PDA

View Full Version : Best Film For Scanner Comparisons...



Scott Rosenberg
16-Mar-2007, 08:32
a friend of mine and i are laying out the test methodology we will employ in a study whereby we compare several scanners to determine which will hold up to enlargment better - smaller formats on dedicated film scanners or larger formats on flat beds. we are currently disscussing film choice and hoped to solicit some opinions from our friends on the forum.

to make color comparison back to the original easier, we want to use positive film. ideally, we'd like to use the sharpest film possible, so thought initially velvia. however, we're also going to be assessing each scanners range, so wanted a film with a little more latitude, which pointed us to Astia. however, i've read that Astia isn't as sharp as Velvia, so that made Provia seem like a good choice, as it's something of a comprimise on both ends... sharper than Astia, but wider that Velvia.

do you guys have any suggestions for film choice? is our reasoning thus far off-base?

any inputs would be greatly appreciated.

PViapiano
16-Mar-2007, 08:45
I've never heard that Astia wasn't as sharp as any other film. Astia is a great film with neutral color and it scans very well.

You have some beautiful images on your website...keep up the great work!

Michael Mutmansky
16-Mar-2007, 08:46
Velvia.

The density range of the Velvia film will tax the capabilities of any consumer scanner, and it will make the differences in DMAX capabilities the scanners the most readily apparent.

This is one issue that is not scalable. Even if you do very little (or no) enlargement, the limitation on DMAX will have an impact on the scanner performance. Many people focus on sharpness, and it is important, but I generall argue that for chromes, DMAX is in some ways equally important.

On the other end, I would recommend Astia (or maybe EPN) as possible the easiest film to scan. A person scanning small source materials may be able to produce a better result with Astia than a larger format scanned on Velvia. It depends on how small and how large.


---Michael

Michael Gordon
16-Mar-2007, 10:00
Scott: Astia has a lower RMS granularity rating than any of the Velvia's and greater resolving power, too. Take a look at Fuji's spec sheets for each.

Ted Harris
16-Mar-2007, 10:55
I agree with Michael .... Velvia. We have been using a Velvia transparency in the ever continuing scan around and the decision was initiall made, in part, for exactly the reasons Michael stated.

Scott ... one suggestion ... you might also want to have a "reference scan" made on a high end scanner as a point of comparison. Be careful when you do this thoughand make sure you have it made at solid high resolution ... at least 2400 for 4x5 and appripriately higher for smaller formats. If you take it to a lab/service bureau you are likely to get a scan measured in megabytes rather than SPI.

Bruce Watson
16-Mar-2007, 11:23
a friend of mine and i are laying out the test methodology we will employ in a study whereby we compare several scanners to determine which will hold up to enlargment better - smaller formats on dedicated film scanners or larger formats on flat beds. we are currently disscussing film choice and hoped to solicit some opinions from our friends on the forum.

to make color comparison back to the original easier, we want to use positive film. ideally, we'd like to use the sharpest film possible, so thought initially velvia. however, we're also going to be assessing each scanners range, so wanted a film with a little more latitude, which pointed us to Astia. however, i've read that Astia isn't as sharp as Velvia, so that made Provia seem like a good choice, as it's something of a comprimise on both ends... sharper than Astia, but wider that Velvia.

do you guys have any suggestions for film choice? is our reasoning thus far off-base?

any inputs would be greatly appreciated.

It sort of depends on what your purpose is. If you want to write up your results for general consumption you might want to use a tranny film like Velvia, a color negative film like 160Portra, and a B&W negative film like FP4+. The scanners will likely do some films better than others.

Else, if you are doing the study for your own consumption to help you decide what scanner to buy, you might get more meaningful results testing with the film(s) you plan to use the most.

Also, I suggest comparing the color of a scan to the original is a difficult and perhaps not terribly meaningful task. Unless your exposures are going to be made only in a studio where you control 100% of the light so that it has exactly the qualities that the film is balanced for, you are always going to have to make some color corrections post scanning. This means that no matter how accurate the scanner, you are still going to do some color correction. The only question left is how much.

If in fact your film will only be shot in the studio with lighting matched to the needs of the film, then you might want to consider profiling the scanners first. The profile should zero out the scanner as it were and give you a more accurate idea of how well the scan file can match the original. IIRC however, profiling only works with tranny films. There's too much variability of density with negatives films to make a good profile. It's been awhile for me, you might want to check and see if that is still true.

Ralph Barker
16-Mar-2007, 12:23
Rather than selecting just one film to "tax the Max" of the scanners, how about using a series of representative films? That might be more directly applicable to various readers' situations. You might also compare the scanned results to densitometer readings from the selected films, so as to more precisely quantify where the scanners fall short. If you also include spot meter readings from the original scene (of the same points chosen for densitometer readings?), that would complete the picture, so to speak, for the comparisons.

Scott Rosenberg
16-Mar-2007, 12:25
thanks for the inputs, fellas.

PV, thanks for the kind words about my work - that's always appreciated!

To the question of films and DMAX, we reasoned that a film with more DMAX would be the better test vehicle to see subtle differences in the capabilities of the test scanners. if the film is the limiting reagent in terms of DMAX, then we wouldn't be able to make any relevant conclusions about the relative performance of the scanners. it was initially our assumption that the best test of the scanners ability to resolve shadow and highlight detail would be a film that could record the widest range of values, thus enabling us to better determine even slight differences between the scanners.

here's an extreme example... if i want to determine which scanner can resolve more levels of grey, would not the most appropriate test target be the one with the most levels available? a target with two values, absolute black and absolute white, would not yield a lot of granularity in the results. however, a target with 256 levels of grey would bring into sharp relief the differences, even fine differences, between the scanners. more to the case in question, if the test shot is set up with a full range of values, from absolute white to absolute black, would Asita, which presumably will have a more subtle transition to total black, not make for a better test vehicle than Velvia, which will goto to detail-less black much more abruptly? if I’m hearing you guys right, this is faulty logic.

Ted, we’ve discussed including a reference scan and will likely do so. There’s no doubt that the highest quality will be attained from the largest film scanned on the best drum scanner. As a data point, that would be interesting to represent the pinnacle of what is possible.

Bruce, we originally were doing this to answer a couple questions that we’ve been wondering about for some time, however, once we started to flesh out the experiment, we saw that the results might interest a broader circle. So, we decided to cast a wider net and incorporate some things that would add validity to the study, but weren’t exactly germane to our specific situation – such as including scanners that we didn’t have as part of our workflow and using the most appropriate film to the study, instead of the emulsions that we most commonly use.

The test shots will be made in a studio setting under very controlled lightening. Every precaution is being taken to eliminate as many sources of variation as possible.

Thanks for the commentary thus far, this has all been very helpful.

Ted Harris
16-Mar-2007, 12:58
Scott,

A couple more things to consider:

1) If your purpose is to deterine if you can get better results from scanning a 6x7 or 6x9 transparency on a Nikon 9000 scanner than you can get from scanning a 4x5 transparency on a 'prosumer' flatbed scanner the answer without doing a huge amount of testing is very likely .... but at about three times the price of the most expensive of the currently available 'prosumer' flatbeds.

2) If you are thinking in terms of assisting others you might want to take a look at http://www.largeformatphotography.info/scan-comparison/ on this Forum's Home Page ... that is the 'scan aroound or Collaborative Scanning Exrcise' that I mentioned in my earlier post. There are many, many scanners currently in the comparison with more being added ... I will be sending the results of the Screen Cezanne to Leigh Perry soon and then the transparency off to get yet another scanner added.

3) If your decision relates strictly to scanning (e.g. you would shoot 6x9 with a roll film back in a Field or Rail camera) then this could be a worthwhile exercise for you. OTOH, I find using a rollfilm back a PITA and much perfer the larger image size of 4x5 ... although I do like the 6x9 aspect ratio and that is the reason I shoot more 5x7 than 4x5. If there are other aspets of the larger film size that appeal to you then remember that, while you will definitely see differences between the Microtek i800 on the probable low end the the Epson v750 on the probable high end, these differences will be incremental, small and subtle and nothing will be equal from scnner to scanner ... that is that while I am sure the v750 will give you the best reolution I am lmost s sure that it will not give you the best dMax or density range, that you will likely get from the i900. If you enjoy this sort of testing for the sake of testing and have the retail outlets near you that will allow you to do such testing (another iffy proposition) then go for it ..... if not, count your pennies and buy a scanner and start scanning ....

Just some more thoughts and not to discourage you but to give you some more info to chew on.

Bruce Watson
16-Mar-2007, 13:09
here's an extreme example... if i want to determine which scanner can resolve more levels of grey, would not the most appropriate test target be the one with the most levels available? a target with two values, absolute black and absolute white, would not yield a lot of granularity in the results. however, a target with 256 levels of grey would bring into sharp relief the differences, even fine differences, between the scanners. more to the case in question, if the test shot is set up with a full range of values, from absolute white to absolute black, would Asita, which presumably will have a more subtle transition to total black, not make for a better test vehicle than Velvia, which will goto to detail-less black much more abruptly? if I’m hearing you guys right, this is faulty logic.

It... depends. I know, I hate that too. Still...

If your scanner has the capability to set the limits on the log amp circuit (most drum scanners do this -- setting the black and white points for scanning directs the software to set the upper and lower limits for the log amp circuits), then your logic might indeed be faulty. Basically, such a scanner will translate whatever density range you present to it, be it 0.3 or 3.0, and convert that into a range of digital values (for an 8 bit scan, 256 levels, 12 bits -> 4096, etc.). That is, an 8 bit scan would result in 256 levels for a target with density range 0.3, 1.0, 2.0, or 3.0. All have 256 levels in the scan file.

If your scanner doesn't have adjustable limits on the log amp circuits (it would read from 0 - Dmax regardless of the density range of the film it's scanning) than your method makes more sense. In this case, the scanner would give maximum levels for a full range target and fewer levels for a restricted density range target. That is, an 8 bit scan might result in 8 levels for a target with a density range of 0.3, 50 levels for the 1.0 target, 125 levels for the 2.0 target, and 240 levels for the 3.0 target. (I'm just pulling numbers out of the air, I haven't done any of the math. I'm thinking it's likely to be non-linear).

So, rather than test with a high density range target, it might be more informative to test with targets of high, medium, and low density ranges. Then you might see differences in how many levels you get for a given density range.

Gordon Moat
16-Mar-2007, 13:17
I don't see how a scanning test would be useful with only one film being tested. Unless the readers only use one particular film, the results would be quite narrow in scope. It would seem that a range of films would provide a better comparison, especially considering the diverse nature of current films.

If you are testing resolution, it seem to me that using a few B/W films would be much better. You could include a comparison of the resolution tests done by C. Perez and others.

Kodak E100VS has a different colour palette than Fuji Velvia. If you are testing colour response, then it might be more useful to test both films. Fuji Astia 100F is very low grain, though perhaps testing a grainier film would also tell quite a bit; the chance of grain affecting noise would be one aspect.

In a work environment, the end printed medium is always a consideration. Depending upon printing requirements, a given scanner might be beyond the capability of a chosen printing method. One thing I have noticed with some consistency is that colour can be more judgemental than sharpness or resolution for many clients; a slightly soft image would not be as problematic as one in which the colours were off or one that was too dark in shadow areas.

The first thing I evaluate in scanning systems is the Dmax. Unfortunately in consumer level scanners the numbers are often way off reality. After checking Dmax, then I look to see if enough true (not interpolated) resolution capability is present.

Another factor is the camera and lens. There will be variation amongst lenses, and some medium format lenses providing greater resolution capability than some large format lenses. You might consider using a Mamiya 7 and one of their well regarded lenses. Another possibility is an optical test using B/W film in each system, then seeing how close each scanner comes to matching the film results. If you want to avoid using cameras, then Kodak (Q-60) and Fuji have targets you can use for scanner testing.

Ciao!

Gordon Moat
A G Studio (http://www.allgstudio.com)

Michael Mutmansky
16-Mar-2007, 13:58
The problem is that every slide film that I'm aware of exceeds the DMAX capability of every consumer scanner on the market, and most do so by a considerable margin. That means that no matter what, you will exceed the range of levels that the scanner can see, so in this respect, you initial logic is a bit off.

If you were to use B&W negatives, it would be possible to make some that have a DMAX lower than tat of the consumer scanners, but then there is a whole host of other problems that crop up.


So when you want to discuss shadow detail with a chrome, you for the most part are really discussing the DMAX value in an indirect manner, not the number of actual levels that can occur in the shadows.


---Michael

Henry Ambrose
16-Mar-2007, 15:32
As currently planned you are stacking the deck against getting results that have much value. Mixing up all the parts of the imaging chain in your test leaves many questions unanswered and leaves you with "when Bubba scans his 4x5 on his flatbed his Epson prints look better than my Nikon 9000 scanned prints from my roll film back."

I have some suggestions:

Isolate the particular characteristics you are testing for and then devise a test that uses a known value or reference material. Use the same reference for all scanners. To further nail down the reference you might have it scanned by the best drum operator you can find.

One thing to test might me sharpness. A good way to do this is to use a the same resolution target film for all scanners. That would likely mean the same roll film negative or positive in all scanners. (I assume you have no interest in 35mm).

For testing color rendition use one of the charts Gordon mentions above. Again, use the same one in each machine.

For density range and dMax use a step wedge, again the same one in each machine.

Stouffer has some targets that will help you.
http://www.stouffer.net/

After doing the above is the time to do your workflow to workflow comparison to see if you can beat Bubba's 4x5 film on the flatbed. :D

If you use

Kirk Keyes
16-Mar-2007, 19:08
Scott - I have a set of Wolf Foust IT8 films in several emulsions, Velvia 50 +100, Provia/Astia, Ekta... They are 35mm, but they have 21 step wedges, color scales, and known densities for each square in the target.

See http://www.targets.coloraid.de/ Look for Set S1.

You're welcome to use them.

Kirk