side by side comparison... large print digital back VS 4x5 color film

[Lenny Eiger]

I sent Michael an email.
Come on Lenny ... you'll crush the poor fool. Think what it would mean for your business.

Nothing like a little violence to soothe the soul...
;-)

Lenny

[Lenny Eiger]

I've always found this claim quite dubious. The size of the film grain clumps (dye clouds in color films) vary widely in most films used to capture real subjects with realistic subject brightness ranges. IOW, there is not one single size of the grain of the film -- film grain clump size is fairly stochastic (clump sizes increase generally with increasing density). And the scanner doesn't vary its aperture size as it scans - the aperture size is fixed. So the claim that the scanner "captures scan sizes based on the size of the grain of the film" (and yes, I realize it's not you making this claim, it's Aztek's marketing hype) is a non sequitur.

More on what film grain is and what it actually looks like can be found in Tim Vitale's excellent paper on the subject.

Bruce,

There are a number of things that work - that don't make apparent sense... I know you have seen the effects of aperture on scans. It may be based on something else entirely, but you know that sample size (aperture) is a huge factor in the quality of the resulting scan. Too small and you get salt and pepper, too large and you get blurry scans.

The Premier has aperture settings every two or three microns, where the 4500 jumps form 6 to 13, to 19, etc. It's just a little less, altho' 13 works for a lot of things in b&w neg or transparency, so it doesn't matter. The 19 works great for color negs. I don't really need the 8 and the 10 - they aren't used that often.

It's better than the Tango, which is set at 11, and can't vary. Some others have small, medium and large settings, which are not particularly specific.

It's a huge difference when you get it right, however. You end up smoothing the grain just a bit but retaining the sharpness. I think the problem is that there are multiple things going on. First of all, you aren't really looking at grain, you're looking at scan samples...

is it clouds we are looking at? I don't really know. I've invited Tim to come over many times, and he almost has... However, the long and the short of it is that the aperture does do a lot...

Lenny

[Bruce Watson]

I don't have a problem with using aperture size as a tool to make a good scan. We agree it's a valuable tool.

What I have a problem with is the idea that one can match the aperture size to the grain clump size. The aperture is fixed for the duration of the scan, while the grain clump size varies (sometimes markedly) over the density range of the film. IOW, there is not a one-t0-one correspondence; the laws of physics prevent it.

What setting the aperture size is then is an art. You and I and any other drum scanner operator has to work with the film and decide which aperture is the best compromise of sharpness and smoothness for that particular film. With my own work the correct aperture setting varies from sheet to sheet of the same film developed the same way. IOW it's not a constant under the best of conditions. It's a compromise, and it's that idea of compromise that they completely leave out of the marketing brochures.

It's just the way their marketing says it. In trying to simplify it into a sound bite they get the physics wrong. I hate it when that happens.

[Lenny Eiger]

It's just the way their marketing says it. In trying to simplify it into a sound bite they get the physics wrong. I hate it when that happens.

I totally agree.

I had many conversations with Phil Lippincott where I would ask a question and get a 2 hour response that had little if anything to do with my question. I think his brain used a language from another planet. However, as I said in a previous post, there are a number of things that don't make sense, yet work a certain way.

I would also agree that setting aperture's and settings is an art, rather than a science, and perhaps that's why the operator is so important. I would add that color management is also an art, way worse than scanning.

Lenny

[Bruce Watson]

I totally agree.

I had many conversations with Phil Lippincott where I would ask a question and get a 2 hour response that had little if anything to do with my question. I think his brain used a language from another planet.

Ah yes. I had some of those oddly frustrating conversations with Phil too. He was certainly a piece of work. Interestingly convoluted thinking. But it seemed to work for him.

However, as I said in a previous post, there are a number of things that don't make sense, yet work a certain way.

This one makes sense to me. I just can't figure out how to reduce a description of what's going on into a sound bite. Good thing I don't have to!

I would also agree that setting aperture's and settings is an art, rather than a science, and perhaps that's why the operator is so important. I would add that color management is also an art, way worse than scanning.

Well, at least color management is not a black art. ;-)

[Lenny Eiger]

Well, at least color management is not a black art. ;-)

Uh, actually, that's what I meant.... ;-)

I bought ColorThink to analyze my profiles and I got all the curves to match up perfectly - then made a horrible print. I called the guy who made it and he said - well, it's just a guide.

A $400 guide... I now make a profile, then print, then make another until the print looks right... and I have iO Table, I know my StudioPrint upwards and downwards and it still doesn't mean its right unless the print works....

Black art? Yes, matey....

Lenny

[Tim Povlick]

I've always found this claim quite dubious. The size of the film grain clumps (dye clouds in color films) vary widely in most films used to capture real subjects with realistic subject brightness ranges. IOW, there is not one single size of the grain of the film -- film grain clump size is fairly stochastic (clump sizes increase generally with increasing density). And the scanner doesn't vary its aperture size as it scans - the aperture size is fixed. So the claim that the scanner "captures scan sizes based on the size of the grain of the film" (and yes, I realize it's not you making this claim, it's Aztek's marketing hype) is a non sequitur.

More on what film grain is and what it actually looks like can be found in Tim Vitale's excellent paper on the subject.

While I would agree film grain clump size can be viewed as a Stochastic function, the

clump sizes increase generally with increasing density

is not the reason why. Can you elaborate on this? I checked "Probability, Random Variables, and Stochastic Processes" by Papoulis.

Back to the main point, scanning and film grain, (or whatever one wants to call it), the aperture is being set not to match simply the grain size of film, but the average size of the smallest grain. In Stochastic terms this would be the expected value of grain size. So I agree with what Aztek says, as long as one realizes it's the average grain size that set's aperture. Therefore one should not vary aperture while scanning, If one is set to 10 microns and along comes a dark area that is "clumped" at 55 micron, the 10 micron aperture will read the density properly.

Attached is an intentionally "bad" scan I did with the Premier on Ecke 25, 4K dpi and 8 micron ap. The light and dark areas show the same general "grain size", at least to my eyes.

FWIW I have worked with the folks at Aztek and found them to be straight talkers. This from an EE of 30 years experience that includes designs using PMT's.


Regards,

(West Coast) Tim

[sanking]

I have seen what seemed to me very significant grain reduction with software like Noise Ninja, without any apparent decrease in apparent sharpness. In what fundamental way does the mechanism of adjusting the aperture with a drum scanner differ from the software? Are we not with the software doing the same thing, i.e. adjusting edge rounding to match the average grain clump size?

Sandy King

Back to the main point, scanning and film grain, (or whatever one wants to call it), the aperture is being set not to match simply the grain size of film, but the average size of the smallest grain. In Stochastic terms this would be the expected value of grain size. So I agree with what Aztek says, as long as one realizes it's the average grain size that set's aperture. Therefore one should not vary aperture while scanning, If one is set to 10 microns and along comes a dark area that is "clumped" at 55 micron, the 10 micron aperture will read the density properly.


Regards,

(West Coast) Tim

[Lenny Eiger]

I have seen what seemed to me very significant grain reduction with software like Noise Ninja, without any apparent decrease in apparent sharpness. In what fundamental way does the mechanism of adjusting the aperture with a drum scanner differ from the software? Are we not with the software doing exactly the same thing, i.e. adjusting edge rounding to match the average grain clump size?

Sandy King

This is a little hard... it took me a while to understand this. I think the two processes are quite different from each other. Tim will have his own understanding, deeper in some ways than mine, but here goes. If someone has a better way to describe this, I'm all ears. I may also not be perfectly accurate in this description and feel free to correct me.

I believe Ninja blurs, at least that's what it did on my test. Maybe Sandy or someone else can fill in here if there is more to this... that's all I know and I don't mean to shortchange it.

The aperture is a sample, it isn't like this, but I imagine an opening to the sensor that's like a vertical shade - or shutter perpendicular to the drum. As the drum goes around, the sensor can see thru this shutter. One can widen the shutter or tighten it.

If you sample many times as the drum goes around, thru this opening, you get a series of samples, that make up the pixels of an image. A Premier has a 38,000 step stepper motor, with that many possibilities around a single circumference. It takes 8,000 of those steps and makes samples.

That's vertical. The drum can move over in increments of a16,000 of an inch at a time. It moves over a specific distance based on the ppi/aperture settings. It occurs to me I don't exactly know which one of these it chooses, but it must be the ppi if the aperture can oversample... If the slit that it looks thru is smaller than the RMS Granularity, the scanner will sample the same grain cloud more than once, creating an effect called grain anti-aliasing. If the slit matches the granularity, then you have a sharp scan with no ill effects. If the slit is larger then the image will be blurred.

It isn't a slit, its a circle in a rotating disc, it's just how I visualize it. There are usually two options that are "correct", one that will be grainy and sharp, and another where the grain is smoothed somewhat, but is still sharp. One can make a 1/8in x 1/8in sample and blow it up full size and watch the effects at different micron sizes. Differences are quite visible.

Here's a pic I made that I think illustrates the "slice" and the grain cloud it is looking at, and why I think this effect is occurring. I have a link to the comparison of how different micron settings affect the image here: http://eigerstudios.com/scancompareMicrons.html

I hope this helps...

Lenny

[Oren Grad]

the aperture is being set not to match simply the grain size of film, but the average size of the smallest grain. In Stochastic terms this would be the expected value of grain size. So I agree with what Aztek says, as long as one realizes it's the average grain size that set's aperture.

This makes no sense in purely mathematical terms. By definition, the expected value will never be the smallest grain size unless the grain is uniform in size. And in that case, talking about the "average grain size" is superfluous.