PDA

View Full Version : Scanner DPI Options



Bart B
15-Sep-2011, 16:25
I'm searching for a decent scanner for my 4x5 B&W negatives. Something that will do a good job on negatives with images resolved in the 30 to 60 lines per millimeter range my lenses produce.

The largest prints I'll make from a 4x5 negative are 20 x 30 inches. From 120 film 2.25 x 2.75 inch negatives, 16 x 20 inch prints are about as large as I'll go. So I'm thinking that a scanner that does 1600 or 3200 dpi should do the job. Such scanners are out there for 70 to 140 bucks used. I think 1600 dpi should suffice so that's what I'm considering.

Question is, is 1600 dpi enough or should I opt for a 3200 dpi one?

Michael E
15-Sep-2011, 16:53
Question is, is 1600 dpi enough or should I opt for a 3200 dpi one?

The question is, does the scanner really deliver the DPI figure in the specs...

Michael

Bart B
15-Sep-2011, 19:05
The question is, does the scanner really deliver the DPI figure in the specs...

Michael, that's a good question. I'd guess flatbed negative scanners deliver close to specs. If not, then a higher spec'd resolution would best ensure you got at least what your minimum is.

If anybody with an Epson 1600, 1680 or 3200 used to scan B&W negatives that could offer setup vs. actual dpi numbers, that would be of great help.

Nathan Potter
15-Sep-2011, 19:10
It can be rather difficult to determine exactly what a scanner delivers from the advertised SPI figure.

Firstly consider that the SPI figure says nothing about the contrast at that SPI.

Next you will not know if that SPI is specified for capturing the full density range of your image. You can always increase the contrast to boost the apparent SPI figure at the cost of sacrificing valuable intermediate densities and/or densities at both ends.

Lastly the size of the aperture used when scanning is seldom specified. The aperture as referenced to the film surface can resolve detail in the image of about twice that aperture diameter. So an aperture of 25 um (.001 inch) will adequately resolve detail about twice that diameter with no bets on the contrast within the 50 um. That's just due to Nyquist limitations.

Your requirement of 30 l/mm (15 lp/mm) at the low end corresponds ideally to about 770 SPI and at the high end 60 l/mm (30 lp/mm) to about 1500 SPI. so you're in the ball park at 1600 SPI. But if you want some real detail from the film you'd better go for 3200SPI which would, I think, from most experiences here, provide you with a visually pleasing 20 X 30 inch print that would stand up under close inspection.

An example of the fallibility of scanner SPI specs can be found by actually measuring the performance of the scanner using a resolution target. I've been doing this with an Epson V750 just so I could understand what I was getting from the scan vs what was on the film. The results don't particularly match the scanner claims, although the claims are of course quite vague and imprecise.

I have a ton of data, but here are two test runs of interest. Focus point 3.37 mm above the platen.
4X5 Film Holder and using Film (with Film Holder). 2400 SPI with no Adjustments or Sharpening.
Scan of 20 um lines and spaces (25 lp/mm; 1270 SPI). Contrast determined by the difference divided by the sum method X 100, with readings obtained by measuring the K values of dark lines and white spaces at 1600% magnification in PS. Post histogram adjustment of K values from 1% to 99% for full density range in PS.

Positive Film setting with RGB scan = 28% contrast at 25 lp/mm (1270 SPI actual)
B&W Negative Film setting scan = 54% contrast at 25 lp/mm (1270 SPI actual)

BTW, consider what happens when the scan is done at 45 lp/mm (2300 SPI) still using the 2400 SPI machine setting:

Positive Film setting with RGB scan = 4% contrast at 45 lp/mm (2300 SPI actual)

At 52 lp/mm (2700 SPI) the contrast falls to zero, there is no discrimination between lines and spaces. All is greyed out. But also note that this is for a full density range condition, 1 to 256.

Nate Potter, Austin TX.

Michael E
15-Sep-2011, 23:35
I couldn't have said it better. I really couldn't. I hardly understand any of it. :-)

This is a good example: According to Nathan's tests, the Epson V750 delivers 2400 spi. According to my german dealer, it delivers ca. 2200 ppi. Epson claims 6400 dpi. Apart from the confusing nomenclature (I think Nathan got it right with his use of spi), it shows that even a very good scanner doesn't necessarily conform to the numbers on the box. It's probably worse with el cheapo models.

Besides that, an Epson V700 or V750 would probably be a very good choice for your application.

Michael

Bart B
16-Sep-2011, 04:08
Nate, thanks for your excellent post.

I meant 30 line pairs per millimeter in my original post; I know it takes 60 lines, half white and half black alternately positioned, to see 30 of 'em across about 1/25th of an inch. Apologies for not being clear.

Found this web site:

http://www.largeformatphotography.info/scan-comparison/

....that, to me, shows pretty good comparisons of scanner results. Compared the Epson 3200 to their V750 and there is a difference. Not much, but it can be seen. Biggest difference is in the shadow areas and that's where the V750 is much better. I don't know if that much difference can be seen with B&W negatives.

Jim Jones
16-Sep-2011, 07:17
. . . Compared the Epson 3200 to their V750 and there is a difference. Not much, but it can be seen. Biggest difference is in the shadow areas and that's where the V750 is much better. I don't know if that much difference can be seen with B&W negatives.

4x5 negatives scanned on an Epson 3250 made 16x20 prints that suited me. However, your standards (and those of others) appear to be higher. Certainly the loss of contrast in dense areas was a limitation.

Nathan Potter
16-Sep-2011, 18:55
I've plotted the contrast vs spacial frequency (and SPI) for 12 data points using a step function resolution mask.
This for an Epson V750 scanner on the B&W negative setting.

While this is a kind of academic exercise it gives a bit of insight into the scanner capability for the indicated settings. This is what you get by scanning at 2400 SPI. Notice that for 30 lp/mm (1500 SPI) the contrast is about 50%. If you can tolerate 10% contrast you get a considerably higher 43 lp/mm (2200 SPI) capability. As suggested by some on the forum here 2200 is about the upper limit of the V750 for B&W. Color scans reduce this number significantly; maybe even halve it.

Bart, maybe this makes it a bit easier to see how this works; it's really fairly complicated especially when you fold in some requirements for maintaining the full density range from the film master.

Nate Potter, Austin TX.

http://farm7.static.flickr.com/6171/6154033421_f5c72e4169_s.jpg (http://www.flickr.com/photos/argiolus/6154033421/)
EPSONcont-web-1 (http://www.flickr.com/photos/argiolus/6154033421/) by hypolimnas (http://www.flickr.com/people/argiolus/), on Flickr

Bart B
16-Sep-2011, 19:24
Bart, maybe this makes it a bit easier to see how this works; it's really fairly complicated especially when you fold in some requirements for maintaining the full density range from the film master.Nate, thanks for the link to the curves. I understand better now that I've seen the data.

'Twould be nice to see the results of scanning a really good B&W negative at different scan lines per inch; 800, 1600, 2400, 3200 and so on. I'm gonna search our nororious web to see if I can find some.

Nathan Potter
16-Sep-2011, 22:42
BartB, I'm a workin on 6400 SPI @ B&W setting on the V750 but that's a lot of data with a lot of pixel by pixel density measurement.

Even though the resolution mask is only two densities, clear and opaque I need to be careful that the full density range 0 to 256 is within the histogram in order to insure that the % contrast is computed about correctly. Even that poses problems due to the higher spacial frequencies being smaller than the scanner aperture.

Nate Potter, Austin TX.

Ken Lee
17-Sep-2011, 03:58
The standard resolution for printed magazines is a 150 line screen, IE 300 dpi.

As Nathan points out, high contrast detail works better, but on average the Epson gets somewhere around 2100 to 2300 spi.

Therefore you can enlarge by a factor of around 7x or 8x at most, and still get 300 dpi.

If some of your negatives contain only 30 lp/mm, then you ought to try this before purchasing the scanner: You may not be delighted with the results, even with a perfect scanner. They may not be up to the standards of Large Format photography. Find out for yourself.

Bart B
17-Sep-2011, 12:58
Found this site:

http://www.largeformatphotography.info/scan-comparison/

Decent comparisons, I think, to compare different scanners using the same negative.

Ivan J. Eberle
18-Sep-2011, 07:52
Ken, the standard for high speed web printing screens (magazines) is 133 dpi. For better quality presses it is 150 and sometimes 200 dpi, rarely higher with the possible exception of fine art repro in books

Bart B
18-Sep-2011, 08:34
When I did some technical documents working at Hewlett-Packard, the graphics group standard for printing marketing photos in brochures and such was 300 dpi. As explained to me, that's based on the human eye being able to resolve no finer than 6 line pairs per millimeter at a viewing distance of about 12 inches. So, 12 lines/dots per mm equals about 300 dpi. A large print (4 x 5 feet) viewed at 6 feet could have 50 dpi and look very sharp to most of us.

Of course, there's contrast and other issues at hand. Checking around the web for this info showed numbers from 180 up through 360 dpi. As long as the dpi value gives a reasonably sharp image for the viewer as seen from his viewing distance, all kinds of numbers will emerge.

Bart B
22-Sep-2011, 14:28
Well, my next phase of scanning 4x5 B&W negatives is using the Epson 3200 I just bought; that is, as soon as it arrives in a box.

It doesn't have the SilverFast software CD, but I'm wondering if it's really needed for scanning B&W negatives. Some web sites say it's not needed as Epson's software does a good job for B&W negs. Yet, a few say SilverFast's needed even for grayscale negs.

I'll post process scanned files with Photoshop Elements 8.

And SilverFast's web sites list several versions available for Epson's 3200.

Ken Lee
22-Sep-2011, 14:53
Ken, the standard for high speed web printing screens (magazines) is 133 dpi. For better quality presses it is 150 and sometimes 200 dpi, rarely higher with the possible exception of fine art repro in books

From what I can tell, lpi and dpi are not interchangeable. LPI appears to be equivalent to line pairs per inch, which is double the number of dots per inch or dpi, IE double the number of pixels per inch or ppi.

According to this discussion of standard line screens on PrintPlanet (http://printplanet.com/forums/computer-plate/12988-standard-line-screen) which describes itself as "the world's largest e-community for print":

"For 150 lpi you need your pictures to be saved in 300 ppi."

Some of the contributors mention that on high-quality stock, they use screens of even higher resolution. On poorer stock, they user lower resolution, naturally.

According to another member (emphasis mine): "Of course, 300 dpi is what we recommend, but just saying the least dpi we can get away with without seeing the "stair-step"."

I have corrected my original post to read "150 lpi screen, IE 300 dpi".

rdenney
22-Sep-2011, 14:55
When I did some technical documents working at Hewlett-Packard, the graphics group standard for printing marketing photos in brochures and such was 300 dpi. As explained to me, that's based on the human eye being able to resolve no finer than 6 line pairs per millimeter at a viewing distance of about 12 inches.

The standard sounds right, but HP really printed brochures using a 300-line process color screen? Color me doubtful--the ink spread would turn that into a mess using web offset.

I think you mean they required photos at 300 pixels/inch for reproduction. That is not uncommon as the starting point for 133-line magazine process color.

Or maybe they were printing production marketing stuff on high-production laser printers rather than using process color lithography. HP lasers have always been based on multiples of 300 dpi.

Rick "who always provided 300 ppi files to magazines" Denney

Ivan J. Eberle
22-Sep-2011, 15:11
No, you were talking about resolution at line screen for magazines, Ken; lpi is the term in the traditional ink printing industry. Where you may be confused is that it was generally considered necessary to oversample when scanning by a factor of 2X from film, so film was scanned at 266 dpi (that is, at the final repro size in print) or greater for a 133 lpi screen. 300 dpi scan at the printed size dimensions would be specified if 150 lpi screens were used. Of course in CMYK color print processes the dots were overlaid in a rosette pattern and screened at various angles to prevent moire so looking at it under a microscope there'd be more than 133 dpi total in the final printed page-- but 133 lpi was the resolution of the half-tone screens used in the vast majority of web printer processes.

Ken Lee
22-Sep-2011, 16:29
Excellent - Thank you for the helpful explanation.

Kirk Gittings
22-Sep-2011, 17:05
I think Sandy King did some actual resolution tests posted in some thread here not too long ago where he found the 750 to deliver 1600 if I remember correctly. That agrees with my tests too. The 4990, 4870, 3200 will all incrementally go downhill from there. I've owned all of them and used them extensively wet, dry, upside down-you name it. With 4x5 i won't use one for prints over 11x14. Just my personal opinion.....

Nathan Potter
22-Sep-2011, 17:55
Kirk, I tend to agree with you. If you look at my recent contrast plot (this thread) I get about a 50% contrast at about 1600 SPI (for capturing a full density range) in B&W. An 11 X 14 print, well executed, ought to yield a museum quality print that will stand close viewing scrutiny.

Nate Potter

Bart B
22-Sep-2011, 18:00
The standard sounds right, but HP really printed brochures using a 300-line process color screen? Color me doubtful--the ink spread would turn that into a mess using web offset.

I think you mean they required photos at 300 pixels/inch for reproduction. That is not uncommon as the starting point for 133-line magazine process color.Well, yes, I did mean 300 pixels/inch (ppi) photos for reproduction. I retired from HP almost 17 years ago; 'tain't easy for me to remember all that stuff.

Thanks for setting me straight.

rdenney
22-Sep-2011, 19:06
Kirk, I tend to agree with you. If you look at my recent contrast plot (this thread) I get about a 50% contrast at about 1600 SPI (for capturing a full density range) in B&W. An 11 X 14 print, well executed, ought to yield a museum quality print that will stand close viewing scrutiny.

Nate, I suspect that when people report ultimate resolution of 200-2400 spi, they are probably talking somehwere down in the 10-20% MTF range. A little sharpening can bring that up reasonably. It's probably okay that the stuff that ends up pushing the 8 line pairs/mm range on that final print has a lower MTF than 50%. What will make the print look sharp is probably down around 4 or 5 line-pairs/mm, where the MTF can be up around 50%. So, that solid MTF at 50% (with lower MTF at a slightly higehr frequency) might still support as much as an 8x enlargement. I personally have not taken mind past 4x, though--16x20 is the biggest print my Epson printer can make from 4x5. 1600 really honest pixels per inch is fairly useful at that print size. Museum quality? Not for me to say. But it's certainly at least as good as anything I could get under an enlarger. (That's anything I could get--for me the relevant comparison is the best I can do in my home darkroom versus the best I can do with my home scanner and printer.)

Rick "who wants to own his own quality even if that limits print size" Denney

Kirk Gittings
22-Sep-2011, 19:29
Kirk, I tend to agree with you. If you look at my recent contrast plot (this thread) I get about a 50% contrast at about 1600 SPI (for capturing a full density range) in B&W. An 11 X 14 print, well executed, ought to yield a museum quality print that will stand close viewing scrutiny.

Nate Potter

Thanks. I totally agree. Above 11x14 I get increasing mush requiring too much sharpening, but it still doesn't look right. Give me a drum scan for anything larger. I know they are expensive but why put all the work into a file if you are going to limit yourself to 11x14? Just my personal view.