Why scan to compare?

[Dan Fromm]

In the thread " LF vs MF lens quality", Stewart Skelt reported on comparisons of scans of "a bunch of 35mm, 6x6, 6x12 and 6x17 shots." Michael Reichmann also scans film before comparing shots.

Pardon my ignorance of digital photography, but would y'all take pity on me and please explain how and why more can be learned from examining scans of negatives/transparencies than from direct examination of the negs/trannies? I've never got it, I still don't get it, and the practice is so prevalent that I fear I'm missing something important.

Thanks,

Dan

[Henry Ambrose]

Hey Dan,

I think its likely because more people have scanners than microscopes. Otherwise the ability to change views and magnifications with a click, side by side comparisons of more than one sample on screen together and the ability to quickly crop out equal film or subject areas are pretty good reasons as well.

Using scans for lens evaluation has its pitfalls since it requires proven and polished technique to compare across formats and maybe even scanners with confidence, but so does making prints to compare results from different lenses.

And if you're all set up to scan and print digitally then what else would you use than the tools at hand?

[Sheldon N]

I think it's because most people want to see for themselves. If you posted a web article with no examples to view, then its a lot harder for people to trust the author's own private conclusions.

The second reason is because most people are now printing their color film in some sort of digital process. Since the film is going to be scanned anyway, it's just an assumed step in evaluating the shot.

I agree with you, though, that the most accurate way to evaluate the film is with a high powered loupe or microscope.

[Marko]

What Henry said.

Even more people have computers than microscopes. The real appeal of Inernet is its immediacy and direcntess. It is one thing to read a verbal description of what some expert or the other saw under their microscope a couple of months before the magazine went to print and quite another to see it the same day on your monitor.

It also allows instant comparison of different opinions and claims. Yes, it is not perfect, but neither is film exposure and development.

[Capocheny]

I think it's because most people want to see for themselves. If you posted a web article with no examples to view, then its a lot harder for people to trust the author's own private conclusions.

I have to agree with you. People need to "see" with their own eyes before believing something... especially on the www.

However, the funny thing is that it's often times really difficult to make any qualitative judgements based on images posted on the net because the quality isn't very well represented on a computer screen!

Personally, I'm like Dan in that I prefer to look at a transparency or negative through a good loupe.

But, in essence, I think Henry hit the nail on the head - More scanners around than microscopes and every person scanning is instantly an expert scanner! :)

Cheers

[Ralph Barker]

Another point that is often (conveniently?) overlooked is that the scanning process inserts another set of optical variables into the comparison. That may or may not be misleading, even if the same scanner is used, theoretically applying the same level of degradation to both scans. But, as mentioned by Henry and others, some method of presentation is "demanded" by the Web-viewing public, even if scientifically inappropriate.

[Gordon Moat]

People have become very use to looking at images on their computer monitor. There is also the practical matter of sending transparencies around so people could look at them through a loupe or microscope. Anyway, though you might find this interesting:

http://www.porteous.net/test/digi.html

Basically, very simple, not very well done, and looks like crap on a monitor. However, I do find it interesting that this site tries to show what a portion of an image looks like through a microscope.

The only (somewhat) respected author who has not posted resolution images is Erwin Puts. He claims near 90 lp/mm from Kodak E100G and Fuji Astia 100F, though interestingly bashes the higher claims from Zeiss for various films. Probably many people would respect what Zeiss publishes, but they rarely have images or illustrations indicating their results.

I think a quantitative approach to imaging is a product of the internet in general, and somewhat influenced by a seemingly endless marketing approach of MegaPixels. The reality is when you investigate many of the big name photographers it is not always apparant what film, camera, scanner, or any other gear they might use. Their images can often stand above their technical merits, or lack of technical merits.

Ciao!

Gordon Moat

[sanking]

Another point that is often (conveniently?) overlooked is that the scanning process inserts another set of optical variables into the comparison. That may or may not be misleading, even if the same scanner is used, theoretically applying the same level of degradation to both scans. But, as mentioned by Henry and others, some method of presentation is "demanded" by the Web-viewing public, even if scientifically inappropriate.


I use a microscope to examine resolution. However, one issue that may not be immediatley apparent is that observation through a microscope will not necessarily show you how the negative will print. For this you might need to view the magnified image through a filter of similar wavelength to the spectral sensitivity of the process.

But on the issue of scanning, does anyone know of a standard protocol on scanning and presenting the results on the net that would be widely accepted as valid and that would be capable of illustrating actual differences in sharpness, and in resolution up to around 60-70 lppm?

Sandy King

[Ralph Barker]

. . . does anyone know of a standard protocol on scanning and presenting the results on the net that would be widely accepted as valid and that would be capable of illustrating actual differences in sharpness, and in resolution up to around 60-70 lppm?

I haven't seen anything, but that doesn't mean much. Each of the Web-based reviewers seem to follow their own procedures. It would be nice, however, if those who run such tests could agree on a standard protocol, and invite public comment prior to finalization - much the same way as ISO committees work.

We might be able to set up a private group sub-forum here for those discussions, if there's interest.

[Stewart Skelt]

I fear that I may have misrepresented myself in the post in which I was quoted. I am not scanning in order to compare. I am scanning in order to print. However, I am scanning a large number of pictures all taken with the same three cameras and with a limited number of lenses, on the same scanner, and looked at on the same monitor. Over time I have built up a good knowledge of which of my lenses are sharper than others.

[Dan Fromm]

All, thanks very much for the replies.

I understand the necessity of scanning for relatively easy and inexpensive sharing of images. No quarrel with that, even though I don't do it.

I'm still mystified by the practice of scanning film for purposes of measuring, e.g., resolution. Since I'm resolutely non-digital (NOT anti-digital), I don't have that option. So when I'm trying to decide which len(es) not to use, I shoot a more-or-less standard target, put the resulting neg/tranny between the light table and a 12x magnifier or under my little stereo microscope, and look at it. What would I gain by scanning? I see and, now, hear only of losses and difficulties and complications and lack of standardization.

Stewart, if I misrepresented your position I apologize.

Thanks again, best regards to all,

Dan

[Bruce Watson]

when I'm trying to decide which len(es) not to use, I shoot a more-or-less standard target, put the resulting neg/tranny between the light table and a 12x magnifier or under my little stereo microscope, and look at it. What would I gain by scanning?
AFAIK, nothing.

I do scan, but for evaluation of negatives, I put the film on a light table and use a loupe.

[Eric Leppanen]

I think there are two separate issues here.

For comparing lenses of a common focal length within a common film format, evaluating chromes or negatives using standard test targets and a high powered loupe is the cleanest solution with the fewest variables. The sole drawback is that other folks have to rely on the competence of the tester to accept the results. Scanning makes test results more accessible but introduces all sorts of subjective variables (what type of scanner? how many Photoshop adjustments? how much sharpening? etc.) that will dilute the value of the results for many folks.

Evaluating lens performance across different film formats is more problematic, in that the smaller format negatives need to be enlarged so that image scaling is normalized (I'm assuming use of real world test targets and not Air Force resolution charts). The most efficient method for doing this is via digital scanning, which introduces all the usual scanning/digital processing variables. Such tests can still be valuable, but obviously require much more scrutiny as to digital workflow.

If one must test using scanning, I think the best approach is to post raw scans on-line so that folks can process the results themselves through their preferred workflows.

[Ted Harris]

If one must test using scanning, I think the best approach is to post raw scans on-line so that folks can process the results themselves through their preferred workflows.


Eric, that is exactly why I believe that sharing the results over the Internet is so futile. What good does it do to compare severly downsampled or downsized files? It makes even less sense to download same and try to work with them. OTOH few will want to take the time to load and view a full sized file. Thoe on a dialup connetion are without hope. Not to mention how much bandwith would be used if, for example, I uploaded a 1 GB file and many people downloaded it.

Secondly, a totally separate point but one worth considering in discussions such as this ... there is no way to 'peer review' posts, we can only go by the work and reputation of the poster.

[Dan Fromm]

Eric, thanks for your comments. You wrote "Evaluating lens performance across different film formats is more problematic, in that the smaller format negatives need to be enlarged so that image scaling is normalized ... "

Pardon my denseness, but I don't get the point of what you call normalizing, at least from the point of view of deciding which lenses not to use on a format. Think of my shootout of a 210/9 Konica Hexanon GRII against a 200/4 MicroNikkor AIS. I concluded that I shouldn't use the MicroNikkor on my Nikons. Of course, it is so much more convenient to use that I still do, but the conclusion stands and I don't see how normalizing would have helped me. Would you please explain further?

Thanks,

Dan

[Eric Leppanen]

Eric, thanks for your comments. You wrote "Evaluating lens performance across different film formats is more problematic, in that the smaller format negatives need to be enlarged so that image scaling is normalized ... "

Pardon my denseness, but I don't get the point of what you call normalizing, at least from the point of view of deciding which lenses not to use on a format. Think of my shootout of a 210/9 Konica Hexanon GRII against a 200/4 MicroNikkor AIS. I concluded that I shouldn't use the MicroNikkor on my Nikons. Of course, it is so much more convenient to use that I still do, but the conclusion stands and I don't see how normalizing would have helped me. Would you please explain further?

Dan
Dan,

You're right. I edited an originally much larger post too severely, and it ended up incomplete and unclear. Let me try again.

I don't use artificial test targets specifically designed to measure lpmm. I use real world targets representative of my work (landscapes and architecture). Because I lack the equipment to configure my LF lenses on, say, my 35mm camera, I generally have to test my lenses in their native formats. This makes comparisons across formats difficult due to the differences in negative size (the scaling problem I mentioned).

Let's say I want to compare lenses from several different formats at infinity using a distant building as a test target. I focus each camera and lens on the building, and then examine the resulting negatives with my 10x loupe to see how well the building's features have been rendered. When dealing with formats as diverse as 35mm, 4x5, 8x10, etc. it is virtually impossible to compose the test shots so that the building occupies the same amount of negative area (e.g., make it a 1/2 inch tall in each negative regardless of format), as this would entail either moving the camera closer/farther to an impossible extent, or somehow configuring all lenses onto a common film format, which I lack the equipment to do.

Therefore, the best I can do is compose the test shots so that their composition is identically registered, i.e. 150mm lens on 4x5, 300mm lens on 8x10, etc. I then scan both images and display them side-by-side on my computer screen at an enlargement factor representative of my prints (the 4x5 shot being upscaled or "normalized" to twice the enlargement factor as the 8x10, so that both images are the same size on the screen), to see how each shot fared. Of course, this is not strictly a lens test anymore; it is measuring the total resolution delivered by the lens plus film format.

My point is that if one is testing lens resolution delivered for a given film area (lpmm), then such a test is best performed with film and loupe only, with no scanning involved. If one is testing total resolution delivered by a lenses in different formats using the full film area of each format (typically to determine maximum acceptable print size), then comparing scans normalized to a common enlargement size becomes necessary.

Does this make more sense? :)

Thanks!
Eric

[Dan Fromm]

Eric,

Thanks for your long and thoughtful explanation. Yes, it make more sense.

Y'r account of shooting, e.g., the same scene from the same position with 150 and 300 mm lenses, brings back to me why I switched from using a 100 marks/mm stage micrometer as a target in my acceptance tests of macro lense to using Edmund's USAF 1951 on glass target. Yes, I have a couple of microscopes. The stage micrometer gave me a good sense of which lenses just wouldn't do at all. The USAF 1951 let me do rankings. And the difference was that the USAF 1951 let me see where each lens that could separate marks 10 microns apart crapped out. I think a target with varying scales of fine detail would do the same for you, the problem is finding one when and where you need it.

I'm still bothered by the idea of asking which of a 4x5 shot and a 35 mm shot of the same scene, both shots composed as nearly identically as possible, will give the better 4x6 or 8x10 or 16x20 print. I didn't think there was much need to ask the question. I mean, that's why I moved up from 35 mm to 2.25 x 3.25 and will eventually go to 5x7.

Cheers, thanks again,

Dan