optimal settings for 4x5's from a 4870

[JoelBelmont]

I am looking to make some smaller prints, probably on Epson's new 3800, both 11x14 in size and possibly 16x20. I know the 4870 is not the best scanner in the world, but I am under the impression that smaller prints will look fine. Would 16x20, from a good 4x5 neg be feasable?

My main question is to get a file for printing at this size, what settings would I use? Just a file that is 16x20 @ 300dpi, or ? I have not used it for this purpose yet, and would appreciate any help in what to do to get the best scan the scanner is capable of.

Thanks,
~Joel Belmont
www.joelbelmont.com

[jim kitchen]

Dear Joel,

Your suggested print size quality should be more than acceptable from a well scanned image. Do you have Photoshop to edit the file before you print?

I would try and retrieve the largest file possible from your scanner, edit the file accordingly in Photoshop, prep the file for printing, and see how your test print looks. There are many schools of thought, regarding a correct minimal scan, and the net is at your disposal to help you achive that goal. The following locations should get you started:

http://www.scantips.com/

http://www.photosig.com/articles/1449/article

There are several qualified users in this group that will help you along...

jim k

[Keith S. Walklet]

Bump Jims comments.

Assuming a color transparency:

Scan at the highest optical resolution the scanner supports (with the 4870, the limit for 4x5 color is 3200spi when all extras--dust reduction, etc. are turned off), no sharpening, 48 bit, scanned into a large color space such as Ektaspace.

Make sure to set the white and black points properly, and I've found bumping the gamma to 1.2 seems to help.

The advantage of the larger scan file is that you will capture more data which will be discarded in the event the final output doesn't require it. Downsampling helps minimize noise.

Once you have the raw scan, you can use the Layer Transfer Technique to speed your workflow.

BTW, 16x20 prints from a 4x5 scanned on that scanner should be excellent if you handle the file properly from start to finish. ie. a good clean scan, proper editing in your image editing program, proper sharpening for output, proper profile for the device and substrate it is printed on, and a good output device.

[Jeffrey Sipress]

Layer Transfer Technique?

[JoelBelmont]

Thanks.

Jim: the first link was more helpful, but even that was a bit to weed through to get what I am after.

Keith: this is the info I need. What the highest settings are on this particular scanner that I can/should use. I am scanning B&W negs, though. Also, is it best to scan in 48-bit, then convert to greyscale in CS2, or just scan 16-bit grey?

It looks like the higest res I can scan a 4x5 is:
for a 16.5 x 21" scan (room to crop to 16x20), 1800 dpi, 2GB file.

It is my understanding that there is no point in printing beyond 300 dpi...?
So it also allows me like an 80 x 100" output size at 300 dpi, 1GB file (and it can go larger.

So my effective question is, and my apologies for being a novice in this area, is what is the ideal size to scan to, given the eventual downsample to a 16x20?

Also, what is the layer transfer technique?

Thanks for your time.

~Joel Belmont

[Keith S. Walklet]

Joel,

Here's a link to a thread where I describe the Layer Tranfer Technique, which was originally shared with me by Rich Sieling of West Coast Imaging.

http://www.largeformatphotography.in...ad.php?t=12528

I don't use my 4870 to scan much B&W, but know that many of those that do, favor scanning in RGB and either selecting the best color channel (red, green or blue) as the data they work with, or just convert to B&W post scan.

The downside to scanning as a RGB image is a large raw scan file and a lower spi threshold (3200spi instead of 4800spi) due to the scanner software memory buffer limit.

The upside is that when one downsamples either by decreasing the physical pixel dimensions of the output, or by throwing away color pixel data when converting to black and white, the result should mean less noise.

When I am scanning, I am not concerned with a particular size output, and just scan to capture as much data as I can at the scan stage, and customize that data in an edited master file that can be used for many different ouput devices and sizes.

For me, that means scanning at the highest optical samples per inch the scanner supports (which on the Epson is 3200spi with a 4x5 in RGB 48 bit mode). You can probally go to 4800 spi if you scanned the file as a monochromatic image.

When I go to print that optimized output (I use a Epson 9600), the file resolution size is 360 dpi, and the printer resolution setting is 1440dpi, as I am using Premium Luster paper. A matte paper (such as Epson Textured fine art) would typically require a lower printer resolution setting.

If you want to read an excellent description of this chain of spi and dpi and what the eye is able to discern, I highly recommend you get a copy of Bruce Fraser's latest "REAL WORLD" book on sharpening. Of the three in that series I own (Photoshop, Scanning, Sharpening), it is the book on sharpening that best addresses this issue.

[Brian Ellis]

The May/June 2005 issue of View Camera magazine contains an article by Ted Harris, one of the moderators and frequent contributors to this forum, that includes the results of various tests he conducted on various scanners, including the 4870. The test results showed that the maximum ppi that the 4870 is capable of resolving is about 2050 ppi. If that's the case then I don't think there's any benefit to scanning at a ppi greater than that with the 4870, the only result will be a bigger file than necessary.

[sanking]

The May/June 2005 issue of View Camera magazine contains an article by Ted Harris, one of the moderators and frequent contributors to this forum, that includes the results of various tests he conducted on various scanners, including the 4870. The test results showed that the maximum ppi that the 4870 is capable of resolving is about 2050 ppi. If that's the case then I don't think there's any benefit to scanning at a ppi greater than that with the 4870, the only result will be a bigger file than necessary.

I don't understand it this way. I use this scanner (Epson 4870) and what I have found is that the actual resolving capability is a certain percentage of the theoretical maximum. For example, if I scan at 2400 spi the acutal resolving power is at *most* only about 50% of that, or about 23 lppm, not 47 lppm. If I scan at 4600 spi the actual resolving power drops to about 40% of potential, or about 35 lppm, not the 90+ lppm that would be suggested by potential. These are actually fairly optimistic results.

My experience in scanning 5X7 negatives iwith the Epson 4870 is that I can print at about 3X negative size with very good image quality. For printing at more than 3X I don't believe the Epson 4870 will do the job.

There appears to be some agreement that the best technique is to scan a B&W negative in RGB and then discard the R and B layers. Whether you can actually do this depends on the capability of your computer to process really big files.

Sandy King

[Keith S. Walklet]

Brian, Ted's observations may be true, but he is referring to "measured" resolution, while I am suggesting that Joel use "4800 spi setting" in the scanner software, which, regardless of what the actual/measured resolution is, instructs the scanner to scan at the resolution that happens to be the maximum optical resolution. What the "claimed" optical resolution of the scanner is doesn't matter. It is just a matter of getting it to scan at the highest uninterpolated resolution it can possibly capture.

I shared my prints with people at the View Camera Conference in June, mainly to support my posts here, because the quality of the results I was able to achieve really surprised me.

And what the prints show is a steady increase in resolution until one goes beyond the optical limit of the scanner (4800spi) into the realm of interpolated (9600spi). I created the sample prints simply to see for myself and share with my students, and didn't measure the results with anything except my own eyes. But the results don't require a measurement to quantify. The improvement in resoultion is easy for anyone to see.

As for reproduction size, I've seen many rule-of-thumb suggestions here, and the truth is that they may be a good starting point, but the results one achieves depend on image content, film size and type, the quality of the scan, image editing technique, proper sharpening and the quality of the final output device. My samples were generated from a 35mm transparency printed to 20x24 because I thought I'd see the biggest difference in results from the smallest piece of film. Instead, it opened my eyes to the potential of the equipment.

I just had an exhibit last month where an individual asked what scanner I was using. I replied TANGO, NIKON 8000 and EPSON 4870. I had ten 20x24s on display created from that mix of scanners and asked them to tell me which was scanned with what scanner. They couldn't. If one does everything right, the results are surprising.

But, if one can start with a TANGO scan, it is akin to playing a round of golf from the ladies' tees. It will be easier to get better results because the raw scan should be better (again, the operator makes a difference). That doesn't mean that one can't play from the blue or white tees. It just might take a little more care/work to get there. ;-)

[sanking]

And what the prints show is a steady increase in resolution until one goes beyond the optical limit of the scanner (4800spi) into the realm of interpolated (9600spi). I created the sample prints simply to see for myself and share with my students, and didn't measure the results with anything except my own eyes. But the results don't require a measurement to quantify. The improvement in resoultion is easy for anyone to see.

As for reproduction size, I've seen many rule-of-thumb suggestions here, and the truth is that they may be a good starting point, but the results one achieves depend on image content, film size and type, the quality of the scan, image editing technique, proper sharpening and the quality of the final output device. My samples were generated from a 35mm transparency printed to 20x24 because I thought I'd see the biggest difference in results from the smallest piece of film. Instead, it opened my eyes to the potential of the equipment.

The increase in actual resolution up to the optical limit of the scanner is also what I have seen in testing this scanner.

As to final print size, there are indeed many things that have to be taken into consideration. Just for the record, I always base my own decisions on resolution figures that will give at least 8-10 lppm at the printing size since I consider this the minimum needed for critical viewing at about 10". In that context about 3X negative side is the highest I would go.

Sandy