Scanning large format with vuescan

[paulr]

I'm wondering if anyone has experience scanning 4x5 negs with vuescan, specifically with an epson 4870.
I'm new to both, and am trying to cut the learning curve down to size.
My project is for a book that will be printed with piezography. The enlargements will be small (up to 3X). I'm tryingto get the highest quality possible.

First question is about multiple pass scanning. Is more than one pass a good idea with this scanner? Not sure if the possible noise reduction is worth the possible misalignment and blurring.

Second question is about digital ice dust removal. I've read in other forums that people were having trouble getting ice to work properly with vuescan. not sure what version they were running though. Any thoughts?

Finally, for BW work, how do you like vuescan compared with epson scan?

Thanks for any help ... any other ideas you can throw my way.

Paul

[Jim Ewins]

I consider vuescan an expensive lesson. But then again I continue to do dumb things. Jim

[julian_4860]

Paul, multiple passes don't work too well on a budget scanner as the head has to go back to the exact same spot each time. More exact scanners use multiple sampling to achieve the same end. Its easy to test, do a scan with and without multiplepassing and see if it is softer by comparing the scans 'at pixel' view.
Vuescan is the best way IMO of getting BW scans. Scan at 'image' 16bit and you get a neg in PS which has had no messing with by the scanner software. make sure you use Vuescan in Advanced mode and set both the blackpoint and whitepoint to '0' so that the histo is unclipped - the default is an 'idiot' mode that gives you clipped black. Don't forget to set the colourspaces away from sRGB too
Digital ice is ok for workprints but it works by softening the image slightly. You want the sharpest image possible , I guess, so leave it alone and spend the time spotting!
There's a big learning curve here, just as in a wet darkroom. There's loads of 'pluggins' and 'auto' this and that, but they all (well nearly all) degrade the image somewhat.

[Leonard Evens]

I've been scanning 4 x 5 for a while now with the Epson 3200. The 4870 is similar but resolves a bit more fine detail. I experimented a little with multiple scans, but I don't find it necessary. For what you are doing, the basic scan, even without sharpening will probably suffice. If you do sharpen, use one of the techniques which isolates regions around edges and sharpens only in those regions.

The 4870 apparently can't resolve much fine detail above 35 lp/mm if it can even achieve that. You can probably scan at 2400 ppi without any visible loss of detail. For b/w you are not likely to have any problems with dynamic range.

If you are careful with your negatives, you should have to do little or no digital spotting (with the clone tool). You shouldn't need to use any form of digital ice. I often need no spotting at all, and even in the worst cases I can do it all in five or ten minutes.

I find Vuescan to be much superior to Epson's scanning software. But it would be nice if Ed Hamrick added curve controls to it. It is on his to do list and eventually I think we will get it.

[paulr]

Thanks for the responses. So far I'm finding that 2400ppi is seeing all the detail that the scanner can resolve. I'm experimenting with scanning at 2400, then downsampling in in ps to double my printer driver's native resolution (at print size). For the biggest print size this is about 1750 ppi. I'm thinking it's best to scan at an even factor of the scanner's optical rez, and to leave the odd resampling to photoshop. Does this make sense?

Julian, you suggested setting the colorspaces away from sRGB. What colorspaces make sense for black and white? Photoshop's color settings have options like Gamma 1.8 grayscale, but I don't see anything like that in vuescan.

Digital ICE is completely butchering my images in vuescan ... I'm seeing a pale ghost of all the detail, offset by several pixels. This doesn't happen in Epson scan. I sent an email to Ed Hamrick asking if this is a bug (with a fix on the way). Using Epson scan, ICE does not seem to degrade the image (except maybe right at the dust) the way despecle or dust+scratches would.

After a couple of nights of playing with it, I think the scanner will work well. It can resolve the grain of my t-max negatives, which is more than any of my darkroom prints have been able to do (up to about 4x enlargements), even with a well aligned enlarger, critical focussing loupe, and modern apo lens. So assuming it can separate the densest highlights in some of the older, contrastier negs, it should be fine for the project.

[Leonard Evens]

You should scan at successive divisors of 2. So for the 4870, the possibilities are 4800, 2400, 1200, etc. If you scan, for example, at 3200, the scanning software will scan at 4800 and interpolate.

Some people recommend rescaling for printing to a simple divisor of the printer's native dpi. This is a bit confusing because dpi is not the same as ppi. Each pixel may represent a very large number of colors but each dot is just one of 4 to 8 colors, depending on the printer. Many dots are needed to represent one pixel. However, before printing, the printer hardware takes the ppi you send it and converts to some native ppi before actually producing dots. For Epson printers, this is apparently 720 ppi. So if you send it 720 ppi or 360 ppi, you eliminate one further interpolation. Your photoeditor preesumably can do a better job of rescaling than the printer firmware or printer driver.

Vuescan controls the gamma through the brightness control, which is really a gamma multiplier. Also, your choice of film type and CI will determine the starting gamma to which the multiplier is applied.

It is highly unlikely that you are resolving grain with your scanner. It is at too high a level to be detected. What you are probably seeing is what is called grain aliasing. If you do a google search, you will find plenty of references and explanations of that term.

As I noted previously, don't bother with digital ice.

[paulr]

interesting article on grain aliasing at
http://www.photoscientia.co.uk/Grain.htm

I did some quick math; based on what I see with a 10X grain focuser, I would expect to need over 9000ppi resolution to resolve the actual grain. Never crossed my mind that it would require so much (not that I need it for anything).

[julian_4860]

Make sure you have the 'advanced' options tabbed and then look in 'output'. You'll see the output space there. I'd set it to adobe rgb and make sure your colour space in PS is the same. If you haven't calibrated your monitor - do it. There's a rough guide for BW on my website http://www.foundobjectsgallery.com/calibrate.htm. I'd set everything to a gamma of 2.2, dotgain of 20percent in PS. Scan for 'image' 16bit. White/black point to 0. Invert the neg in PS adjust the blackpoint in 'levels' so that only one pixel is fully black, and none clip when you set the white point. Then use curves to set global contrast. Use masks for any local dodging/burning.
The piezography website used to have a tutorial online for how to setup a softproof for BW. You might also find the bwdigitalprint group on yahoo useful

[paulr]

Thanks everyone, especially Leonard and Julian.
Very helpful ... there aren't enough days in the year to experiment with every single variable.
One dumb question: what does "CL" stand for in the film selection menu? It obvioulsy adjusts gamma but I don't know what the number represents.

[Brian Ellis]

"I consider Vuescan an expensive lesson."

Are you sure you're thinking of Vuescan? I think I paid about $50 for it, it's probably the least expensive program on my computer.