Ideal film/developer-drum scanning B&W Landscapes

[Jack Brady]

After 20 years absence from Photography, I'm back. I've shot 2 1/4 and 4x5 in the past of landscapes strictly in B&W. Used to use Pyro developer with propriety formular- basically what is now WD2D. I've noticed in using my new scanner that grain can be an issue with the negatives I shot 25 years ago - then it was not a problem as I used Fred Pickers Cold Light Head for my printing.

Films have come and gone, hopefully some have improved, in the last 20 years. I've reviewed many posts here but find very little on scanning and the ideal film/s.

I would appreciate your suggestions for what B&W film AND developer I should give consideration to. I know there is a lot of personal preferances involved but ideally I'm seeking advise from individuals who are scanning their files and printing up to and including 30x40 size prints for the fine art market, but I'll listen to most anyone that has something to share!

I prefer to go with one film and developer for my work and learn it's traits in detail rather than constantly changing.

The key factor is that the B&W negs will be scanned on my Howtek 8000 drum scanner, which can scan down to 3 microns, so grain is a serious consideration for me; far more concern on grain and latitude than film speed. Currently I'm finding I'm getting best scanned images at 8 and 10 microns from my old negatives.

To date, I've purchase the following for my use: Howtek 8000 HiResolve drum scanner and the Jobo CPP2 with lift to do my own developing; X-Rite 810 densitometer to calibrate my negatives to the Zone System. All future shooting will be on 120/220 film and 4x5 - I don't care if I have to load my 4x5 or go with QuickLoads/ReadyLoads - I'm far more concerned with results rather than convenience.

Once negative is scanned, I'll stay in the digital domain with CS2 and be printing on the Epson 4800. No darkroom printing.

Your guidance and learned experience is appreciated.
Jack

[Ron Marshall]

Jack, I was in your situation about six months ago. As I don't have extensive experience I will direct you to a site that specifically adresses your questions.

http://www.butzi.net/articles/tuning1.htm

For fine grain I recommend Kodak T-max 100, or Fuji Acros 100. Both are great films, but I prefer Acros since it does not require any reciprocity correction up to 2 minutes.

[Ken Lee]

When you develop your negatives for scanning, leave them a little softer than you might otherwise, because with a digital workflow, we can always increase contrast. The limiting factor, even with a high-end scanner, will be the range of brightness, more so than resolution, which can be captured.

I suggest TMax 400 for 4x5, developed in Pyrocat HD. After 30 years absence from cold light heads, Tri-X, HC-110, etc. I found this combination to be the modern-day equivalent. If the grain is appropriate, then use if for roll film as well. If not, then Ilford FP4 will suit you nicely, with very similar flexibility for zonal expansion and contraction.

Don't take my word for it (although I have plenty of sample images which may help persuade). Have a look at the research work done by Sandy King and others. It is thorough, compelling, and consistently reliable.

[Bruce Watson]

Jack,

You've asked this question several times in several places. What is it about the answers you are getting that you don't like?

[Jack Brady]

Bruce,

The absense of a large enough sampling of suggestions from folks that are actually scanning with a high end scanner.

That and the idea that this site may be more on target with the work I'm doing.

Jack

[Jack Brady]

Ron,

What are you using for a developer?
Using a Jobo?
Is so, suggested times?

I've just ordered some Acros 4x5, so I'll be doing testing next week.

Thanks,
Jack

[Frank Petronio]

Just my two cents... every successful photographer I know tests before making a "final" decision, and then those "final" decision are never forever.

Since your Howtek and Epson are fixed variables, why not try a round of different films? You may find that color neg maybe better, since the idea of calibrating to the Zone System is only going as far as capturing information onto your negative, not in making a trad print. Then again, isn't the Howtek an 8-bit scanner so that you're only scanning the middles and not the ends?

[Michael Mutmansky]

Jack,

What you are possibly going to experience with a drum scanner is grain ailasing effects; the condition that occurs when the sampling size is close to the inherent grain size of the film. This effect does not occur with typical flatbed scanners because they never perform anywhere near the actual resolution they claim. Recent tests on consumer oriented scanners sho that they all max out at about 2200 ppi, which is below the grain size of all but the fastest, grainiest films.

However, you have to also consider thatif you are shooting 4x5, it is unlikely that you will ever need to sample above about 2000ppi for a reasonable print, unless you print very large. That's not to say a flatbed can perform equally to a PMT scanner, but that you may not ever need to sample as small as you have mentioned, except in cases when you are using a smaller source negative.

Since you have a drum scanner, I recommend you change nothing about your negative contrast, because the scanner will easily be able to handle the density of a normal silver negative. Developing to a low contrast will ultimately INCREASE the apparent grain and noise effects in the image by extrapolating the data outward in the digital realm, which may cause the subtle noise artifacts to become more prominant.

Most scanners, even consumer flatbeds, can easily handle the range of density in a traditional silver negative, that may go from about .15D up to about 1.2D or so. They should even be able to handle negatives developed for alternative processes reasonably well, and drum scanners can easily handle the very high densities of negatives developed for pt/pd without difficulty (at least mine can).

If you are concerned about grain, shoot the tightest grain film you can find, TMAX 100 or Acros, and develop it in a staining developer. The stain is not grain-based, so the density it adds will not contribute to the grain ailasing effect. I would not recommend a developer like Rodinal in these situations, as the grain will be very defined and will probably give you fits in the scanning stage.

Beyond that, you may have to oversample the scan a little to knock down the grain ailasing effects. This will soften the scan slightly, but sharpening in PS later on will compensate.

I recommend you make some reasonable decisions about what you want to be able to do with your 4x5's first, and then make some tests based on the magnification that you will need to do to meet the output that you desire. There's no reason to take the approach beyond what you can reasonably output, because the files become very large, and much slower to manipulate. This means you probably won't be scanning at 4000ppi with 4x5, and therefore the grain ailasing may not truely be an issue, but more of a theoretical problem. And even then, there are ways to deal with it.

---Michael

[Michael Mutmansky]

Then again, isn't the Howtek an 8-bit scanner so that you're only scanning the middles and not the ends?

What? ...

Frank, an 8 bit scan has nothing to do with where in the source densities the scanner operates, only the number of possible 'steps' that the scanner is capable of placing the data into for the output file. PMT scanners are ANALOGUE devices up until the point where the data is converted for output by the A to D converter. Some of the older ones are only 8 bit, but internally, they may be 12 bit.

---Michael

[Jack Brady]

Frank,

The Howtek HiResolve 8000 has the following specs:
8000 Optical dpi
17 apertures - down to 3 microns
Density 0-3.9 OD
12 bits per color; 4096 levels of gray
Effective scanning area: 11.5"x12"
Reflective and Transparent.

I use the Aztek DPL Pro which gives a 16 bit Tiff Grayscale file.

I'm finding that I'll scan 2 1/4 x 2 1/4 at 8K res and 8 or 10 microns.
4x5's are working best at 4K res at 8 or 10 microns.

I have scanned several 4x5's at 8K - the detail is awesome BUT the file size is typically in the 3 GB range. Even with 8GB in my G5/2.7GHz machine it is just too large of a file to be able to work with any layers and feel comfortable about it.

Jack

[Ron Marshall]

Jack, I hand roll a Jobo 3010 using D76 1:1. At 68 degrees, 50 revolutions per minute, normal development is 8 minutes. I then scan on an Epson 4990. I only print up to 16x20.

[Eric Leppanen]

Jack,

I use Tmax 100 for B&W drum scanning applications, due to its small, tight grain and sensitivity to development variations (which I like for Zone System work). I have tried Acros but it tends to be a bit contrasty in Xtol developer (used by my B&W processing lab, I don't do processing myself), otherwise it would also be an excellent choice. In theory Delta 100 should be another option, as it is similar to Tmax.

[Kirk Gittings]

" Developing to a low contrast will ultimately INCREASE the apparent grain and noise effects in the image by extrapolating the data outward in the digital realm, which may cause the subtle noise artifacts to become more prominant."

This may partially explain a phenomena that I experience also. That phenomena is: if I scan a normal negative at a tonal range that is in the ballpark of what I know I want in the final print (when I know I will want allot of midtone contrast in the final print) rather than a low contrast generic flat scan . I get less noise by sampling at the desired tone than appying a steep curve to a tonally flat file in PS. This is especially true when I want to add allot of drama to a sky. If I appy a steep curve to a snapshot that I am going to use to burn in with, oftentimes I get huge noise increases in the transitional tone areas that have been "extrapolated outward" by the steep curve.

Does that make any sense?

[Ted Harris]

Jack,

You've got good information for people who do it. Fromt he technical side pay particular to Michael. Also, Kirk has lots of good info and you will find useful stuff at Paul Butzi's website.

[Scott Schroeder]

Kirk,
I find your evaluation very interesting. Ironically, this something I have been trying to pin down a solution for. I shoot Ilford Delta 100 and Acros 100 and develop by Jobo with pyro. If I understand you correctly, you are saying it is better to adjust contrast in the scan (or close to your final vision) than in PS. Let's say a dramatic sky with puffy thounder clouds. By applying the contrast during scanning, you are avoiding the artifacts (noise) that start to show if you burn too much?

This is quite interesting. Anyone else have something to add to this?
Thanks

[Eric Leppanen]

" Developing to a low contrast will ultimately INCREASE the apparent grain and noise effects in the image by extrapolating the data outward in the digital realm, which may cause the subtle noise artifacts to become more prominant."

I've seen this too when having Piezography prints done. I initially processed my film for low contrast to record as much shadow detail as possible, but found that later adding contrast digitally (through curves or sharpening) sometimes resulted in a weird effect where the print came out looking like a charcoal painting rather than a photograph. I can only assume this was the B&W version of posterization (where excessive digital manipulation results in loss of tonal separation and detail). There's no substitute for having your negative as close as possible to your desired look, so that the need for digital manipulation is minimized.

[Michael Mutmansky]

Kirk,

Yes, it does. That is part of the reason that I believe it is important to get is reasonably close to right in the scan, and not leave it to PS manipulation later on. Every manipulation to the file after it's creation destroys data in some manner, with the exception of some basic rotaional transpositions and other minor things like that.

Curves, Levels, color adjustments, etc. all ultimately are destroying data that was in the original file, which is why it is so important that the manipulations are made as Levels adjustments, and are not applied directly to the background.

In the case of Curves, as you have suggested, the adjustment of the midtones causes a seperation in the steps somewhere in the curve, and a compression of the steps in another part of the curve. When the seperation is drastic enough, it will cause an increase in the apparent 'grain' that is visible. In this case, I mean noise more than grain, but it does ultimately come off looking like digital grain.

This is partially exacerbated by the fact that most people consider a digital file to be a continuum, rather than a stepped data source. Further, PS is apparently not truely 16 bit, it actualy operates at 15 bits. Even more, most B&W files don't have anywhere near the theoretical number of possible levels of gray in them, most have much fewer than 1/2 and high bit files (16 bit) may often have 1/3 or 1/4.

Additionally, the problems most easily crop up in areas where there is a discontunity in the tonal structure of the file, because this is where a large step is already being made in the data.

This can be made worse if the source file is a general scan to cover the entire density range (the raw scan output approach), because then the useful data does not cover the entire range of the image file, and if there were possibility for some interpolated data within the useful data set, it was thrown out in favor of capturing useless data beyond the image density range.

Suddenly, 16 bit doesn't sound like too much, does it? Image how careful you have to be when scaning at 8 bit? My drum scanner has a 12 bit output, and writes a 16 bit file so it is useable in PS, but it only has 12 bits of true data in it. Because of the fine control over the input data, I am able to easily produce good scans that require only minor adjustments to the fie, and these types of scanning artifact problems do not show their head too much. I still have to be careful in open sky areas, though.

Jack,

The Epson 4800 is a 17" printer, correct? With 4x5 film, you will need to scan at about 1750ppi to get a 17x21" print from a slightly cropped negative at 360dpi. If this is the limit of your printing capability, then you may want to reconsider the resolution you are getting in your scans.

---Michael

[Jack Brady]

I just took a break from CS2 working on an image that I scanned - 2 1/4 B&W scanned at 8K with 8 aperature. Had to get away from the image! Came over to my PC and started reading the replys.



I too am finding that if I scan a negative for low contrast/flat look from the scanner, when I take the image into CS2 I have to do a lot of "rebuilding" of the contrast with Curves.

Amazing the timing of Kirk's and Michael's comments!

The image I'm working on right now was taken in 1982 on Cape Cod - it's of an old boat bow on the sand with nice clouds in the sky. Scanned flat as someone has recently told me to scan that way and do all of you manipulation in CS2.

Working with it in CS2 and trying to get the sky right is introducing all kinds of digital noise!

I'll rescan the neg in a couple of days and go for the "finished look" with DPL Pro, then bring over to CS2 and see if I get the same digital noise.

I'm betting I won't.

This mornings work is putting merit to Michaels comments!

Michael - yes, the Epson 4800 does up to 17" wide. What I'm doing is scanning and working at the resolution of the largest print I think I'll ever do - that would be on the 9800 when it is released. Now, I'm pumping 999 resolution to the 4800 when I print and getting fairly good looking results ( I still have not standardized on an archival paper - I like the traditional glossy look but just purchased a sample pack of the Red River papers to test. I just tried the Hahnenuhle Photo Rag, Smooth Surface, and did not like it.)

Thanks for the dialog - I'm learning a lot and appreciate the input!
Jack

[Kirk Gittings]

Thank you for that Michael. I have found that to be true in practice but could never explain to anyone why it was.

[Brian Ellis]

Sory to dispaly my ignorance but I have a couple questions about some of the above responses:

Kirk said: "I get less noise by sampling at the desired tone. . . "
What does it mean to "sample at the desired tone" and how do you do it?

Michael said:

". . . which is why it is so important that the manipulations are made as Levels adjustments, and are not applied directly to the background. "
Do you mean it's important to apply Levels on a New AdjustmentLayer? If not then I don't understand what you're saying.

Michael also said:

"Additionally, the problems most easily crop up in areas where there is a discontunity in the tonal structure of the file . . . "
What's a "disconinuity in the tonal structure?" Do you mean an abrupt contrast change, e.g. a silhouetted tree against a bright sky?

" My drum scanner has a 12 bit output, and writes a 16 bit file so it is useable in PS, but it only has 12 bits of true data in it."
How does one go about determining this kind of thing? Is it peculiar to drum scanners or can it be determined with a flatbed?

Thanks for the question and responses, they've been very helpful except for the things above that I need some help with.

[Kirk Gittings]

This explanation by Michael is so very very important that I would like it written up and added to the archive. I know you are busy Michael but it would be of great use to me in the future with students etc.

[Ken Lee]

"Developing to a low contrast will ultimately INCREASE the apparent grain and noise effects in the image by extrapolating the data outward in the digital realm, which may cause the subtle noise artifacts to become more prominant."



"... adjustment of the midtones causes a seperation in the steps somewhere in the curve, and a compression of the steps in another part of the curve".



Wonderful explanation !

I'm glad the original question was asked, even if it was for the nth time - I learned something most valuable. Thanks !

[Scott Schroeder]

I agree, excellent information! Now I need to go reread Michael's VC article. I think it will make more sense this time.

[Kirk Gittings]

Brian,

I am simply but confusingly stating that if I want a final print with rich mid tone contrast (like the old Tri-X HC110 look that I like so much), I do not want to do a flat low contrast "raw" scan which I have to apply a steep curve to later on. I want to scan it (sample it) at a tonal separation (mid tone contrast, end points etc.) close to the final tones that I want in the print.

[paulr]

I'll seccond the recommendation for tmax. Grain in the 100 speed is smaller than can be resolved at 2400 ppi, and the modulation characterics of the film are so outstanding that you can presum zero loss of quality in the film stage in enlargements smaller than 4x.

I'd recommend against developing for adjacency effects. Doing so is like sharpening the image on the negative. The problem is that the effectiveness of any sharpening operation is completely dependent on enlargement size. Photoshop gives you much more flexibility to get it right for the particular size you print, and to do so with less loss of information.

Personally, I would not worry about the amount of information being lost to a single, moderate application of a curves layer (which is all the tonal adjustment a decent negative should require). Regardless of whether you're getting 16 or 15 bits, you're getting a lot of them. In my own tests I have had to commit acts of uncommon brutality on a 16 bit scan before problems started showing up in the histogram. Even more to get problems to show up in a final print. At any rate, I haven't compared the virtues of low vs. high contrast negs for scanning. My tmax negs are on the low contrast side ... they print on grade 3 paper. And they scan great. But I do not think you'd have problems seeing into dense highlights on negatives that are a couple of grades more contrasty, even with a consumer scanner. Even a fairly dense bw neg is a lot less dense than a typical chrome.

If you accept that tweaking the tones with a curve layer won't kill you, then your film/developer combination is absolved from having to provide the perfect tonal scale. You can chose your developer for more workmanly qualities: speed, grain, resolution, how nice it smells, etc.. Acknowledging Kirk's point, it makes no sense to take this to extremes. And it makes all the sense in the world to get your scanner software set up to deliver a raw scan that's in the ballpark.

[Kirk Gittings]

"Even more, most B&W files don't have anywhere near the theoretical number of possible levels of gray in them, most have much fewer than 1/2 and high bit files (16 bit) may often have 1/3 or 1/4."

Michael,

Does this suggest that one should perform most of the workflow on scanned b&w neagatives in RGB?

[Michael Mutmansky]

Brian,

You know, when I wrote that, I knew it didn't sound right, but I missed what my error was.

Yes, I mean it needs to be done as an adjustment LAYER, not Level as I wrote in the above.

-------

The discontunity I mention is the region on the negative where there is a substantial density change due to a luminance change in the subject. The problem is, there are not only ideal (abrupt, instantaneous) contrast changes (like a tree silhouetted against a sky), but situations where the transition is rapid, but covers some visible range of the negative.

These rapid contrast change regions often have very distinct steps to them, and they can be visible because they are more than a single pixel wide, as an instantaneous change may occur. Now, add to that some digital noise from the scanner, and when a substantial Curves adjustment is made, problems with banding, digital hash, and other scanning artifacts can show up.

-------

As far as I know, the specifications of the flatbed scanners can be trusted (to a certain extent). Some of the older pro scanners have 14 bit CCD technology, so they are probably only able to output 14 bits of data to the file, even though it is a 16 bit file. What is essentially done is the remainder of the file is filled with "0's".

All of the modern consumer scanners are 16 bit, but again, that doesn't mean you are getting 16 bits of data in the file. That is the theoretical limit of the data that the CCD sensors can provide, and it doesn't mean that they will actually deliver that amount.

Besides, most images are not evenly distributed with tones in a manner that will give you data in all 16 bits anyway, so all of these things are essentially theoretical in the first place.

Don't worry too much about the bit depth issue, I was simply making a point that most drum scanners are not capable of the bit depth of the flatbeds, but they are still capable of very high quality results when operated by a knowledgeable pilot. The same goes for many of the scanners out there, within their reasonable limits (and you need to know what these limits are for your particular scanner).

---Michael

[Michael Mutmansky]

Jack,

Now, I'm pumping 999 resolution to the 4800 when I print and getting fairly good looking results

I suggest you do a few tests on this. My experience is that it is better to do the interpolation in the computer rather than in the printer. The older Epson printers operated at 360/720/1440 ppi in native mode. That means that if you send them a file that is not at one of these resolutions, the printer has to do an interpolation, either in the hardware, or in the driver.

I was originally under the impression that the Epsons were native at 360ppi, but I have seen some evidence that they may actually be native at a higher rasolution, possibly 720ppi, but I doubt it would go as high as 1440ppi.

I suggest you do some clip tests from a high resolution file set to print at the same size, and send the printer a raw (999ppi) file, and then send the same one downrezzed to 720, and then another downrezzed to 360ppi. Then compare.

You will have to sharpen each file independantly, and you should start with an unsharpened source for the downrezzing so you don't cause sharpening artifacts. Also, when you do the final sharpening, you have to remember that the amount you do needs to be based on the output resolution, and what is best for printing at a particular size will probably look to be too much on the monitor, especially if you are printing at a very high resolution like 720ppi or more.

My guess is that the 720 downrezzed file will print the best on the printer, followed maybe by the 360ppi, and then then 999ppi file. The differences will show up in edges and lines mostly, but there will probably be a visible difference.



---Michael

[Michael Mutmansky]

Kirk,

Does this suggest that one should perform most of the workflow on scanned b&w neagatives in RGB?

Yes and No.

I considered this possibility a while back, because using RGB is essentially like multisampling in one step. It takes three different CCD scans of a particular point on the film and then in PS you could combine them by converting to grayscale, and you will suddenly achieve over one more bit's worth of data in the file. This could be beneficial in a negative that is prone to noise in the scan, or an image that happens to have the kinds of details that noise shows up in.

However, with the typical flatbed scanner, there is enough of a lack of perfest registration that using all three channels will soften the image, sometimes considerably. The same goes for drum scans, but the registration is probably better enough that it may be worth considering under some circumstances.

It is possible to actually have this increase the noise level as well, depending on the scanner capabilities. On many scanners, the Red channel is the most prone to noise. If you did an RGB scan where the G and B channels were pretty good, but the R channel had a good bit of noise, you would be better off not using the R channel in the composite scan.

I have tried a bunch of these variables. I think that the results are going to vary for each scanner type, and may also vary based on the conditions of the electricity in the house, etc. but if you care about reducing the noise, multisampling and combining the G and B channels in the Channel Mixer will probably produce the best result.

It won't necessarily produce the sharpest result, however. If you have the ability to do multiple sampling in a single scanner run, then the best will probaly be to scan that way, and then use only a single channel from the output.

Let me think about the multichanel benefits for the Curves manipulations a bit. There may be a benefit for times when banding or noise shows up. I may have an approach that would help alleviate that.

---Michael

[Bruce Watson]

This is interesting. In my testing, I've found just the opposite of what Mr. Mutmanski has found. In particular, I've found that I get better results, including less graininess in the print, by decreasing developer time and thus negative density. In Zone System land, my normal development is N-1 to maybe N-1.25 which seems to be about optimum for me.

This is doubly perplexing, since Mr. Mutmanski and I are both using nearly identical drum scanners - he's using an Optronics ColorGetter Falcon and I'm using a ColorGetter 3 Pro IIRC. Both using the same ColorRight Pro 2.0 software to drive the scanner IIRC.

Here are the steps I took to come to this conclusion. First, I went to the literature. Haist and Henry and a handful of other researchers. I learned what I could about what determines graininess and acutance. What I found was that grain clump size (dye clouds if you want to think in terms of color films) is controlled by a handful of factors, some of which are interrelated.

The biggie is the film's ISO rating. Fast films have bigger grain than slower films. Next on the list seems to be developer type (solvent vs. acutance). It's a trade off. To get smaller grain size, you give up some acutance. To get higher acutance, you get larger grain. Next is density. More density means larger grain size. Most of us know this instinctively as we seldom have grain problems anywhere but the highlights. This is one of the big reasons for the continuing popularity of transparency films, particularly with the 35mm crowd -- in transparency land, the dense part of the film is the shadows where it is hardest to see grain problems, while the highlights show the smallest grain. After that, you end up with the interlocking parameters of processing temperature and processing time. Notice that agitation doesn't directly effect graininess - only through the effect it has on processing time. That is, if you want smaller grain and agitation is your only tool, then use constant agitation (counterintuitive, that, but it has scientific backing).

Armed with this knowledge, I ended up with the following. First, Tri-X. Reason: I need the speed. Second, Jobo CPP-2 and 3010 tank. Reason: I want consistency beyond what I can get with other methods. Third, XTOL. Reason: It straddles the line between solvent and acutance developers, exhibiting less solvent behavior as you dilute it, and it dilutes very well. That, and I found I had problems controlling HC110 with constant agitation.

Given this, I proceeded to find my EI and normal development time per the Zone System. I started out with XTOL 1:3, 20C, 10 minutes, about 30rpm. From there, I started reducing time and therefore reducing DMax and therefore contrast. At each stage, I would scan at 11x (10x with some room for cropping if I need it) with an output of 360 dpi, or nearly 4000 ppi scanner resolution (The scanner software wants the degree of enlargement at your output size and resolution; it doesn't expose to the operator the resulting scanner resolution). This results in a file about 550MB, 16 bit grayscale. I would take a 30 cm square chunk out of the full size image and print it on my Epson 7600 using selenium PiezoTones. This in turn went up on my proofing wall. I posted them side by side under the same lights.

I found small but steady improvements in the print going down to about 6.5 minutes. The scanner's densitometer told me that I was getting zone VIII values in the 1.0 range at 6.5 minutes, down from about 1.35 at 10 minutes. That's a bit more than N-1 in my book.

What I found in my prints (sections out of what would have been 125 x 100 cm (about 50 x 40 in) prints I could print the whole thing) was this. As developer time went down, graininess went also went down. That is, it went from being barely visible to being invisible. I want to emphasize here that it's not a big change, but it's noticeable if you are looking for it. Acutance improved very, very slightly.

Some of these images I've now had printed on canvas at 125 x 100 cm. The full size prints verify what I saw on the 30 cm sq. sections. At this level (about 10x enlargement), I can see no grain, even when I take my glasses off and put my nose almost on the print (about 22 cm (8.7 in). Likewise, acutance is excellent. Likewise, tonality is excellent.

So, that's what I've found works for me. Tri-X, XTOL 1:3, 6.5 minutes at 20C, Jobo CPP-2, 3010 drum, resulting in N-1 or a bit less development.

When I scan, I take good care in setting my white and black points and my output contrast curve. In Photoshop, I touch up the white and black points with a levels correction. I often find that I don't need a curve correction at all. I do much less dodging and burning (local contrast corrections) than I did making darkroom prints, and I get better shadow detail and highlight detail than I could in the darkroom. Make of this what you will.

[paulr]

"This is doubly perplexing, since Mr. Mutmanski and I are both using nearly identical drum scanners ..."

Hmmm. Two people using nearly identical materials and equipment, but getting completely different results.

Sounds a bit like photography ; )

[paulr]

"Next on the list seems to be developer type (solvent vs. acutance). It's a trade off. To get smaller grain size, you give up some acutance. To get higher acutance, you get larger grain"

This seems like as good a rule as any with traditional emulsion films like tri-x. However, my tests have challenged whether this is still the case with t-max (or whether the relationship is still as simple).

I did find that rodinal produced the expected higher grain and edgier look on t-max. But when I experimented on variations of D-76, I found that accutance increased slightly as solvent levels were reduced, but grain size actually went down too. My current developer is a homebrew designed specifically for tmax 100. It has the lowest sulfite level possible for the working solution, and the results are for all practical purposes grainless. This would not be the case at all with Tri X or Agfa Pan or FP4.

Another difference is response to activity. Traditional emulsions get grainier if you use a high-activity developer/short development time combination (vs. a longer time in weaker developer to get the same density). They also seem to develop tones better with more time and less activity. I find t-max to be the opposite. Kodak hints at this by suggesting warmer developer temps, but to really take advantage of what the film can do, really goosing the developer activity and keeping the time down to 5 minutes yields wonderful results. Whether you're scanning or printing onto traditional materials.

[Michael Mutmansky]

Bruce,

I don't think your results are inconsistant with mine, it's just that we have different input variables that make direct comparison difficult.

I agree that developing less will reduce the grain size a bit, but much more significantly, using a slower, tabular grain film will improve it considerably.

At that point, the grain will be small enough that it will not be resolved at normal scan sampling sizes, whereas with Tri-X, you will have much more considerable grain to contend with.

I may not have been precise enough in my posts above, so I will clarify a bit. With a drum scanner, the CI of the negative will not be as important as with a flatbed, due to the inherent low noise levels. There will probably be a subtle benefit to using a lower CI in the negative for grain purposes when shooting a faster film. With slower, tight grain films, I don't believe the difference will be significant.

However, with a flatbed scanner (and this is what I was responding to primarily) the increased manipulation in the file may raise the noise floor considerably, and this would be a larger detriment than the slight benefit that is gained by reducing the grain size. Also, consumer flatbeds can't resolve to grain level anyway, so grain ailasing effect is not going to rear it's head with these scanners.

I believe the majority of problems with a flatbed scanner are related to noise, and not grain issues, and it makes sense to tackle that as the primary source of digital hash in an image.

Also, it seems that it makes sense to be developing negatives so they are useable for other purposes, like traditional darkroom applications. My suggestion was not that the film be developed more than normal, but that normal development may be better due to the noise consideration.

Since Jack is talking about using Acros (an excellent film, and my choice for 4x5 work, BTW), I don't belive that reducing the CI of the negative will make an impact on the grain unless he is doing considerable magnification and very high resolution scans, well above 4000ppi, if even then.

I do agree with your conclusions for Tri-X film, and although I haven't tested it myself for this purpose, I expect you are right on the money for your needs, as Tri-X grain does easily become rendered when scanning at 4064 in the Optronics units.

-----

Folks, Bruce and I are talking about differences in scanning results that are well beyond the capabilities of flatbed scanners, and only really apply for very high magnification scanning and printing. These are likely irrelevant discussions for flatbed users. The results of the scans that Bruce did are probably extremely subtle, and while the differences are visible such that a direct A/B comparison is possible, they represent the absolute last percent or two of an image's quality.

I expect that the differences he compared will become negligable when using Acros and a staining developer. I would like to hear if Jack does a comparison test. An extreme case would be a low-contrast (non ABC) stained negative or a sulfite-based developer negative compared to a high contrast Rodinal sheet, as these are probably the two ends of the spectrum as far as grain size goes.

---Michael

[Bruce Watson]

Michael,

Jack's original question is about drum scanning, film / developer combinations, 30 x 40 inch maximum print size, and grain worries.

I think the root of my contribution to this discussion is just this: I can and routinely do make grainless and razor sharp 40 x 50 inch inkjet prints from drum-scanned 4x5 Tri-X.

If I can do that with Tri-X, any film and developer combination that Jack decides to use is likely to work well for him.

One of the joys of LF photography is that it frees you from having to think of films in terms of graininess at all. You can choose a film / developer combination for how well it generates the tones you want in your prints, and how it responds to the controls you choose to use. This in my book is a much nicer way to photograph.

[Michael Mutmansky]

Bruce,

Yes, that's a good way to sum it up for Jack, and a very appropriate general statement that answers the question that he originally posted.

As the discussion evolved, a lot of other issues were introduced that may cloud the picture a bit, so this is a great statement to make a clean answer to the OP.

--Michael

[Jack Brady]

WOW - what a wealth of information!
You folks are great!
This has turned into a very informative exchange and has been handled by all in a very professional way.

Bruce, you now have the answer as to why I posted my question here - if you compare the comments that I received at other forums with this thread, then you'll understand why I have now made this forum my default Home Page!

There is much to absorb in this thread for me - I've read the thread a couple of times and will study in more detail tonight.

To answer one questions about my printing resolution procedure - I scan 4x5 at 4K - take in to CS2 and do my "magic". I then, in CS2 drop the res to match the print size I'm going to using the Edit->Image Size tool. I then save the resized image in a separate folder that hold resized for priint image files. This way, I never touch the 4K original file and use it to "clone" my print resolution files. I will start printing at 720 and see how that looks - makes sense to me. I just went to the max of 999 as I learned that is the max that CS2 and pump out the the Epson 4800.

I've just received some Acros 4x5 from Jim at Midwest - I should have my X-Rite 810 delivered by the end of the week. Thus, next week I'll start test with the Acros 4x5 sheets and both the WD2D+ and the PMK Pyro from PF.

I'll be happy to post what I find as results.

[eman]

I just took a break from CS2 working on an image that I scanned - 2 1/4 B&W scanned at 8K with 8 aperature. Had to get away from the image! Came over to my PC and started reading the replys.

Working with it in CS2 and trying to get the sky right is introducing all kinds of digital noise!

I think scanning flat has it's merits depending on the scanning software, scanner, etc., you're using and how wide a tonal range you're trying to cover.

My drum scanner isn't as good as yours and I'm working in OS 8.6 to accommodate the software. I bring as much contrast into the preview as I dare to avoid clipping which usually ends up being fairly flat. My scanner's preview leaves a lot to be desired so I have to be very conservative w/contrast. Bringing lots of contrast in PS will increase grain or noise if one goes too far (and I've had that happen) but typically here it does not. Since I'm working in OS 8.6 I don't calibrate that monitor and in fact the monitor I use there isn't the greatest either, but is adequate for the job of previewing and one learns what it's capabilities are through use.

Working w/a modern OS to do the scans and working on the monitor you'll be editing with gives some distinct advantages over my workflow. Your scanning software probably has much more control than mine with a much more accurate preview, so with your workflow, you shouldn't have to go too flat. With my workflow I do and it works out well.

It's frequently mentioned that good drum scans depend on the scanner operator. By working with your machine and testing these different approaches and working with images of differing tonal ranges, it seems you're going through the proper steps to achieve that end.

Eric

[Lenny Eiger]

Eman,
You just restarted a thread from 5 years ago.

Last time I talked to Jack he had just bought a Hasselblad 50 megapixel thing. He's gone digital.... a lost soul... ;-)

Lenny

[eman]

A super big WHOOPS!
A doggone Google Alert brought me here and I didn't even look at the posting dates. It's so seldom I have anything to offer on the subject. Wouldn't ya know it's from '05. :(
Gomenasai


Eman,
You just restarted a thread from 5 years ago.

Last time I talked to Jack he had just bought a Hasselblad 50 megapixel thing. He's gone digital.... a lost soul... ;-)

Lenny