Scanner comparisson page and drum scan limits? [Archive]

SCHWARZZEIT

3-Sep-2010, 01:32

In other words at which ppi is the practical resolution of the film reached with the drum scan?....
It really depends on the lens and film combination you're using but first of all on what's practical to you.

Take a sharp Zeiss lens on high resolution 35mm film like Spur Orthopan UR (ASA 12) which is based on an Agfa microfilm emulsion, and you're able to get finer detail on film than any scanner can detect. Recently we did a test comparing this combination to Fuji Acros on 4x5" shot with the 110XL. The 35mm microfilm was surprisingly close in terms of overall detail. You would need something in the range of 14,000 spi optically to get the finest information from that microfilm digitized.
On 8x10" I found the best lenses offer diffraction limited resolution on film at f/22 in the center for high contrast detail. So getting all the detail into the file requires a scan of up to 3500 spi resulting in a 5.5 GB file. Of course for lower contrast detail the effective resolution on film is lower.
With their smaller image circle 4x5" lenses are usually slightly sharper. So 4000-4500 spi will probably be sufficient to get all the detail.

But in practical terms with normal modern fine grained films you'll get most of the information from the film with an optical scan at 2000-4000 spi provided that the scanner still has a high MTF within that range. It's only areas with high contrast detail that require higher sampling resolutions.
If, however, you're planning to print really large it's always better to scan at least to the desired printing resolution. The tonal transitions are much nicer when scanned optically from film than interpolated by software.

BTW when it comes to scanner resolution the latest Flextight X5 is in the same league as the best drum scanners for 35mm film. It was tested recently at ScanDig (http://filmscanner.info/en/HasselbladFlextightX5.html) and came out with an effective resolution of 6900 ppi. Unfortunately the text is in German only, but when you scroll down you can take a look at the scan sample of the USAF1951 target.

-Dominique

Lenny Eiger

3-Sep-2010, 09:38

In other words at which ppi is the practical resolution of the film reached with the drum scan?....

It depends on the quality of the drum scanner, and its ability to set the aperture to match the grain size on the film. There is a huge difference between flatbeds and drums. Flatbed's scan blurry, and the image is then sharpened (sometimes quite successfully for the size). Drum scanner's start out sharp. You can get 8,000 ppi out of them, but of course, that depends on how you quantify it. There are a lot of variables, and where one of us is clear about one part of the equation we may be not so clear about another. What I can tell you is that things look better in a scanner capable of 8,000 (or 3 microns) vs one that is set to 4,000. In sharpness, with test targets, a Premier beats a Tango hands down. In real life scanning with real film it's hard to quantify, harder to prove and as sharpening gets better, sometimes hard to justify. However, the old rule of "can't get more than 4,000" isn't true, in my opinion.

I have been fortunate to have a drum scanner, I like to start out sharp. Contrary to what Dominique referred to, an Imacon/Hassleblad is closer to a flatbed scanner - in that it uses ccd technology. It has a better lens than most flatbeds and the results are better. But it does not match a drum scan.

There is one other issue that I will mention that is a pet peeve of mine. We constantly talk (ok, haggle) about resolution. We have a lot of sharp lenses these days, skilled camera operators who know how to tip a front standard (for example) and good film and developers. My scanner gets all the sharpness the film has, I zoom in to the grains and can see it. The amount of sharpness in a digital print from a scan of this type is limited by the printers ability to put down dots in a tight enough pattern, without the slightest banding, overinking, etc.

The real issue for me has been full tonal reproduction. There are some inherent conversion issues that plague the digital world. A PMT (photo multiplier tube) is capable of picking up a few photons of light inside the dark box. They are extremely sensitive, sensing about 20K steps per channel or 64K for RGB. The ccd's pale in comparison. Consider what you work with in Photoshop, where you have steps from 1 to 255. That's quite a dip in tonal representation. Getting "all of the juice" requires a strategic approach.

Many people make a scan to look like the print they are going to make. This is a mistake. I was doing some scans with a friend last night and we were playing with pulling out the tonal separations inherent in the film. As we raised and lowered the midtones, we could see some of the tones blending together. We moved the values until they separated, which may have been lighter or darker than the actual print in that area. The clear goal is to supply the raw materials (tonally separated) so that someone can make a great print from the scan (after a little adjustment).

As has been said many times on this forum and others, this is the reason that "a scan is not just a scan". The operator is the key vs the machine. Working with someone on a long term basis to pull out all the "juice" from your work is the answer to a lot of these questions, especially for people who want exquisite prints. (Or practicing one's self - with a good drum, if finances allow.)

There is nothing wrong with doing one's own scans. Anyone can learn, of course. However, while I am not sure your question is the right one to ask, there is certainly more than 2,400 in the film, on a couple of fronts.

This was a longer answer than the question, I apologize for the long-windedness and I hope the info is useful to some.

Lenny Eiger
EigerStudios

Ken Lee

3-Sep-2010, 10:37

If a large format lens delivers 60 lines per millimeter on film, we consider it pretty good (http://www.hevanet.com/cperez/testing.html).

Since there are 25.4 mm per inch, that is 60 x 25.4 = 1524 lines per inch. To effectively resolve 1524 lines per inch, we need a scanner that can resolve 3048 lines per inch, isn't that right ?

Given an image made with such a lens, if we scan at greater than 3000 spi, what are we getting ?

Or is my math wrong ?

Lenny Eiger

3-Sep-2010, 11:25

Given an image made with such a lens, if we scan at greater than 3000 spi, what are we getting ?

Or is my math wrong ?

Wen you have a device that can see in to the grains - and can reproduce them as samples you get what the film has. I have seen this kind of math used to describe things, way too often, but it simply does not match to real life...

Once again, the conversation is about sharpness. What about tonal reproduction?

Respectfully, I just don't think this is right, or the whole picture.

Lenny

Nathan Potter

3-Sep-2010, 11:58

Ken raises a critical point and one that has been nagging me. I think his math is correct if one talks about lines per mm on a high contrast target (that is near 100% contrast). But what about a target that has much less contrast say 50% or 25%, or even a sine function target?

Now if we want to resolve the fine density gradations between lines on a sine target we will need some degree of higher resolution than is simply implied by the target lines/mm - maybe a much higher resolution. The same logic also holds for small variations in density on film, that is dimensions beyond the resolving power of the taking lens. Another way of saying this; is it useful to scan at a resolution that will detect artifacts that emanate from a resolvable point on the original subject and that has been recorded on the film.

I wonder if this is something that Lenny is alluding to when he comments on the value of extreme resolution scanners that employ PMTs, like Aztec.

In fact is the issue of scanner resolution tied more to the resolving power of the film than to the taking lens?

Nate Potter, Austin TX.

Ben Syverson

3-Sep-2010, 12:16

Since there are 25.4 mm per inch, that is 60 x 25.4 = 1524 lines per inch. To effectively resolve 1524 lines per inch, we need a scanner that can resolve 3048 lines per inch, isn't that right ?

Given an image made with such a lens, if we scan at greater than 3000 spi, what are we getting ?
Ken, your math and assumptions are both correct, with one big caveat. 3048 is the absolute smallest number of samples necessary to represent 1524 cycles. In reality, various factors prevent your scanner from resolving every pixel perfectly. Some scanners will be able to resolve 1524 cycles per inch by increasing the number of samples (ie, scanning at 4000 DPI/SPI or beyond). However some scanners won't see any improvement, because the scanner optics or sensor are already limiting resolution at 3000 SPI. For example, my Epson flatbed will never resolve 60 lp/mm, no matter what the SPI is set to.

sanking

3-Sep-2010, 14:57

In fact is the issue of scanner resolution tied more to the resolving power of the film than to the taking lens?

Nate Potter, Austin TX.

In my opinion, no, the limit to resolution is always the weakest link in the chain. In the case of a scan, the chain is 1) film, 2) lens and 4) the optical scanning resolution. I assume optimum conditions to get maximum resolution, i.e. camera on tripod, lens at best aperture, etc.

If you scan at an optical resolution beyond the potential of the film and the lens you look deeper and deeper into the film and see finer and finer grain clumps. But this is not resolution, and it is not sharpness, though depending on how it is processed it may appear to produce sharpness, when in fact is is only an illusion of sharpness that could be achieved with other methods of processing.

When I write "optical resolution" I mean the true effective resolution, not the stated resolution in dpi. As others have pointed out there is a huge difference in effective yield between various scanners.

Sandy King

Ken Lee

3-Sep-2010, 15:06

If you scan at an optical resolution beyond the potential of the film and the lens you look deeper and deeper into the film and see finer and finer grain clumps. But this is not resolution, and it is not sharpness, though depending on how it is processed it may appear to produce sharpness, when in fact is is only an illusion of sharpness that could be achieved with other methods of processing.

Thank you - That confirms my experience when scanning 35mm film at 4800 spi with a Minolta Dimage Multi Pro scanner: once I exceeded the limit of the taking lens + film, all the rest was just... grain.

Nathan Potter

3-Sep-2010, 18:29

Well I can agree with what Sandy says but that was not quite my point. I put it poorly. So to explain another way.

If I have a very fine grain film where even clumps cannot be resolved by the scanner and in fact the resolution on the film is totally determined by the film, then is there some kind of advantage in scanning beyond the resolution imposed by the taking lens? The advantage being this kind of elusive "quality" that Lenny mentions about very high resolution scans.

Nate Potter, Austin TX.

Ken Lee

3-Sep-2010, 19:40

As Sandy said, "the limit to resolution is always the weakest link in the chain"

Let's say we are taking a photo of the earth from outer space, with a camera that can resolve 1 meter at most. Can we read the print on a newspaper ?

Ken Lee

3-Sep-2010, 19:43

"If I have a very fine grain film where even clumps cannot be resolved by the scanner and in fact the resolution on the film is totally determined by the film..."

Perhaps I misunderstand, but that sounds contradictory. If the film is so fine-grained that it's undetectable, then the taking lens (or the tripod) determines resolution, not the film.

Hovmod

4-Sep-2010, 02:28

SCHWARZZEIT

4-Sep-2010, 06:29

However, the old rule of "can't get more than 4,000" isn't true, in my opinion.
I agree. A scanner designed for higher resolutions usually performs better at lower resolutions than a scanner being at its limit at 4,000 spi. Any idea how the old Linotype-Hell scanners performed at their stated maximum resolution of 24,000 spi?

Contrary to what Dominique referred to, an Imacon/Hassleblad is closer to a flatbed scanner - in that it uses ccd technology. It has a better lens than most flatbeds and the results are better. But it does not match a drum scan.
I was only comparing the X5 in terms of resolution on 35mm film. Tonality is a different story. That's where I'd prefer a high-end drum scanner any time. I was comparing X5 scans with my own ICG scans at the highest optical resolution of 8,900 spi. I really have to say that the X5 was on par. I don't know if a Premier could do so much better when even the X5 is quite close to its 8000 spi spec. If you could show some sample how a Premier scan looks at 8,000 ppi maybe the difference compared to other scanners is obvious.

My scanner gets all the sharpness the film has, I zoom in to the grains and can see it.
Seeing the grain pattern in a scan does not necessarily imply that you're getting all the detail. The grain clumps in a single emulsion are of various size. To detect some detail hidden in the smallest grain clumps you would have to use a very small aperture which is going to cause grain aliasing from the larger grain clumps. Your choice of aperture is a compromise between maximizing the amount of detail and minimizing artifacts. This may not be a problem for large format film because the LF lenses are unable to transfer enough contrast in the high spatial frequencies for most films to capture. But think about the resolving power of the best 35mm lenses. Choosing a high res film for these lenses can really make a difference in terms of resolving power. Like in our tests with microfilm we could record more than 250 lp/mm on film. You would require a sampling spot smaller than 2 microns and a very high MTF at that frequency to digitize these fine structures.

The real issue for me has been full tonal reproduction. There are some inherent conversion issues that plague the digital world. A PMT (photo multiplier tube) is capable of picking up a few photons of light inside the dark box. They are extremely sensitive, sensing about 20K steps per channel or 64K for RGB. The ccd's pale in comparison. Consider what you work with in Photoshop, where you have steps from 1 to 255. That's quite a dip in tonal representation. Getting "all of the juice" requires a strategic approach.
May I ask where you got those number about the 20K steps per PMT channel?
As far as I understand the PMT signal generates an analogue current which is routed through a log amp before going into the A/D converter where it's digitized and the steps are introduced. But in general I agree that you start off with a strong signal from the PMT.
One thing though that made me wonder was a comment by Karl Hudson in the ScanHi-End newsgroup where he made a statement about the Tango being able to resolve down to 4 microns with a 15.9 micron aperture.

Also remember there has to be SOME light falling on the photomultiplier to get any kind of dynamic range. How many photons can pass through 3 microns? I have no idea but I'm sure a lot more pass through a 15 micron aperture ...and after all the PMT's are going to "white out" on clear drum and set the 6 volt peak signal there. Then it's going to block the light and adjust the op amps to zero volts offset there. What's in between is your dynamic range. The less light you have to start with on white-out, the less dynamic range your scan will have.
It got me thinking how useful a 3 micron aperture really is if it compromises the signal's dynamic range. Maybe this is another factor besides the grain aliasing causing the noise you get on 3 micron scans. How often do you actually use the 3 micron aperture?

Many people make a scan to look like the print they are going to make. This is a mistake. I was doing some scans with a friend last night and we were playing with pulling out the tonal separations inherent in the film. As we raised and lowered the midtones, we could see some of the tones blending together. We moved the values until they separated, which may have been lighter or darker than the actual print in that area. The clear goal is to supply the raw materials (tonally separated) so that someone can make a great print from the scan (after a little adjustment).
If you're working with 16 bit/channel, is there any advantage applying the tonal separations in the scanner software compared to what you could do with a raw scan in Photoshop?

Damn, with all these quotes the my original post went over 10,000 characters. Sorry about the length.

-Dominique

SCHWARZZEIT

4-Sep-2010, 06:48

In my opinion, no, the limit to resolution is always the weakest link in the chain. In the case of a scan, the chain is 1) film, 2) lens and 4) the optical scanning resolution. I assume optimum conditions to get maximum resolution, i.e. camera on tripod, lens at best aperture, etc.
It is not that simple that the system resolution is limited by the weakest link in the chain. It's rather a function of contrast of all the links combined, and the weakest link reduces the overall contrast more than the others. Please see below my response to Nate's post where I'm going to elaborate on that point further.

If you scan at an optical resolution beyond the potential of the film and the lens you look deeper and deeper into the film and see finer and finer grain clumps. But this is not resolution, and it is not sharpness, though depending on how it is processed it may appear to produce sharpness, when in fact it is only an illusion of sharpness that could be achieved with other methods of processing.
That's correct, but it's hard to define where the resolution of a lens and film combination stops because the variance is huge depending on the contrast at hand. When you're photographing a test target with a line pattern it may give you an idea what your capture system is capable of when it comes to 2-4 stops contrast. The great work of Chris Perez and Kerry Thalmann may serve as a base as well for LF lenses and TMX or similar film. And when you're scanning the MTF of the scanner needs to taken into account. A scanner that maxes out on a scanner test target at 60 lp/mm will probably be unable to detect some barely resolved detail on film at that frequency.

Ken raises a critical point and one that has been nagging me. I think his math is correct if one talks about lines per mm on a high contrast target (that is near 100% contrast). But what about a target that has much less contrast say 50% or 25%, or even a sine function target?

Now if we want to resolve the fine density gradations between lines on a sine target we will need some degree of higher resolution than is simply implied by the target lines/mm - maybe a much higher resolution. The same logic also holds for small variations in density on film, that is dimensions beyond the resolving power of the taking lens. Another way of saying this; is it useful to scan at a resolution that will detect artifacts that emanate from a resolvable point on the original subject and that has been recorded on the film.

I wonder if this is something that Lenny is alluding to when he comments on the value of extreme resolution scanners that employ PMTs, like Aztec.

In fact is the issue of scanner resolution tied more to the resolving power of the film than to the taking lens?

Nate Potter, Austin TX.
For real world photography it is an interesting question how a lens, a film and a scanner perform on a target of lesser contrast. A sine target gives a much more realistic result about the performance than a simple line target, but these are very expensive.

When I tested my ICG drum scanner with a high contrast scanner target it was able to resolve more than 160 lp/mm. But with a test target shot on film the effective resolution was much less because the contrast of the finest detail on film was too low for the scanner to detect at the highest spatial frequencies. Thus I'd take the lp/mm values from scanner tests with a grain of salt though it might work as a relative comparison between the machines. But it's really the MTF that is important. Unfortunately these are not published. A drum scanner's MTF depends on the mechanical precision, the optics and the effective aperture. A real world photograph has much lower contrast in high spatial frequencies. To detect the highest spatial frequencies captured on film you would need a scanner that has a close to 100% MTF on those frequencies.

Example:
Let's just say we're trying to record some detail at 60 lp/mm that has a contrast of one stop. At f/22 and 1:20 magnification ratio a perfect diffraction limited LF lens transfers 14.7% contrast for that frequency. A modern slide film like Fuji Astia 100F transfers about 38% contrast at 60 cycle/mm according to the published MTF. I'm taking a transparency film for this because it doesn't compress the tonal range like most negative films do. So our one stop real contrast in the scene becomes just 1/18 of a stop on film if everything else is perfect. The scanner needs a very high MTF at the 60 lp/mm frequency to detect that sort of detail. A high res drum scanner has the necessary optics and mechanical precision to capture such fine detail. But the higher you go with your desired resolution you'll see that even the best drum scanners have their limitations. If you take negative film with its compressed tonality it's even harder for the scanner to detect those very subtle differences in density where the finest detail is hidden.

As Ben pointed out oversampling helps to get a more precise image of the lower spatial frequencies. I can see it even in blurry areas with no detail at all that the tonal transitions are nicer if the scan was made at a higher optical resolution. If your desired print size requires a scan beyond 4000 ppi than I would always recommend to scan for the native resolution of the printer.
Last week I was scanning very old 35mm slide film from the 60s that is going to be blown up to 2 x 3 meters. I used the maximum optical resolution of the scanner at 8,900 ppi which is enough for a 105 ppi print at that size. The client told me the printer wants 195 ppi. I could have used the 12,000 ppi setting of my scanner but it is in fact only 12,000 x 6,000 so I found the 8,900 x 8,900 to be the best solution I could offer, also considering scanning hours and file size.
Lenny, would you have used the 16,000 x 8,000 setting of your Premier in a similar case?

-Dominique

sanking

4-Sep-2010, 07:19

BTW when it comes to scanner resolution the latest Flextight X5 is in the same league as the best drum scanners for 35mm film. It was tested recently at ScanDig (http://filmscanner.info/en/HasselbladFlextightX5.html) and came out with an effective resolution of 6900 ppi. Unfortunately the text is in German only, but when you scroll down you can take a look at the scan sample of the USAF1951 target.

-Dominique

The text came up in English for me! Guess something out there knows whether we speak English or German?

Sandy

Ron Marshall

4-Sep-2010, 07:32

Some scan comparisons (scroll down):

http://www.clarkvision.com/articles/scandetail/index.html#introduction

peter ramm

4-Sep-2010, 08:58

It is not that simple that the system resolution is limited by the weakest link in the chain. It's rather a function of contrast of all the links combined, and the weakest link reduces the overall contrast more than the others.
-Dominique
I am going back a long ways, but I believe system MTF is the product of the discrete component MTFs. Improving the weakest link is often the easiest way to get a proportional improvement.

More generally, the expert eye has always been the best judge. Scanner SNR is directly related to so many factors that providing a broadly applicable quality metric would be very difficult. I do not recall ever seeing a thorough analysis back when PMT and CCD scanners were relevant to some lucrative markets. People bought what made their data look good - as they do today I guess.

Practically, I believe a PMT-based system can be engineered to have better SNR at small apertures than any area-based detector. LN2-cooled or amplified CCDs can actually detect single photons and have some advantages (primarily speed), but a PMT-based detector is more sensitive (better SNR) at low flux levels.

I am sure the engineers on this board can put it better but here's my simple summary. At anything under NASA price levels, PMTs maintain better SNR at small sampling apertures. This yields better MTF = perceptually better rendition of intensity variation overall and less visible "grain" in light image areas. Drum scanners are a fast way to position a PMT over large areas so we like 'em for LF. Wish I had a drum to play with today.

Ken Lee

4-Sep-2010, 09:10

Some scan comparisons (scroll down)

What item in particular please ?

SCHWARZZEIT

4-Sep-2010, 09:26

The text came up in English for me! Guess something out there knows whether we speak English or German?

Sandy
That's great. Somehow I must have missed that they have the review text up in different languages.

-Dominique

SCHWARZZEIT

4-Sep-2010, 10:20

More generally, the expert eye has always been the best judge. Scanner SNR is directly related to so many factors that providing a broadly applicable quality metric would be very difficult. I do not recall ever seeing a thorough analysis back when PMT and CCD scanners were relevant to some lucrative markets. People bought what made their data look good - as they do today I guess.
It probably wasn't necessary to do a scientific analysis outside of an engineering level. There was the Seyold Report (http://www.kar.fi/Skannaus/pixelperfect2_seybold_2000_vol30_nro1.pdf) where a bunch of prepress scanners were evaluated.
I agree that the expert eye is the best judge. Looking at the scans from different machines and playing around with the files, trying to see where the limitations are is probably the best you can do to make a buying decision.
The scanner comparison page on this site is also a great resource to get an impression. But I guess some differences would be more obvious if the scans were done at a higher resolution. One more suggestion would be to host the original scans somewhere for those interested to play with the files. Today hosting space is much cheaper than it was when this comparison started.

Practically, I believe a PMT-based system can be engineered to have better SNR at small apertures than any area-based detector. LN2-cooled or amplified CCDs can actually detect single photons and have some advantages (primarily speed), but a PMT-based detector is more sensitive (better SNR) at low flux levels.

I am sure the engineers on this board can put it better but here's my simple summary. At anything under NASA price levels, PMTs maintain better SNR at small sampling apertures. This yields better MTF = perceptually better rendition of intensity variation overall and less visible "grain" in light image areas. Drum scanners are a fast way to position a PMT over large areas so we like 'em for LF. Wish I had a drum to play with today.

Thanks for your input, Peter. As Lenny pointed out the influence of the scanner operator should not be underrated.

-Dominique

Lenny Eiger

4-Sep-2010, 11:24

Any idea how the old Linotype-Hell scanners performed at their stated maximum resolution of 24,000 spi?

Sure. They didn't. For starters, I don't mean to pick on you, or make you defend yourself. Just in case you feel that way, this isn't personal at all.

I used to be very upset at Imacon's quoting of certain numbers that were actually out of range. They suggested their DMax was 4.8, I believe, which exceed the max on film, of 4.6. How did they actually get that result? I railed against them in forums for lying to us about their capabilities. That is, until one fellow took the time to explain to me how they got their result. There was a mathematical formula based up the probable capabilities of such a device with its components, etc. It was theoretical number. They were quoting theoretical while others were quoting tested values. It got worse, when Epson suggested it can do 6400 real resolution, which we all know is too high. Now its all just a mess of one kind of disinformation vs another.

In conversations with Aztek they said that their stepper motor was capable of 18,000 steps around the circumference of the drum. That means that they could have taken samples and produced a much higher number of pixels. Phil Lippincott said to me that he decided against this because he felt the other components could only do 8,000 and he would rather have delivered 8,000 real samples vs some other result, theoretical or otherwise. The long winded point I am trying to make is that there are two sets of numbers. One is the theoretical capabilities of the scanner and the other is the real resolution in real terms as measured by say, a piece of film.

I had occasion to work on an old Hell 299 (I think that's the number) and was probably the first person to make enlarged negs for platinum printing with a scanner - back in 1980. I'm familiar with that scanner. I also am familiar with the 3400 series, and their scans aren't up to par with later drum scanners.

I was comparing X5 scans with my own ICG scans at the highest optical resolution of 8,900 spi. I really have to say that the X5 was on par. I don't know if a Premier could do so much better when even the X5 is quite close to its 8000 spi spec. If you could show some sample how a Premier scan looks at 8,000 ppi maybe the difference compared to other scanners is obvious.

I have an assistant who did a lot of work with 4x5 scans on an Imacon before he found me and we spent some time comparing the scans. They aren't bad. But they aren't drum scans. They were a bit soft. I am curious as to what ICG you have. I'd love to see a scan off a 380... just to see how good it is.

I am happy to supply some samples. I'm not sure what format. It appears that one doesn't see it unless one is looking at the whole thing.... As much as I appreciate the efforts here on the scan comparison, I find it very difficult to make real sense out of it. When things are compressed that much, the differences disappear.

May I ask where you got those number about the 20K steps per PMT channel?

Haddon Stevens Lippincott. Phil passed away a couple of years ago and Haddon is now the head of the firm.

One thing though that made me wonder was a comment by Karl Hudson in the ScanHi-End newsgroup where he made a statement about the Tango being able to resolve down to 4 microns with a 15.9 micron aperture.

This is what's frustrating. There are 24,500 microns in an inch, roughly. One might not imagine that at 15.9 microns you would not resolve to 4,000, but only 1597.5. The micron settings are different for different scanners. I usually do chromes and b&w negs at 13, they start to get soft at 16. I wouldn't claim that I could do my best work at 16. But Karl might.... because the way the settings they have things are different. I have been told that it is fixed at 11 microns, then told it isn't. I don't know Karl personally and have no reason to doubt him.

On the other hand, some would say that you resolve on the amount of microns you have and I would say not. When I scan at 13, I shouldn't be able to get more than 2,000 resolution on anything. But that isn't how it works. I get much more. As you stated, one can't scan at 3 microns or you will anti-alias. But how does one explain the results? Either there is no more after 2,000 or there is a lack of understanding on the actual process. I would suggest the latter.

If you're working with 16 bit/channel, is there any advantage applying the tonal separations in the scanner software compared to what you could do with a raw scan in Photoshop?

I think there is. Especially with DPL. Can I prove it? Probably with time I could. However, I can tell you that if I get things right in the scan, things separate well and the prints are fairly easy to make. I just did something off a raw scan and I ended up with 14 adjustment layers. There is a huge loss in going to Photoshop. It's going from 20K separations down to 250. If you give it tonal ranges that are separated in the way you want it to be, there is a real benefit when the conversion happens. I am stating my opinion, rather than stating it as a real fact. It's whats worked for me so far. I'm still learning.

Lenny

SCHWARZZEIT

7-Sep-2010, 06:32

Sure. They didn't. For starters, I don't mean to pick on you, or make you defend yourself. Just in case you feel that way, this isn't personal at all.
I didn't take it as personal offense. On the contrary, I do welcome real honest opinions, and hopefully we can all learn from each other.

I used to be very upset at Imacon's quoting of certain numbers that were actually out of range. They suggested their DMax was 4.8, I believe, which exceed the max on film, of 4.6. How did they actually get that result? I railed against them in forums for lying to us about their capabilities. That is, until one fellow took the time to explain to me how they got their result. There was a mathematical formula based up the probable capabilities of such a device with its components, etc. It was theoretical number. They were quoting theoretical while others were quoting tested values.
Yes, it is the theoretical Dmax value for any 16 bit/channel workflow. It doesn't take into account any CCD noise issues. I haven't tested a Flextight scanner for real Dmax. I heard they are doing quite ok compared to the Nikon Coolscan film scanners. FWIW, a client of mine who owns an Imacon 646 told me that the drum scan I did for him of a Velvia 50 slide had much better shadow detail than he could get from his Imacon. But I haven't compared it myself.
Aztek has an honest approach by giving the real Dmax value in their public specs for the Premier.

It got worse, when Epson suggested it can do 6400 real resolution, which we all know is too high. Now its all just a mess of one kind of disinformation vs another.
I see your point, and of course I'd prefer the manufacturers to state some true optical resolution for their machines but it seems they rather give the specs for their stepper motors and let the customers in the dark. But drum scanner manufacturers are no real exception on this.
So in the end we have to see for ourselves how these machines perform and draw our own conclusions if the quality is worth the money.

I had occasion to work on an old Hell 299 (I think that's the number) and was probably the first person to make enlarged negs for platinum printing with a scanner - back in 1980. I'm familiar with that scanner. I also am familiar with the 3400 series, and their scans aren't up to par with later drum scanners.
These scanners were probably made for a different set of computer hardware. It's only for a few years since we're able to process huge files. It's still very time consuming.

I have an assistant who did a lot of work with 4x5 scans on an Imacon before he found me and we spent some time comparing the scans. They aren't bad. But they aren't drum scans. They were a bit soft.
I guess there are differences between the newer and older Flextight models.

I am curious as to what ICG you have. I'd love to see a scan off a 380... just to see how good it is.
I have a 370HS. When I bought the scanner from ICG I asked them about the difference between the 370HS and the 380. I was told that the 380 has a slightly faster top speed at 2,000 rpm on lower res scans. But I wouldn't see a difference in the quality of the scans. Exactly the same optics and components are used to build both machines. This came directly from ICG.
I could supply some scan samples if you like.

I am happy to supply some samples. I'm not sure what format. It appears that one doesn't see it unless one is looking at the whole thing....
I think a crop where you personally think that it's representative in showing how sharp the scanner sees the grain at 8,000 spi would be great. But I know what you mean that it's more impressive to see the whole scan.

This is what's frustrating. There are 24,500 microns in an inch, roughly. One might not imagine that at 15.9 microns you would not resolve to 4,000, but only 1597.5. The micron settings are different for different scanners. I usually do chromes and b&w negs at 13, they start to get soft at 16. I wouldn't claim that I could do my best work at 16. But Karl might.... because the way the settings they have things are different. I have been told that it is fixed at 11 microns, then told it isn't. I don't know Karl personally and have no reason to doubt him.

On the other hand, some would say that you resolve on the amount of microns you have and I would say not. When I scan at 13, I shouldn't be able to get more than 2,000 resolution on anything. But that isn't how it works. I get much more. As you stated, one can't scan at 3 microns or you will anti-alias. But how does one explain the results? Either there is no more after 2,000 or there is a lack of understanding on the actual process. I would suggest the latter.
I'm sure there there's something else than just interpreting the aperture size as the smallest resolvable spot. The aperture is a hole in a disc, so how does the micron number describe the hole? It could be the radius, the diameter or something else that defines the spot size. I asked this question several times and was always told that it is the diameter. If it is the diameter of the analysis spot than the smallest resolvable detail could be as fine as half that size because the spots can overlap. This way smaller details are resolved with some loss in contrast, just like diffraction spots.
Another approach could be that the analysis spot is optically magnified before it is projected onto the resolution aperture, and the solution is probably different for each manufacturer. It seems that we cannot draw conclusions on the absolute resolving power of the scanner just from the aperture size.
BTW, in Phil Lippencott's scanner test (http://www.scannerforum.com/) the Tango came out with an optical resolution of 4096 ppi.

I think there is. Especially with DPL. Can I prove it? Probably with time I could. However, I can tell you that if I get things right in the scan, things separate well and the prints are fairly easy to make. I just did something off a raw scan and I ended up with 14 adjustment layers. There is a huge loss in going to Photoshop. It's going from 20K separations down to 250. If you give it tonal ranges that are separated in the way you want it to be, there is a real benefit when the conversion happens. I am stating my opinion, rather than stating it as a real fact. It's whats worked for me so far. I'm still learning.
I think the difference may be that some corrections can be done more easily in a dedicated scanner software beforehand. Thus it's much easier to get to the print file from a preadjusted scan than working from a raw scan file. On the other hand I like the idea that I have the full unmodified gamut of a raw scan to work with. I only made a few tests where I compared a preadjusted scan with a raw scan after both files went through Photoshop. The results were very close, and I couldn't really decide which I liked better.
It's not that you have only 250 separations per channel in Photoshop. Internally a 16 bit file has roughly 16K values per channel. If you have 20K separations to work with in DPL then your Premier must have a greater than 16 bit A/D converter. Otherwise you would lose some steps during A/D conversion.

-Dominique

Lenny Eiger

7-Sep-2010, 11:26

So in the end we have to see for ourselves how these machines perform and draw our own conclusions if the quality is worth the money.

Yes, "you're on your own" is the response we are getting because there is no independent testing agency. Aztek's testing was apparently independent, but it is old. Their efforts were not supported by the rest of the industry, in fact one scanner maker told them to take off the results or they'd be sued. FLAAR, which attempted to be such an agency, or at least promote it that way is corrupt. Their endorsements are bought and paid for.

This said, there are a lot of factors. I have a scan sample someone gave me from a Scanmate 11,000 that should not be possible. The samples are too tight for their "stated resolution". Did the fellow do some post-processing - I don't know. The scan project here undertaken by Leigh Perry was an attempt to make a real comparison, but there is quite a bit of downsizing and post processing done. Really mediocre (being polite) scanners show up as quite reasonable on the web.

It's not that you have only 250 separations per channel in Photoshop. Internally a 16 bit file has roughly 16K values per channel. If you have 20K separations to work with in DPL then your Premier must have a greater than 16 bit A/D converter.
-Dominique

They have told me its a 16 bit. Must be the effects of rounding numbers up and down... for discussion.

Regarding raw vs adjusted at scanner scans- one thing the software does (DPL) does that's great is that they have a "CMS" file attached to each scan. This gets loaded into the scanner at scan time and adjusts the range of the scanner. This is very different from doing a raw scan, like most scan software does. The pro version allows you to create your own cms file/profile for each image. It's not needed for most chromes, and only some color negs, but its great for b&w negs. It's like having a different scanner for each image. At least that's what I've been told and the premise I am working on. It appears to be working that way, but there's plenty I don't know...

Lenny

Tim Povlick

7-Sep-2010, 12:35

I am reading this thread with great interest. One point of discussion I do not understand is separation. How does a 16-bit number have less than 20,000 "separations" ?

_ .. --
Tim

Lenny Eiger

7-Sep-2010, 12:51

I am reading this thread with great interest. One point of discussion I do not understand is separation. How does a 16-bit number have less than 20,000 "separations" ?_ .. --
Tim

Tim,
I am curious myself. According to latest Photoshop we are in 16 bit - so there should be 65,536 possibilities, if my math is correct. However, when we move a curve up and down, we only have 0-255 as choices. This would appear quite sledge-hammer-ish as a tool. In terms of b&w, I also don't think my printer can separate out that many tones. My gray ramps are perfectly smooth, but the resolution is low - 1440x1440. So if I only have the tones that can be generated with that resolution, I believe its quite limited. I don't know what the number would be... do you?

Lenny

Ken Lee

7-Sep-2010, 13:01

One has to ask whether any monitor can accurately render that many shades per channel, and whether we could even see them if it could :)

Lenny Eiger

7-Sep-2010, 14:51

One has to ask whether any monitor can accurately render that many shades per channel, and whether we could even see them if it could :)

Agreed. The key, IMO, is to get as many shades on the paper as possible. (I try not to imagine the monitor is of any use at all. I only look closely at the print, which is what open would do in a darkroom.) To get as many shades, one needs good ink, and good eyes. It helps a great deal if the scan's zones are separated... in my experience so far...

Lenny

Ken Lee

7-Sep-2010, 15:09

If I recall correctly, printer drivers for inkjet are 8-bit only.

This means that we have lots of room to get things right in the deep color space - before things get collapsed by the printer, the driver, the RIP, and ultimately, by the even more limited gamut of the paper/ink combination.

Tim Povlick

7-Sep-2010, 19:33

Tim,
I am curious myself. According to latest Photoshop we are in 16 bit - so there should be 65,536 possibilities, if my math is correct. However, when we move a curve up and down, we only have 0-255 as choices. This would appear quite sledge-hammer-ish as a tool. In terms of b&w, I also don't think my printer can separate out that many tones. My gray ramps are perfectly smooth, but the resolution is low - 1440x1440. So if I only have the tones that can be generated with that resolution, I believe its quite limited. I don't know what the number would be... do you?

Lenny

Hi Lenny,

Your math is correct, there are 64k possibilities (separations?). As to the photoshop, I think if you are in 16-bit mode the tone curve is presented to the user as 0 to 255 they are expanding it to 16-bit internally and I see your point that in so doing one has only a crude control over the curve that is 16-bit with an 8-bit control. If they allowed user input of a curve one would generate the desired curve to perfection, but that would be a 64k table to fill. Ouch. My only suggestion is be very precise with the curve tool in PS. Maybe someone can suggest a better suggestion.

I checked the Hamamatsu site concerning the PMT's spec's. I don't have the PMT number used in our Premiers but a PMT I looked at had the following features. If one took dark current as a "step" and maximum current as the max limit as the number of steps, then you ended up with a PMT that had 20,000 steps. Interesting result (matched Haddon's number). A more interesting was take the minimal value the PMT could sense and use that as step size. To my way of thinking this is the proper value to use. The PMT then had 6,000,000 steps. In otherwords, a 22-bit DAC could realistically be used to capture a color channel. That would yield 66-bit RGB. No way to print that :-)

_ .. --
Tim

SCHWARZZEIT

7-Sep-2010, 23:40

They have told me its a 16 bit. Must be the effects of rounding numbers up and down... for discussion.
Lenny, I confused the number of values in 14 bit and 16 bit. My mistake! 64K values is of course correct for 16 bit. So a 16 bit ADC is sufficient for the 20K separations of the PMT.

Regarding raw vs adjusted at scanner scans- one thing the software does (DPL) does that's great is that they have a "CMS" file attached to each scan. This gets loaded into the scanner at scan time and adjusts the range of the scanner. This is very different from doing a raw scan, like most scan software does. The pro version allows you to create your own cms file/profile for each image. It's not needed for most chromes, and only some color negs, but its great for b&w negs. It's like having a different scanner for each image. At least that's what I've been told and the premise I am working on. It appears to be working that way, but there's plenty I don't know...
Ok, I think I get how the CMS files work. On my ICG scanner there are basically just two hardware modes that define the range. One is called standard range which is good for all but very dense negatives and the other is called extended range where the full 4D+ range of the scanner engine is used. The hardware difference between these modes is that they calibrate to different values. Or in other words there's a difference in gain. I think the CMS files work similarly on your Premier but you can adjust them precisely to the density range of your original. Whether you really see a difference in the final print file is a different matter.

As to the photoshop, I think if you are in 16-bit mode the tone curve is presented to the user as 0 to 255 they are expanding it to 16-bit internally and I see your point that in so doing one has only a crude control over the curve that is 16-bit with an 8-bit control. If they allowed user input of a curve one would generate the desired curve to perfection, but that would be a 64k table to fill. Ouch. My only suggestion is be very precise with the curve tool in PS. Maybe someone can suggest a better suggestion.
On an 8-bit curve in PS you may be able to see the change of single values. On a 16-bit curve smaller adjustments would be invisible when the changes are lost on its transfer to the display device. If however you feel that you need finer adjustments on your curves you may reduce the opacity of each curve layer or fade the adjustment after applying the curve directly to the image.

-Dominique

Tyler Boley

8-Sep-2010, 11:21

sanking

8-Sep-2010, 11:33

really? You guys really find moving mid tone value, oh say, 124, up to 125 too sledge-hammerish? You really want a curve tool that has tens of thousands of steps because 124 to 125 was just overwhelming? You put those two prints next to each other and the difference is not only visible but far to great?
Man, I'm losing touch with these discussions more and more each day...
Tyler

I agree with you Tyler. Sounds like these guys are working on some planet other than Earth because they are discussing subtleties of tonal values way beyond the capability of the human eye.

Sandy King

Peter De Smidt

8-Sep-2010, 12:15

Recently, my neighbor asked me to photograph his house and then use Photoshop to colorize sections of it so that he could determine a paint choice. He gave me some paint samples to match.

My spectrophotometer reads to tenths, but photoshop curves/readouts are whole digits. Being off by a couple of tenths was clearly visible. Thus, for some things, such as very close tolerance color matching, Photoshop is a bit of a sledgehammer. I've never found this to be a problem doing regular printing.

Tyler Boley

8-Sep-2010, 12:16

by the way, "16 bit" Photoshop mode works in 15 bit + 1 unit, 32,769 values per channel.
Tyler

Bruce Watson

8-Sep-2010, 13:47

Really? You guys really find moving mid tone value, oh say, 124, up to 125 too sledge-hammerish? You really want a curve tool that has tens of thousands of steps because 124 to 125 was just overwhelming? You put those two prints next to each other and the difference is not only visible but far too great?

Man, I'm losing touch with these discussions more and more each day...

I promised myself that I wasn't going to add my voice to yet another thread like this (emphasis on another). Oy, failed again. But yeah, what Tyler said.

I would only add that the tones on the print are the result of the entire system, not just photoshop. I wonder how much an effect paper batch variability, ink batch variability, ambient temperature and relative humidity, etc., have on the tones. I suspect it's a significant fraction of the difference between 124-125.

So are you guys making new profiles for every print you make? Sorry, please don't answer that. If you are, I don't want to know.

I'll leave this thread with a story. I seem to remember a story Fred Picker told, went something like this. When he was a hungry young photographer, he worked (apprenticed?) for some excellent darkroom printers. And Fred, being Fred, could be a PITA at times. So one of the experienced old guys challenged him to print as many individual shades of gray as he could. Take as long as he needed, use whatever techniques he liked.

Turns out Fred was able to come up with something like 54 or 55 distinguishable shades of gray. Try as he might, that's where he seemed to top out. This surprised and impressed his teachers, but Fred was usually one to exceed other peoples' expectations, apparently.

My take on this is that the number of tones you can make might be interesting but it's only a small part of what makes a photograph.

Tyler Boley

8-Sep-2010, 14:05

Peter, forgive me but I'm not following... tenths of what? Most Photoshop tonal tools work in 256ths, that won't more than cover your couple of 10ths by orders of magnitude?
Tyler

Peter De Smidt

8-Sep-2010, 14:15

Peter, forgive me but I'm not following... tenths of what? Most Photoshop tonal tools work in 256ths, that won't more than cover your couple of 10ths by orders of magnitude?
Tyler

Sorry, I made a mistake. My spectro reads to hundredths. So for example, one paint chip measured L 95.26, a 0.39, b 1.76, but entering lab numbers in Photoshop is limited to whole numbers. Using L 95, a 0, b 2 did not produce a visually matching color.

Lenny Eiger

8-Sep-2010, 14:40

Ken Lee

8-Sep-2010, 14:52

Unfortunately, the printer and the paper/ink combination have a limited gamut, but before we squeeze the final image into that limited gamut, it's best if our tools distort the color space as little as possible, all along the way.

Not distorting, translates to retaining as many steps as possible.

Nobody is claiming that we can see or manage all those subtle steps.

We just want to retain as many as possible, so that things remain as "analog" as possible, as long as possible.

Greg Miller

8-Sep-2010, 15:01

Attached is a step wedge chart in 8 bit. Perhaps my eyes just suck, but if there is a huge difference between 2 adjacent tones (1 full step apart), then it is not apparent to me on my monitor or on paper. I'll provide the full res .psd file if anyone wants it.

Ken Lee

8-Sep-2010, 15:51

Greg Miller

8-Sep-2010, 16:26

The problem isn't when we view a complete 256-step wedge.

The problem is when we start out with a small color space, and proceed to make lossful adjustments. This introduces gaps, or banding. With further adjustments, banding become increasingly apparent.

256 tones is nice to finish with (and that's all we can ever get in the end, because printer dirvers are only 8-bit), but if we start out with only 256, the number of steps can only decrease as we perform adjustments.

On the other hand, if we start with 64 thousand tones per channel, we could lose 1000 steps and they wouldn't likely be noticed. If we start with 256, it doesn't take many losses to become apparent. It's a familiar look when (8-bit) JPG files are overly "corrected" by amateur digital photographers. That's why higher end cameras allow Raw capture, and modern imaging software supports it. Google 16 bit Workflow to learn more.

Performing adjustments in a large color space, prevents banding. That's why we convert to the smallest color space, last - just as we do with sharpening.

http://www.kenleegallery.com/images/tech/steps.jpg

Starting with 256 discrete tones, but working in 16 bit will not introduce many, if any, banding issues. I'm quite familiar with 16 bit.

Working in a large color space, such as Prophoto RGB, will encourage more banding, because the tonal neighbors must be spaced further apart than with a small color space (both color spaces will have the same total number of discrete tones) - your gamut is wider but the tones must be spaced wider as well. If you are working with a low gamut image (such as a foggy scene), you are better off working in a small color space, such as Adobe sRGB, because the tonal values will be closer together and have a lower probability of banding.

Peter De Smidt

8-Sep-2010, 16:26

<snip>
Performing adjustments in a large color space, prevents banding. That's why we convert to the smallest color space, last - just as we do with sharpening.

http://www.kenleegallery.com/images/tech/steps.jpg

Is that right? If we have a 16 bist per channel RGB image, aren't the number of colors/shades equal no matter whether we use sRGB or ProPhoto? Isn't what happens is that the number of steps are the same, but there's bigger jumps between the steps in ProPhoto, since the same number of steps has to cover a bigger area, as it were. If so, then aren't you more likely to see banding with the larger color space? I'm sorry if I have this wrong, as I'm no color space expert.

Oops. I see that Greg types faster than I do. :)

Greg Miller

8-Sep-2010, 16:27

[QUOTE=Ken Lee;626685]Performing adjustments in a large color space, prevents banding. That's why we convert to the smallest color space, last - just as we do with sharpening.

This is not correct (see my previous post). Working in a larger color spaces encourages banding. Working in higher bit depth discourages banding.

Ken Lee

8-Sep-2010, 16:48

Yes, I stand corrected. I meant bit-depth. Not color space.

My post should have read as follows

...Performing adjustments in higher bit-depth, prevents banding. That's why we convert to the smallest bit depth, last - just as we do with sharpening.

Greg Miller

8-Sep-2010, 17:12

And my conment was directed to the other comments about "124, up to 125 too sledge-hammerish". You have to look pretty carefully to detect the difference in 1 step on a step wedge. In a "real" photo, and without a direct A to B comparison, it would be pretty darned difficult for a human to reliably identify that a few pixels ended up 1 value off and had a few neighbor pixels with the same value. If an entire section of pixels all ended up with the same value that would be different, but I think it would be pretty rare for that to happen.

Tyler Boley

8-Sep-2010, 17:52

Guys, before you jump all over everyone-- there is some serious compression of tones that happens on occasion, at least. I'm not sure where exactly it occurs. I assume its in Photoshop, but it could be in the rip. There's a certain effect that I get when I use an 8x10 piece of film that adds an extreme amount of textural quality. I think it should be able to happen at smaller formats, certainly 4x5, given resolution of lenses, excellent film and developers, the best kinds of scanners, etc. All the stuff we all use.

I realize the eyes can only go so far, monitors only os far, etc. I continue to struggle to get more than I am getting from smaller film...

This doesn't have to do with world views...

Lenny

Lenny, two givens-

1) "some serious compression of tones... ...happens on occasion"
2) the Photoshop curves tool has control points in increments of 256, both
in and out, no matter the number of levels in the
file.

One has little, if anything, to do with the other.

another two-
1) world views
2) facts

Again, one has little, if anything, to do with the other.

As a note of interest perhaps, your multi K setups will indeed
differentiate between more than 256 levels of gray (given enough room on
the paper), unfortunately StudioPrint converts to 8 bit on the fly behind
your back, before continuing on with the magic it does. Don't ask me to
write the novel required to explain how I arrived at that. So, there is
one bottleneck, though I don't think the problem you mention is at that
point. Still, I'm in agreement that there should be no downscaling of
levels of gray anywhere in the data path if at all possible, but PS tools are a different issue.

Peter, I had no idea you were referring to LAB, that's an entirely
different can of worms, and you are right about the increments of control
points available with it in PS.
At risk of offending you as you may be well aware already, there are other
perhaps more useful tools. Despite it's silly Tron like interface, i1
Share (free) will let you measure in accurate LAB values and tell you the
equivalent RGB value in any selected working space, for example.
Tyler

Lenny Eiger

8-Sep-2010, 18:03

Lenny, two givens-

1) "some serious compression of tones... ...happens on occasion"
2) the Photoshop curves tool has control points in increments of 256, both
in and out, no matter the number of levels in the
file.

As a note of interest perhaps, your multi K setups will indeed
differentiate between more than 256 levels of gray (given enough room on
the paper), unfortunately StudioPrint converts to 8 bit on the fly behind
your back, before continuing on with the magic it does. Don't ask me to
write the novel required to explain how I arrived at that. So, there is
one bottleneck, though I don't think the problem you mention is at that
point. Still, I'm in agreement that there should be no downscaling of
levels of gray anywhere in the data path if at all possible, but PS tools are a different issue.

Tyler

There's part of this that is just interim, or incomplete thinking. I've been working for a long time to get to a fully textured appearance. I have been successful at getting there with an 8x10 camera. However, I believe I should be able to get there with a smaller piece of film - at least one as large as 4x5. Somehow it doesn't work that way - at least not yet. There's a smoothness I haven't gotten to. I'm grasping at straws to figure out where the bottleneck is. I have very smooth gray ramps in my environments but the magic is only there with the big camera - at least so far. More testing to come.

I do know that StudioPrint is 8 bit. They just told me, thankfully no novel needed.

My statement about world views had to do with the comment you made about "losing touch with these discussions every day..." It was an attempt at levity.

Lenny

sanking

8-Sep-2010, 19:40

There's part of this that is just interim, or incomplete thinking. I've been working for a long time to get to a fully textured appearance. I have been successful at getting there with an 8x10 camera. However, I believe I should be able to get there with a smaller piece of film - at least one as large as 4x5. Somehow it doesn't work that way - at least not yet. There's a smoothness I haven't gotten to. I'm grasping at straws to figure out where the bottleneck is. I have very smooth gray ramps in my environments but the magic is only there with the big camera - at least so far. More testing to come.

Lenny

Have you ever considered the possibility that the lack of smoothness might be due to the type of scanner you are using? Maybe you should experiment some with one of those inferior flatbeds? Could me that Premier is just looking too deeply into the film and finding things that were never meant to be found.

Sandy

Tim Povlick

8-Sep-2010, 21:08

Have you ever considered the possibility that the lack of smoothness might be due to the type of scanner you are using? Maybe you should experiment some with one of those inferior flatbeds? Could me that Premier is just looking too deeply into the film and finding things that were never meant to be found.

Sandy

Hey Sandy, I think your on to something there. How about a SF plot where a film scanner looks so deeply into the film it can see future events.... Queue the Twilight Zone theme.

---

Well, I am from Earth, although I have been to Mars (PA) to answer some of the previous questions about what planet we (Lenny and I) are from. I won't speak for Lenny on that one. :) :)

The comment about the tone curve tool being 0 to 255 is because in 16-bit it's mapping 0 to 255 to 0 to 32768 (thanks to Tyler for point out PS uses 15bits of actual data, confirmed in a PS text I referenced). Therefore if one changes 124 to 125 then this is mapping to 15872 to 16000. One could counter and say "doesn't matter as it gets mapped to 8-bits eventually". If that's your workflow np for me.

One concern I have is the graph is so small when making the adjustments to the curve, it would be real nice if PS allowed one to stretch the window larger. Will I continue to use the curves tool - YES.

_ .. --
Tim

Lenny Eiger

8-Sep-2010, 21:44

SCHWARZZEIT

9-Sep-2010, 03:08

There's part of this that is just interim, or incomplete thinking. I've been working for a long time to get to a fully textured appearance. I have been successful at getting there with an 8x10 camera. However, I believe I should be able to get there with a smaller piece of film - at least one as large as 4x5. Somehow it doesn't work that way - at least not yet. There's a smoothness I haven't gotten to. I'm grasping at straws to figure out where the bottleneck is. I have very smooth gray ramps in my environments but the magic is only there with the big camera - at least so far. More testing to come.
I think you're right that the issue you have is about film real estate. If you have a certain print size in mind e.g. 40x50" the grain structure of your 4x5" film going trough a 10-12x magnification while it's only 5-6x for an 8x10" sheet.
It depends on the film you're using but even on modern flat crystal emulsion films you can already perceive a loss in smoothness when enlarged ten times. The grain already interferes with the texture of your image.

Which type of film have you used for your comparisons so far?

-Dominique

wfrascher

9-Sep-2010, 05:27

[QUOTE=Tim Povlick;626767]

One concern I have is the graph is so small when making the adjustments to the curve, it would be real nice if PS allowed one to stretch the window larger. Will I continue to use the curves tool - YES.
_

Tim, get with some digital astronomers and import your pictures with their programs. They've been able to stretch for more then a decade. I guess to say stretch would be wrong, you can zoom in or out. :-)

Peter De Smidt

9-Sep-2010, 06:19

With optical printing, I see a slight loss in quality in moving to more than a 4x enlargement with EI 100 film. Why would things be different digitally? (This is not to say that max quality always wins the day. Sometimes the impact of a bigger print more than makes up for a slight lessening of quality.)

Lenny Eiger

9-Sep-2010, 10:28

I think you're right that the issue you have is about film real estate. If you have a certain print size in mind e.g. 40x50" the grain structure of your 4x5" film going trough a 10-12x magnification while it's only 5-6x for an 8x10" sheet.
It depends on the film you're using but even on modern flat crystal emulsion films you can already perceive a loss in smoothness when enlarged ten times. The grain already interferes with the texture of your image.

Which type of film have you used for your comparisons so far?

-Dominique

Dominique,
I am currently using Delta, and on occasion TMY-2 (for low light). It isn't the film, however, grain patterns or scanning sharp.

Here's my example: let's say you have a telephone pole in your image. It's one of those that has a seemingly endless number of shades of brown from the aging of its creosote coating (I presume they do this, or did, in the EU, and that you have these objects over there).

Let's also suggest that on a 4x5 piece of film you can take a square sample (crop) that will represent the width of this telephone pole and that with a 4x5 the width and height of this sample is .64 centimeters, or close to 1/4 of an inch.

Resolution is not the issue here. The edge of the pole will be sharp in all med-large formats, at least. However, if you compare the tonal reproduction capacity of this 4x5 vs the capacity of an 8x10, the larger film will reproduce the same square of the image in 2.54 centimeters, or a full square inch. There is far more tonal information in an inch of film vs a 1/4 of an inch.

I go over and over this in my mind. I think I am almost there, my next experiment should provide a path to see what exactly is and is not possible between the two formats with identical (both Rodenstock Apo-sironar-S lenses, both normal for the format, and matching film and development, with an appropriate subject that is tonally rich.

Lenny

Peter De Smidt

9-Sep-2010, 10:41

Lenny, maybe you're trying to make too big of prints from 4x5 for your quality requirements? A 8x10" print from a 4x5" negative ought to look the same in quality as a 16x20" print from 8x10" film.

Lenny Eiger

9-Sep-2010, 10:55

Lenny, maybe you're trying to make too big of prints from 4x5 for your quality requirements? A 8x10" print from a 4x5" negative ought to look the same in quality as a 16x20" print from 8x10" film.

I do print large, but its on rare occasions. Some images are quite dramatic and convincing when they get large and others just look silly - my opinion about my own work. That said, my target "image quality size" is 16x20.

So far, with 8x10 I can tell that the sun in the image is "summer sun" and I know that it is in the late afternoon. A long exposure at the ocean reveals tracks of spray that feel like it. Rocks are so textured you feel like you could touch them in the print. Those cues are important to me. I think 4x5 will do it, I'm getting older and would be much happier to carry around a little 4x5 on hiking trails (would be thrilled if the Mamiya 7 would do it... altho' while the lenses are extremely sharp they don't appear to have enough depth of field for me - 22 vs 45-64). I'm having to adjust my thinking to approach this. Good news is that the film and dev is working wonderfully.

Thanks for your suggestions, all.

Lenny

Tyler Boley

9-Sep-2010, 12:16

snip...
I'm in agreement that there should be no downscaling of
levels of gray anywhere in the data path if at all possible, but PS tools are a different issue.
snip...
Tyler

There's part of this that is just interim, or incomplete thinking. I've been working for a long time to get to a fully textured appearance. I have been successful at getting there with an 8x10 camera. However, I believe I should be able to get there with a smaller piece of film - at least one as large as 4x5. Somehow it doesn't work that way - at least not yet.
snip...
Lenny

well since my thinking is an easy target, perhaps you can help me complete it. So then, the curves tool is a sledge hammer with small film and not large?
Tyler

Lenny Eiger

9-Sep-2010, 12:30

well since my thinking is an easy target, perhaps you can help me complete it. So then, the curves tool is a sledge hammer with small film and not large?
Tyler

Tyler, you been working too hard. I was referring to my own thinking being incomplete. I'm going for a certain look that only 8x10 has delivered - so far. I'm just wondering where things get lost. I think I was looking in the wrong place, in fact, I'm sure of it. I should be cured now.

Lenny

Tyler Boley

9-Sep-2010, 12:48

Lenny, I'm picking on you. I know what you are after and how difficult it is. I'm merely trying to correct misperceptions indicated in some posts about the curves tool potentially hammering files or compressing levels, not being incremental enough, or in general being part of the problem.
In this pursuit of B&W excellence, it's important to not divert into areas that are NOT the problem and stay on point since the goal is so elusive, hence my earlier remarks about world views vs facts. We all love to hate Adobe, in this case I don't think they are the problem, given careful and proper use.
Tyler

Lenny Eiger

9-Sep-2010, 13:16

In this pursuit of B&W excellence, it's important to not divert into areas that are NOT the problem and stay on point since the goal is so elusive, hence my earlier remarks about world views vs facts. We all love to hate Adobe, in this case I don't think they are the problem, given careful and proper use.
Tyler

You're right, of course. I find it hard out here to do all this on my own. Like you, I have had to research things one way and the other (by trial and error), to figure out how StudioPrint works, where it does and where it doesn't, along the rest of the software, the paper, the inks. I am sure I have a whole pile of additional suppositions that aren't correct based upon incorrect conclusions I've made during testing.

And yes, I love to hate Adobe as well. Using version 5 at the moment, on my brand new Mac tower wondering where the "themes" are - it looks way too much like a PC app. (Sorry, PC guys, it just isn't my thing.) 5 is a little better, but still a pain in the ars.

Lenny

sanking

10-Sep-2010, 08:04

Dominique,

Let's also suggest that on a 4x5 piece of film you can take a square sample (crop) that will represent the width of this telephone pole and that with a 4x5 the width and height of this sample is .64 centimeters, or close to 1/4 of an inch.

Resolution is not the issue here. The edge of the pole will be sharp in all med-large formats, at least. However, if you compare the tonal reproduction capacity of this 4x5 vs the capacity of an 8x10, the larger film will reproduce the same square of the image in 2.54 centimeters, or a full square inch. There is far more tonal information in an inch of film vs a 1/4 of an inch.

Lenny

My experience is that all things being image quality will always be better with the larger format when comparing film of the same type in 8X10, 4X5 or MF. So if you are comparing the same file in 8X10 and 4X5 it is kind of a no brainer that the 8X10 image is going to look better because of the smooth tones, certainly in any print over 16X20 and perhaps even at that size depending on how close you put your nose to the print. And that would be true even if the resolution of the 4X5 negative it 2X that of the 8X10 negative because even if you have the same amount of detail the structure of the film will be smoother with 8X10 because of less enlargment.

The only way 4X5 can match 8X10 in image quality (or Mamiya 7 can match 4X5) is by the user of a higher resolution film with finer grain. I know for a fact, at least to my own satisfaction, that Mamiya 7 negatives on Fuji Acros, if well exposed and developed, equal or beat print quality from 4X5 Tri-X or TMY-2 negatives. Naturally we are talking about an apples to apples comparison where both cameras are used on a tripod, at a shutter speed that does not cause vibration, and at an optimum aperture.

The characteristics of B&W film are fairly straight forward. There is grain, resolution, dynamic range, curve type and spectral sensitivity. Did I miss something? Image quality itself does not result from any one of these characteristics but from all of them, and how the negatives matches the printing process. I think often people lose their way because they place too much emphasis on just one or two aspects of image quality and lose sight of the whole.

Sandy King

rdenney

10-Sep-2010, 13:14

I'd like to describe something in analogue terms that I think we may miss in these discussions, and as a test of my own thinking.

At various points in the discussion, the notion of a chain has been mentioned, with resolution being determined by the weakest link in the chain. But I don't think a chain is a good analogy. The chain concept suggests that if any link is strong enough, it has the same effect as if it is infinitely strong. If the resolution of, say, the lens is greater than the film, then the lens no longer has any effect on what we see. I don't think that is correct.

The point behind thinking in terms of modulation transfer is to think of waveforms. The scene presents a pattern of waveforms, some of which have, from the perspective of the camera location, infinitely sharp edges, and some of which have gradations.

A lens always renders the scene with gradations--if we look at a small enough piece of the scene that it projects. A lens imparts its own look in the way that it converts the scene's edges and gradations into gradations that do or don't appear to be sharp (even for scenery elements that are in exact focus, which, of course, most aren't). Lines per millimeter is a gross way of measuring this effect, in that it only considers sharp scenery elements. MTF is much better, because it looks at the way the lens modulates the edges and gradations of the scene at different spatial frequencies. But at high enough spatial frequencies, the MTF is always less than 100%--there is a point at which the lens will turn any sharp detail into a smooth gradation.

The film has a completely different way of integrating the information. It imposes its own patterns on the scene, with its own modulation effects. How it does so is what makes each film different, and, from an artistic perspective, this leads to photographers having preferences. Some tonal gradations imposed by the lens may be rendered as a sharp edge again by the film, if it lacks the ability to render a gradation at that spatial resolution. A film may render a straight detail projected by the lens as a ragged edge, where the raggedness of the edge (if it is fine enough) creates the impression of a gradation, and in so doing might add to or obscure the gradation projected by the lens.

If we enlarge, then the enlarger lens and the enlarging paper will each put their characteristic stamp on the result.

It's not a chain. Rather, it's a series of transparent overlays, with each imposing a different way of representing fine detail.

Scanning film also imposes a different pattern on the image. The film's ragged pattern imposed on the gradations projected by the lens will get sliced in a regular grid pattern of samples, and each sample will integrate whatever it sees in that sample. And with many scanners, the samples will overlap, so that each sample will integrate the content of that sample, plus some information from the film in neighboring samples. And scanners that use CCD arrays will scan different colors spatially offset from one another. In that way, the scanner also imposes a modulation by superposing it's sensor frequency and color array on top of the modulation frequencies that are on the film.

So, we have a superposition of wave patterns--the scene, the lens, the film, and the scanner. Each works at a different frequency, and each does different things at different frequencies. Only when one step in the process is grossly less able to handle important frequencies than the others does it become dominant. If one step has much greater resolution than the others, then it has less effect on the integrated total.

In practical terms, if our goal is to render the integrated image as recorded on the film with no further effect, the scanner must be able to see finely enough to describe the grain shapes. So, a film that can resolve x lines/mm might need to be scanned at 10x or 100x to render those grains accurately. Any less will cause the result to see the modulation effects of the scanner, and any less will cause a change in the way gradations are rendered in fine detail.

If our goal is to produce pleasing prints, then we can make use of or work around the modulation effects of the scanner. If that is our goal, then we can determine scanning resolution on the basis of what produces pleasing results for the prints we will make. The rendering of fine detail in a print will be the integrated or superposed modulation of the scene, the lens, the film, and the scanner, with none imposing such an effect as to undermine our aesthetic intent. This is the situation that leads to the chain analogy, but for one of those steps to dominate the result negatively, its modulation has to be a gross effect compared to the fine effects of the other steps. So, again, a scanner might have to have ten times the effect on a given detail than the lens for it to dominate the final result.

A step cannot dominate the result by being too finely resolved. If we modulate accurately at finer levels, the gradations at grosser levels will be accurate. The finer the spatial resolution, the fewer shades of gray are needed to still be accurate at much lower spatial resolutions. There is a spatial resolution where we can provide only binary black or white and still end up with a completely smooth gray scale at lower resolutions. If we get banding at lower levels, it is because our data has been represented at lower resolution, not lower bit depth. But I'm not talking about pixel counts, but rather clusters of pixels all the same color--this is what we call banding or posterization--, which is another way of lowering spatial resolution. Given that we can't practically scan at 100 or 1000 times what we intend on a print in terms of spatial resolution, we end up worrying about accurately integrating larger samples. Our samples are so large that we are forced to represent them as accurate integrations of what they are sampling.

So, for low-res scanners, we want lots of color depth and accurate sampling. The higher the resolution of the scanner, the less accurate each sample needs to be.

I suspect a PMT drum scanner outperforms flatbeds more because of its very high resolution than because of its accuracy for each sample. It may have resolution some large multiple of other steps in the process, but by so doing, improves accuracy without needing each sample to be more accurate.

An Epson flatbed, on the other hand, has low spatial resolution similar to or lower than other steps in the process. Thus, it must represent each sample accurately. But it does not--there is still too much bleeding from one sample to the next, or from one color to the next within the CCD pattern. A Nikon film scanner probably does not have better CCD technology than the Epson, but performs better in terms of broad tonality because of the higher resolution. I know this is weird thinking, but I believe the science supports it.

So, before we can ask about limits, it seems to me we need to set standards for the result we want. If we want everything a given film offers with no influence from scanning, I think we will need extremely high scanning resolution, perhaps one or two orders of magnitude greater than what we expect from the film. The greater the spatial resolution, the less we have to worry about accuracy, it seems to me. The less the spatial resolution, the more we have to worry about accuracy and color depth.

Rick "whose Epson isn't quite accurate enough for enlargements greater than about 4x" Denney

Ken Lee

10-Sep-2010, 13:26

rdenney

10-Sep-2010, 13:43

"The chain concept suggests that if any link is strong enough, it has the same effect as if it is infinitely strong. If the resolution of, say, the lens is greater than the film, then the lens no longer has any effect on what we see. I don't think that is correct."

I may be mistaken, but I thought that the chain analogy suggest that a chain is only as strong as its weakest link. If the film can resolve a lot, but the lens is blurry, then what we get is as good as the lens. If the film introduces artifacts of its own, then as long as they are smaller than the lens can resolve, they won't be seen, because the lens can't see them.

That is what I am arguing against. There is no hard threshold. The effect of the weak link does not mask the effects of the other links unless it is grossly weaker than the others. With a chain, any link that is even marginally stronger than the weakest link might as well be infinitely strong--it won't be the link that breaks. But an image is not a matter of broken or not broken. It is the summation of layers of subtleties, with each layer applying a filter on the underlying layers. The scene represents the base layer, and the lens, film, scanner, sharpening, other processes, and printer are superimposed on top of that layer. One of those filters might be marginally lower in one measure without masking the effects of underlying layers, or it may interact with those effects in ways that produce a third and unexpected effect (moire aliasing is one example of this).

Thus, the assumption that a scanner does all the job it needs to because it is marginally more highly resolved than other elements in the process is too simplistic. It will still impose effects. It may have to have a resolution orders of magnitude greater to impose no effects.

Rick "who runs into this effect when looking at traffic simulation models" Denney

Ken Lee

10-Sep-2010, 15:20

"The effect of the weak link does not mask the effects of the other links unless it is grossly weaker than the others."

Excellent - Thanks for spelling that out.

sanking

10-Sep-2010, 16:29

That is what I am arguing against. There is no hard threshold. The effect of the weak link does not mask the effects of the other links unless it is grossly weaker than the others. With a chain, any link that is even marginally stronger than the weakest link might as well be infinitely strong--it won't be the link that breaks. But an image is not a matter of broken or not broken. It is the summation of layers of subtleties, with each layer applying a filter on the underlying layers. The scene represents the base layer, and the lens, film, scanner, sharpening, other processes, and printer are superimposed on top of that layer. One of those filters might be marginally lower in one measure without masking the effects of underlying layers, or it may interact with those effects in ways that produce a third and unexpected effect (moire aliasing is one example of this).

Thus, the assumption that a scanner does all the job it needs to because it is marginally more highly resolved than other elements in the process is too simplistic. It will still impose effects. It may have to have a resolution orders of magnitude greater to impose no effects.

Rick "who runs into this effect when looking at traffic simulation models" Denney

I believe what you write is true. I think this also explains why the others layers would minimize the impact of the scanner if it happened to be one of the especially strong layers. You see this kind of thing in practice all the time in that most people are not able to see much, if any, difference between two 16X20 prints from the same 4X5 B&W negative, one scanned with an Epson V700, another with a very high end drum scanner. And that is because when we evaluate prints it is a global response, not a response to one or two discreet values. We need to understand this and not obsess too much about one or two single links in the chain.

Sandy King

Lenny Eiger

10-Sep-2010, 18:41

I believe what you write is true. I think this also explains why the others layers would minimize the impact of the scanner if it happened to be one of the especially strong layers. You see this kind of thing in practice all the time in that most people are not able to see much, if any, difference between two 16X20 prints from the same 4X5 B&W negative, one scanned with an Epson V700, another with a very high end drum scanner. And that is because when we evaluate prints it is a global response, not a response to one or two discreet values. We need to understand this and not obsess too much about one or two single links in the chain.

Sandy King

I don't know - I mean I think printing in carbon is pretty obsessive. Certainly not so easy to get it just right. As an artist, I am looking for something. Maybe all the links have to function at a high level for it to work. I can tell you that I'm not interested in photographing with a Canon D-whatever. There are plenty that suggest that this is just fine, and why should I fuss over film and lug around this LF camera. Certainly most here would disagree.

Where is the line of its just fine? Of course, the line is where each person places it. I really like the quality that I get when everything works. (Of course, its a lot better when the image is worthwhile as well.)

For me, it can't matter whether everyone can see the difference visually. I think they do see it, they see an artist who's produced something he or she is willing to stand behind. Especially when there's a consistency to a portfolio. It comes across.

Lenny

sanking

11-Sep-2010, 06:45

I don't know - I mean I think printing in carbon is pretty obsessive. Certainly not so easy to get it just right. As an artist, I am looking for something. Maybe all the links have to function at a high level for it to work.

Lenny

I did not say that we should not obsess about our art. What I wrote was, "We need to . . . not obsess too much about one or two single links in the chain."

Carbon printing is hardly one or two links in the chain. Rather, it is the end process of a very intensive method of making hand made prints that includes many links. Changing any one of those links may or may not diminish the artistic quality of the final product. Scanning a 6X7cm B&W Mamiya 7II negative with an Epson V700 instead of with a dedicated film scanner when the final product is a print 16X20" or larger would surely diminish the artistic quality for me. Scanning a 5X7 B&W negative with an Epson 700 when the final product is a print 16X20" would not diminish the artistic quality for me. The reason is that important image detail would be lost in the first case, in the second case not. So obsessing for me would be worrying about scanning the 5X7 B&W negative with a drum scanner when in point of fact it won't make a hair on a rat's ass difference in the overall quality of the final print.

I can not say what would or would not be obsessing for you or anyone else, but for my own work I have, for the most part, learned which links require obsessing and which ones do not. But then I am often surprised by real life results.

Sandy