fatter, because Gaussian widths are defined a
> relatively long way up the peak.  If you just plug the Airy width into
> the 1/R^2 formula you will automatically over-estimate the error (and
> hence underestimate the resolution) because the implicit model
> Gaussian you are using is simply too fat.
>
>     Imagine trying to predict a system resolution by actually
> convolving the functions.  If you convolve a Gaussian and an Airy
> function of the same width (defined conventionally) you get a curve
> which is almost exactly like the original Gaussian. 

I think this is it. The convolution of a sinc [sin(x)/x] function is very close to the convolution with an impulse if the sinc is of higher frequency than the components of the test signal.

If the resolution is higher for one component then that component will drop out of the resolution formula and it matters little which power is used in adding the resolution:

1/(1/10 + 1/100) = 9.09
1/(1/100 + 1/10,000)^.5 = 9.95

and the results are within 10%.

So, I cast my vote that 1/r is the right choice. Sorry for earlier confusion (no pun intended).

And in cases where the film resolution is far higher than the lens one gets into the problem of the resolution bars being at the same frequency as the lobes on the Airy disc and the lens appearing to have twice the resolution it really has.

--
Nicholas O. Lindan, Cleveland, Ohio [email protected]
Technical Management Consulting & Engineering Services:

[Ed. note: Mr. Erwin Puts is a noted lens tester and author of Leica and related photo articles and resources such as CD-ROMS...]
From Leica Mailing List:
Date: Sun, 5 Nov 2000
From: "Erwin Puts" [email protected]
Subject: [Leica] APX25 replacement

Is there life after APX25? A scientifically conducted study by several German authors in 1990 gave these results. Technical Pan had a resolving power of 250 lines per mm and APX25 of 180 lines per mm. These were the best. A step below these two were almost on equal footing: PanFPlus and Tmax100, with a small finegrain edge for PanF. These data in another notation. TP and APX could resolve details of a width of 4 to 5 micron.

That is the maximum definition for a lens like the Apo-Summicron-R 2/180, which defines the practical upper limit for the moment. The recently designed Leica lenses however will be slightly below this level (in the region of 7 to 10 micron), where the PFP and TMX are located. Older Leica lenses have a limit at 20 micron generally speaking.

My own comparisons (prints at 20x and microscope analysis at 100 and 400 times enlargement) do indeed show a theoretical advantage of the TP and APX25 emulsion.(Both however are challenged by the new BW film, I am currently investigating).

For most situations, the PFP would be a worthy replacement of the APX25 with the additional plus of a full stop more speed and it is a true speed of ISO50. Because the APX25 has a fairly steep curve, the sharpness impression is excellent. The PFP however has a moderately steep curve, giving a smoother graded tonality.

The sharpness impression is a bit less, but even so an enlarger lens like the Apo-Rodagon-N 2.8/50 would have to perform at its best to get this level of detail on the print at 20x enlargement.

You could try to do a comparison shooting with some of your APX25 films and the PFP to see where the differences are and how important they are.

For most of my purposes I can easily switch between APX25 and PFP. If the problem is demanding I have TP with CC40 filter to compensate for the enhanced red sensitivity.

Erwin

[Ed. note: 900!!! lpmm films...]
From Rollei Mailing List;
Date: Thu, 23 Nov 2000
From: [email protected]
Subject: Subject: [Rollei] SEMI-O/T: Slow-speed, fine-grain B/W film now history?

Regarding Efke the German importer, www.fotoimpex.de, has announced on his website that they won't get film until October/November, because the factory is moving to the outskirts of Zagreb. Probably it takes longer than anticipated to resume production.

Generally speaking, b&w film is a niche product and traditional slow speed film is a niche within this niche. And I read on www.fototechmag.com "that recent improvements in 100-speed black-and-white films [Delta/Tmax] have made slower-speed films increasingly irrelevant".

On the other hand, a small German firm has announced a new, utlra-high resolution film (www.gigabitfilm.de). 9x12cm and 4x5" is already on sale (ASA 25, 900 lp/mm, DM 170 for 50 sheets including special developer). The 35mm version (ASA 40, 720 lp/mmm, DM 17 per roll including developer) is announced for mid December. If only 20% of these company's promises are true, and beta-version tester confirmed it meets 100%, Agfa bailed out just in time. But no 120/220.

Hans-Peter

From Rollei Mailing List;
Date: Fri, 27 Apr 2001
From: Richard Knoppow [email protected]
Subject: Re: [Rollei] Lens Resolution, Old and New Lenses

you wrote:

>David Morris  wrote
>
>> Someone over on the Cosina/Voigtlander list wrote in
>
>> to say that the  1960s Tessar on a Rollei T roughly
>> resolves twice as much as a Tessar of the 1930s.
>> Similarly, todays lens designs might resolve three
>> times as  much as a 1930s lens.  One reason for this
>
>> is the ever increasing resolution of films over time
>> to which lens designers have responded.  Is this
>> broadly correct?
>
>Since I was the one who made the comment on the other
>list, I'll expand a bit here so that folks don't need
>to reconstruct what I may have meant.
>
>The Rollei-T vs 1930's Tessar comment is based on the
>tests I posted here a couple of years ago; sample size
>was one T and two 1930's vintage 'flexes.  This was
>system resolution -- Pan-F developed in PMK; at best
>center resolution the T matched the Planar on my F.
>The 1930's Rolleis do much better than the two Ikonta
>Tessars I checked (no surprise to anyone on this
>list); it is also no surprise that the Tessar was
>better than most 1930's lenses.
>
>The film comments are more objective.  Kingslake in
>several articles from the 1940's mentioned that
>Panatomic-X resolved just over 50 lppm.  Published
>figures for Delta-100 and TMX today are around 200.
>
>Marc, do you have the Zeiss factory specs on
>resolution from the 1930's?
>
>=====
>John Lehman
>College, Alaska USA

Some care must be excercised in comparison of old and new resolution numbers for film. The conditions for the resolution testing may not be the same. The target contrast is very important, resolution will be significantly higher for 1000:1 than for 30:1 contrast. Where a resolution number is given alone there should be some specification as to how it was determined. Modern films often have modulaton transfer curves published, which provide much more information.

Also, somewhere in one of the Kodak publications, its mentioned that resolution numbers pre about 1950 are inaccurate due to an error in determining the resolution of the special lens used. After that time a different lens and fixture was used. I think this resulted in higher numbers but don't clearly remember.

Of couse this doesn't apply to other manufacturers, but most of them did not publish any kind of resolution or granularity data at all.

FWIW, here are some resolution numbers from the 1943 edition of the _Kodak Reference Handbook_ and for comparison from the 1949 film booklet with the "redermined values.

Film                                           1943    1949
Verichrome (not pan, the old stuff)             45 l/mm 90 l/mm         
Plus-X Roll film                                50 l/mm 95 l/mm
Super-XX Roll and pack                          45 l/mm 90 l/mm
Plus-X Pan 35mm                                 55 l/mm 95 l/mm
Panatomic-X 35mm and Bantam                     60 l/mm 100 l/mm
Super-XX 35mm and Bantam                        50 l/mm 90 l/mm

Resolution is not given for sheet films.

All the 1943 tests were made using D-76 full strength.

The 1949 tests were made using D76 for Plus-X 35mm, Super-XX 35mm, and Panatomic-X and DK-60a for the others.

Here are some modern films from Kodak data sheets. I am giving the 50% point of the MTF curves. These may not be comparable with the above.

Plus-X Pan 35mm and Roll Type PX and PXP        55 l/mm D-76
Tri-X Pan 35mm and Roll Type TX                 52 l/mm D-76
Technical Pan                                   150 l/mm   Technidol
Technical Pan, Later data sheet                 100 l/mm   Technidol Liquid
T-Max 100 TMX                                   120 l/mm   D-76

Note: The above for T-Max 100 is from the MTF chart. Resolving power is given in the text as 200 l/mm for a 1:000 target and 63 l/mm for a 1.6:1 target

Both Technical Pan and T-Max have thin single coated emulsions with little internal light scattering. The older films all are double coated and have considerably more scattering. This accounts, at least in part, for the improved resolution.

----
Richard Knoppow
Los Angeles,Ca.
[email protected]

From Rollei Mailing List:
Date: Fri, 27 Apr 2001
From: Richard Knoppow [email protected]
Subject: Re: [Rollei] Lens Resolution, Old and New Lenses

you wrote:

>you wrote:
>>
>>  Also, somewhere in one of the Kodak publications, its mentioned that
>resolution numbers pre about 1950 are inaccurate due to an error in
>determining the resolution of the special lens used. After that time a
>different lens and fixture was used. I think this resulted in higher
>numbers but don't clearly remember.

Should have read this over before posting it. The info was from one of Kingslakes books on optics. The original lens was designed in the late 1920's or early 30's. Kingslake says that they discovered it had unsuspected aberrations so a new testing lens was designed and constructed.

I think this is partly a reflection on the tremendous rigor Kingslake provided when he took over the optical department in 1938. It is the lenses from Kingslake's period which are so very excellent.

I believe a paper was published on the design and construction of both the new lens and a new camera, or test fixture, it was used in but I don't have a ready citation.

Obviously something drastic was wrong with the original lens, the resolution numbers are nearly doubled. To get that much change the lens resolution must have gone from about equal to the film to many times greater since the overall resolution is approximately equal to 1/T = 1/F + 1/L Where T is total resolution, F is film resolution, and L is lens resolution.

Kingslake describes a special lens with 500 lp/mm resolution, which is, I think the above lens. This sort of resolution is possible with white light and very careful design but is more easily possible for single wavelength lenses.

----
Richard Knoppow
Los Angeles,Ca.
[email protected]

From Contax Mailing List:
Date: Wed, 4 Apr 2001
From: [email protected]
Subject: RE: [CONTAX] Med Formt

"MF lenses don't have near as high MTF as 35mm lenses (simply because of optical properties that go hand in hand with having to cover a larger image plane), so though you have a larger negative, you actually have to use "more" of it to equal 35mm 'quality'."

True...most of the time. But if one compares exceptional lenses, such as the 100 Makro for 35mm, or 120 for 645, they both will outperform the film itself. In that case, then film size alone is indeed the deciding factor. On lenses the quality of Zeiss (and I'm talking about both 35mm and MF lenses), the MTF factor will still be considerably offset by the larger film size. MF always wins! (Unless one has a Holga, I guess.)

D.

[Ed. note: thanks to Dave Mason for sharing these comments...]
Date: Sat, 21 Jul 2001
From: [email protected]
To: [email protected]
Subject: Re: How much sharpness is enough?

Bob -

Interesting to speculate on the effect of better film.

I've read that it takes ~3 photons to expose each grain, irrespective of grain size. That suggests a deterministic relationship where a certain lens diameter (absolute diameter, irrespective of focal length) transmits an unvarying number of photons from a fixed subject in a fixed exposure time.

Holding subject brightness and negative density constant, and the relationship between focal length and film size constant, graininess in the final print should be constant, irrespective of film size. If you use a larger film format you have a dimmer image so you need faster film to hold exposure and negative density constant. To capture the same 3 photons/grain you need proportionally larger grains; actually the negative needs the same number of grains regardless of size!

Perhaps the true advantage of large format is that for a constant lens diameter, aberrations tend to be better-corrected in a lens of longer focal length and smaller f-ratio, assuming constant complexity and cost. This makes it practical to use a lens of larger absolute diameter and expose more/smaller grains, so the negative ends up with more information (higher resolution, less grain, smoother tones). However this is only possible when you aren't constrained by depth of field, as preveiously shown.

Improved film with a better speed/grain tradeoff envelope seems to accomplish the same thing.

- Dave M

Robert Monaghan [email protected] writes:

> the big next step in film is probably the 10X faster for same grain;
> see http://www.cnrs.fr/cw/en/pres/compress/emulsionsphoto.html
>
> a thinner emulsion, or individually color sensitized grains might
> raise
> film in color to today's Tmax and higher lpmm 

Date: Tue, 31 Jul 2001
From: [email protected] (VT)
Newsgroups: rec.photo.equipment.35mm
Subject: Re: Some lens tests (Leica, Nikon and CV)

[email protected] (Robert Monaghan) wrote:

>provia 100 should do circa 60 lpmm in real world (1.6:1+) contrast
>situations, see http://www.smu.edu/~rmonagha/mf/lenslpm.html per mfg data

Bob,

A 1.6:1 contrast ratio is actually less about 2/3 of a stop difference - real world scenes tend to have higher contrast than that - that's why the film manuafcturers state that this is the _low_ contrast resolution.

The high contrast resolution figure at 1000:1 is equally unrealistic - that's close to 10 stops - although film can commodate over 11 stop range - other than edges, details hardly ever attain such high contrast (other than in artificial test target conditions).

So the only thing we can really say is that the "practical" resolution of film is somewhere between the low contrast resolution and the high contrast resolution - that's why they are given - so one can get a range.

So Fuji Provia 100F has a "practical" resolution somewhere between 60lp/mm and 140 lp/mm.

Fuji Velvia 50 has a "practical" resolution somewhere between 80lp/mm and 160lp/mm.

>it is very hard to break the 40 lpmm barrier, as Erwin Puts has noted;
>http://www.smu.edu/~rmonagha/mf/limits.html even on tripod, 50lpmm is  hard

This I would agree with even taking modified figures from above and guessing at a "practical"/between resolution - it is still hard to break the 50lp/mm "barrier" when the combined ssytem resolution (lens+film) is used - then degradation due to less than ideal conditions (handheld etc).

Therefore I think you can actually still be "generous" in film's practical resolution without losing the essence of what you are tryiing to say.

best regards,
--
Vincent
Remove HAT when replying

Date: Mon, 06 Aug 2001
From: "Christopher M Perez" [email protected]
Newsgroups: rec.photo.equipment.medium-format
Subject: Re: Lens Sharpness

I'd heard the same things over the years. So I bought a lens/'camera system' resolution chart (USAF from around 1950) and tested a number of lenses on various cameras in different formats.

Bottom line: - For a standard contrast scenes (6:1 light to dark), current fine grained films resolve between 120 and 140 l/mm. Color rendition and other 'curves' ignored.

- All lenses, whether 35mm, 120format, or 4x5 inches are diffraction limited to 123 l/mm (green light of a specific wavelength) at f/11. Or 174 l/mm at f/8.

I contend, based on these tests and practical experience of fiddling with cameras of various formats for 30 or so years, that the biggest limiting factor to final image resolution is the lack of a solid camera mount. ie: Use a tripod whenever possible.

Now, these 'tests' have ignored many of Contax' and Leica's and Voightlander's lens 'bouquet' comments and I'm not interested in stepping onto sacred territory. Lens designs can skew the final image in various directions. And my 'tests' have also ignored lens designs and coating that bring the contrast levels up, thus helping a final image appear 'sharp'. Finally, many lenses, when taken off the camera and tested in an 'aerial' fashion, will return higher l/mm readings. But such tests are, in the real world, completely bogus. They only show what a lens is capable of, and not your entire camera system (ie: mirror slap, bad film plane alignment, poor focusing, or image shift as you stop the lens down).

But it is the case, for me, that regardless of format, f/11 is the practical limit for final image sharpness. So you can see why I like to play with the larger formats: more image area on the neg/trans, which is the second most important contributor to final image sharpness (after using a tripod).

Does this help? Or is this confusing?

- Chris

"Gianni Piccoli" [email protected] wrote

> Hi, I have read that to obtain the maximum sharpness from the medium format
> lenses you need to close them at f11 or f16, in comparison with the  lenses
> for the 35 mm format that do their best at around f8, is this true ?
> Why this difference ?
>
> Thanks, Gianni Piccoli
> [email protected]

From: Zeljko Kardum [email protected]
Newsgroups: rec.photo.equipment.medium-format
Subject: Re: Questions about medium-format vs. 35mm
Date: Thu, 14 Mar 2002

"Q.G. de Bakker" wrote:
> MF photographers too are different. Most use more than one brand, and don't
> suffer from the same partisan attitude so overwhelmingly displayed in
> rec.photo.equipment.35mm.

True.

After fiddling with various MF cameras for a past couple of years, I
find out that with MF equipment it' much more important what film you
use than which lens/camera brand ;-)

Kardum
http://www.kardum.com/



From: "Jeremy 1952" [email protected]
Newsgroups: rec.photo.equipment.35mm
Subject: Ansel Adams and Lens Quality
Date: Thu, 23 May 2002


I recently borrowed a copy of Ansel Adam's classic book "The Lens" from my
public library and was surprised to read Adams' comment that today's lenses
were better than film's ability to record the images.  Essentially, he said
that almost any lens from a major camera manufacturer would do just fine,
and that it wasn't necessary to go on a quest to locate the finest lenses.
He gave an example of lenses he used that were over 40 years old, that
produced excellent results.

He was commenting in the context that good photographs are the result of the
photographer's skill, not necessarily of the lens.  While I agree with that
in principle, I have always thought that there was room for improvement in
lens quality.  Does Adams know something I don't know, or are we wasting our
time seeking out better lenses, as if they were the "Holy Grail??"



From: "Q.G. de Bakker" [email protected]
Newsgroups: rec.photo.equipment.35mm
Subject: Re: "35mm frame sized sensor"
Date: Sat, 25 May 2002


Mxsmanic wrote:

> That's why only amateurs buy Technical Pan, Kodachrome, and Velvia.

By the way: i've just unearthed an old post from my files in which Zeiss'
Kornelius Fleischer mentioned that testing at Carl Zeiss, in search for a
replacement for the Kodak Ektar 25 film they used at Zeiss as test film (200
lpmm), showed that Velvia (exposed at ISO 40) reaches 160 lpmm. Very good.
However, Kodak Portra VC, rated at ISO 160, is showing the same 160 lpmm!
I know, i know, one is a reversal film, the other isn't. But it illustrates
that the "slow = high resolution" dictum is rapidly losing validity.


From: [email protected] (Richard Knoppow)
Newsgroups: rec.photo.equipment.large-format
Subject: Re: Negative size or enlargement ratio?
Date: Wed, 22 May 2002


Robert Feinman [email protected] wrote:
>Folk tradition says that the larger image size allows for more detail,
>but if you actually do the experiment you may be surprised to find that
>the 35mm image is sharper since most small format lens have higher
>resolving power than large format lens. (Assuming same type of film,
>development and choosing lens stop not affected by diffraction).
>So I'd go with the less magnification theory.

 This simply isn't true.

  First, lenses for larger formats than 35mm are often of equal
resolution. Even old lenses like the Kodak Ektar is capable of more
than 60 lp/mm and may be capable of 100 lp/mm. Even some large format
lenses are capable of 60lp/mm.

  High resolution on the negative requires that both lens and film be
capable of high resolution. The final resolution is a convolving of
the two transfer funtions. As a rule of thumb is about 1/T = 1/F + 1/L
where T = Total resolution, F = Film resolution, and L = Lens
resolution. Where the two are equal the final resolution is one half.
Its easier to get higher print resolution from a larger negative
because, even if the lens is not as good, more of its resolution wil
be recoreded on the film. For 35mm the film is the limiting factor,
for 4x5 the lens is the limiting factor.

  Tonal rendering is something else. Its dependant on the number of
silver particles available to record the image. Where more particles
are available what we might call the tonal resolution is greater. In
this respect 35mm film is about at the limit of film to deliver good
continuous tone quality. Virtually any larger size negative will look
smoother.
....

---
Richard Knoppow
Los Angeles, CA, USA.
[email protected]




From: [email protected] (Gene A. Townsend)
Newsgroups: rec.photo.technique.nature,rec.photo.digital,rec.photo.equipment.large-format
Subject: Re: Film versus Digital: a summary of results
Date: Mon, 03 Jun 2002 

 "Roger N. Clark" [email protected] wrote:

>Hello.
>I hope I can shed some light (no pun intended) on the
>seemingly never ending film versus digital debates.
>
>I've summarized my research results and derived equations
>of digital camera megapixel equivalent as a function of
>film speed and film format.  I also present graphs of
>digital camera megapixel equivalent for different films
>with 35mm, 6x4.5cm, 4x5 inch and 8x10 inch formats.
>
>Feedback is welcome.
>
>Roger Clark
>http://www.clarkvision.com


Wow!  Looks like you did a really great job comparing digital color to
film color.  However, I dispute your simple method of dividing color
pixels by 3 to arrive at figures for monochrome.  It's too bad you
didn't compare monochrome film to monochrome pixels.  Monochrome film
has every bit as much information as color, since there is much more
image detail for monochrome.

A few comments...

You should probably have used a better quality fixed focal length 35
mm lens, and used it at more like F5.6 for these tests.   Similarly,
in 4 by 5, a better quality lens at around F22 would give closer to
ideal results.  This would effectively raise the effective digital
pixel count by maybe a factor of 2.

Your results show comparatively how digital and film color resolution
work when the final image is viewed on a computer screen, which is
certainly useful for web-based comparrisons.  However, a test
comparing the subjective quality of prints made from film versus
digital would be more useful to most who argue about these kind of
things.

In photography, it's been known for years that resolution is film
limited in  the smaller formats.  In the art world, resolution doesn't
matter at all.  In commercial work, everything is a compromise.

Much higher resolution films are available for monochrome.  For
example, I have used the old Agfa 8E75 holotest films (a "holographic
emulsion") which resolves around 3000 lpm, but has a film speed of
only about 0.5.  That is the fundamental reason that higher resolution
35 mm films don't exist.  Remember Kodak Ektar 25?  There is little
pracrtical use for films that are slower than 25.  Even a 25 speed
film is limited by things like camera shake, wind induced subject
vibrations, etc.  So, the lower limit of film resollution is limited
by practical real-world considerations, rather than theoreticals.

Regards,

Gene A. Townsend


From: [email protected] (Bill Hilton)
Newsgroups: rec.photo.equipment.medium-format
Date: 03 Aug 2002 
Subject: Re: K64 - would it sell?


>From: [email protected]

>I'll agree with you that E6 film is superior to Kodachrome with respect to
>color.
>
>One thing I can't agree on is that E6 film, Provia included, can match the
>resolution of Kodachrome.

This is something that's easily quantifiable.  Here are the resolution numbers
expressed in line pairs/mm for high contrast and low contast lighting
situations:

high contrast

160 lp/mm Velvia
140 lp/mm Provia 100 F
100 lp/mm K-25
100 lp/mm K-64

low contrast

80  Velvia
63  K-25, K-64
60  Provia 100 F

Also useful to look at the grain numbers, expressed as RMS.  An increment of
one means the grain is twice as large ...

8  Provia 100 F
9  Velvia
9  K-25
10  K-64

When you factor in the other Fuji E-6 advantages like fast turn-around, lower
cost, better reciprocity characteristics, and wider range of formats it's easy
to see why the Kodachromes lost so much market share in the 1990's.

Bill


From camera makers mailing list:
Date: Sat, 10 Aug 2002 
From: Robert Mueller [email protected]
Subject: [Cameramakers] High Resolution film  (Black and White)

http://www.gigabitfilm.de/download/datasheet_small_format.pdf
http://www.gigabitfilm.de/html/english/technical_information/descriptions.htm

Some of you enjoy pushing the art of photography as far as possible and the
above pages might provide help if you desire maximum sharpness for your
photos or for testing lenses.  (Many test results are limited not by the
lens but by the film; here is a way to largely eliminate this limit.)

Bob


From: Michael Briggs [email protected]
Newsgroups: rec.photo.equipment.large-format
Subject: Re: MTF and lp/mm
Date: Thu, 15 Aug 2002


Bruce Wilson wrote:
>
> I'm busy writing a database applet for recording shooting data, and it has a
> few pages for doing calculations. One of those pages does a quick estimation
> of final print resolution, in line pairs per mm, from the acceptable circle
> of confusion, diffraction (due to the aperture), lens resolution, film
> acuity, and film bulge. I've chosen to go with lp/mm (at 50% MTF) as my unit
> of calculation, being a little more accessible to the mind than pure MTF
> values while in the field.

With experience, this won't help much, and might distract you from my
important issues like composition.

>
> My problem is that while the system MTF is easy to calculate:
>
>     MTFsys = MTFcof x MTFdiff x MTFlens x MTFfilm x MTFbulge
>
> I'm not exactly sure how to handle the equivalent calculation in terms of
> lp/mm.
> Currently I'm using a geometric mean formula, using lp/mm as my units:
>
>     1/sys = 1/cof + 1/diff + 1/lens + 1/film + 1/bulge
>
> Trouble is, I can't get all the equations in a form that allows me to derive
> this exactly. I'm worried that there might be some weighting factors that
> creep into the calculation and skew the results from the equally-weighted
> geometric mean.

The reason that you can't derive the second equation exactly is that it
isn't exact.  Norman Korn discusses this:


http://www.normankoren.com/Tutorials/MTF.html

--Michael


From: [email protected] (Richard Knoppow)
Newsgroups: rec.photo.equipment.large-format
Subject: Re: MTF and lp/mm
Date: Thu, 15 Aug 2002


"Bruce Wilson"  wrote:

>I'm busy writing a database applet for recording shooting data, and it has a
>few pages for doing calculations. One of those pages does a quick estimation
>of final print resolution, in line pairs per mm, from the acceptable circle
>of confusion, diffraction (due to the aperture), lens resolution, film
>acuity, and film bulge. I've chosen to go with lp/mm (at 50% MTF) as my unit
>of calculation, being a little more accessible to the mind than pure MTF
>values while in the field.
>
>My problem is that while the system MTF is easy to calculate:
>
>    MTFsys = MTFcof x MTFdiff x MTFlens x MTFfilm x MTFbulge
>I'm not exactly sure how to handle the equivalent calculation in terms of
>lp/mm.
>Currently I'm using a geometric mean formula, using lp/mm as my units:
>
>    1/sys = 1/cof + 1/diff + 1/lens + 1/film + 1/bulge
>
>Trouble is, I can't get all the equations in a form that allows me to derive
>this exactly. I'm worried that there might be some weighting factors that
>creep into the calculation and skew the results from the equally-weighted
>geometric mean.
>
>Would any of you have some insight you could lend that might satisfy my mind
>on the matter?
>--
>Bruce Wilson
>http://chem.dynu.com/photo

 You can't really combine resolution numbers this way. You have to
convolve the curves. For a very simple example where only two limits
are involved one can take 1/T = 1/F + 1/L  Where T=total resolution,
F= film resolution, L=lens resolution. Or one can take the sqare root
of the sum of the squares. Neither is really correct since neither
takes into account the shape of the variousl MTF curves. The error
will become larger as more elements are added.

 I don't know exactly where to refer you, probably to an advanced text
on optics.

  Remember than the MTF curves for film and lenses are different in
shape and don't add in any simple fashion.

---
Richard Knoppow
Los Angeles, CA, USA.
[email protected]



From: Stacey [email protected]
Newsgroups: rec.photo.equipment.35mm
Subject: Re: Image size versus resolution and contrast
Date: Sun, 06 Apr 2003 


"William D. Tallman" [email protected] wrote:

>William D. Tallman wrote:
>
>> Please note that I have cross-posted to other related news groups. For
>> those who find these posts to be an onerous duplication, I suspect that
>> they might generate different threads of opinion, such that might prove
>> interesting. It is those differences I seek.
>
>Argh!! Wetware failure.. shoulda been separate posts!!! All replies are
>showing up here... oh, well... 
>
>What I'm getting from the responses is that larger format yields better
>images. Obviously, this is irrefutable. I guess the point of my query was
>lost because I failed to include an important detail: I've been out of
>active photography for over a decade now, and so haven't kept up with the
>changes as they've come. The exclusion of that point made my post
>somewhat... pointless(?), it appears.
>
>It would seem, then, that the best 35mm gear of the present moment will hold
>its own with less than the best medium format gear of 3 or 4 decades ago,
>here including film as gear. 

Bottom line is 3 or 4 decade old 35mm gear using modern film will look
about as good as modern medformat gear using 3 or 4 decade old film.
The film is what has advanced so much, that and autofocus/zoom lenses
in 35mm. My early OM-1 gear with prime lenses still produces as good
an image as a 2003 camera will using the same film.

I've been using a 1950's folding "welta" 35mm camera with a 50mm
zeiss tessar lately for fun and the results are as good as any 35mm
camera I've seen in outdoor shots. The cameras just haven't advanced
that much besides in convenience/automation.

What you have noticed is: if you were happy with medformat of 30-40
years ago, 35mm will produce about the same results today. Just don't
look at medformat or lareg fomat shot on these same films and you'll
be happy.


Stacey


From: [email protected] (Hemi4268)
Newsgroups: rec.photo.equipment.medium-format
Date: 22 Jun 2003
Subject: Re: Lenses - yesterday, today and tomorrow

>Film was a Fuji Provia 100F.

You should not use film to test a lens. What you are testing is the FILM.

Larry



From: Kennedy McEwen [email protected]
Newsgroups: alt.comp.periphs.scanner,rec.photo.equipment.35mm,rec.photo.equipment.medium-format
Subject: Re: New Nikon Coolscan 9000ED
Date: Sat, 15 Nov 2003 

Dennis O'Connor [email protected] writes
>"John Eyles" [email protected] wrote ...
>> >> you appear to have several scale factors out here. Firstly, at normal
>> >> photographic contrasts, no 100ASA colour film and lens combo can resolve
>> >> 100lp/mm, more like 80 max and 50 typical.
>
>See
>http://www.zeiss.de/de/photo/home_e.nsf/0/73d528c09b620a11c125697700548cd6?OpenDocument
>
I don't see anything on that page which actually specifies a resolution
of a 100ASA film/lens combination.

It correctly presents the resolution of diffraction limited optical
images, then quotes an esoteric limit for a specific obsolete 25ASA
film, for which the source of the data is not even referenced.

See the following link for how little quantitative information Kodak
offer concerning the resolution of this film's replacement!

http://www.kodak.com/global/en/consumer/products/techInfo/e40/e40Contents.shtml

Compare that bland statement to the resolution figures and MTF charts
provided by Fuji for Velvia, currently the highest resolution colour
photographic film available, at:
http://www.binbooks.com/books/photo/ic?l=5EE16AF7AF

Here you will see how specmanship really works: this film can only
resolve 160cy/mm at an original image contrast of 1000:1, which is great
if you are photographing backlit silhouettes! At a more typical
photographic contrast source of 1.6:1, the resolution limit is 80cy/mm
and even then, the MTF is plummeting into noise, as shown in chart 20.
This is the 50ASA version of Velvia. The 100ASA version is similar
performance as shown by the Japanese data sheet at
http://www.fujifilm.co.jp/ppg/datasheet/163AR096A.pdf

Hence MY (not John's) original statement of 80lp/mm maximum and 50lp/mm
more typical of a 100ASA film/lens combination is unchanged.
--
Kennedy


Newsgroups: alt.comp.periphs.scanner,rec.photo.equipment.35mm,rec.photo.equipment.medium-format
Subject: Re: New Nikon Coolscan 9000ED
Followup-To: rec.photo.equipment.35mm
Date: 14 Nov 2003

"Dennis O'Connor" [email protected] writes:

> See
> http://www.zeiss.de/de/photo/home_e.nsf/0/73d528c09b620a11c125697700548cd6?OpenDocument

Are you referring to:
"Today's high quality color films do reach resolutions in the region of
140 line pairs per millimeter with Kodak Ektar 25 leading the field at 200!"
?

This resolution is for a contrast of 1:1000. This is only
possible for back-lit test patterns. Even a black-and white
test chart well lit from 45� will not exceed 1:50.

So, for normal subjects, going beyond 100 lp/mm is
basically impossible with color film.

Regards,
Chris

--
Bokeh test images: http://www.bokeh.de/en/bokeh_images.html





End of Page