A Site for Photographers by Photographers

Community > Forums > Digital Darkroom > Scanning>Technique > How to scan 35mm Slides

How to scan 35mm Slides

Chris Watts , Nov 08, 2008; 01:24 p.m.

I am in new territory here. Please could someone advise me about the most reasonably precise way to scan 35mm slides. I recently scanned a few on an Epson 4990 flatbed and was appalled at the output in CS3.

I apply no sharpening or adjustments when scanning photos, and enlarge in the scanner software rather than interpolating in CS3. But this was a disaster for my slides.

I do not understand how to turn this slide into a 12inch by whatever inch for output. I do not understand whether I need to enlarge by 1000% or scan as is at 10000 DPI so that I can interpolate in CS?

How do you get quality with a slide when it is so small in the first place?




    1   |   2     Next    Last

Richard "Dick" Tope , Nov 08, 2008; 01:58 p.m.

Probably everything you need to know can be found at http://www.scantips.com/.

Mendel Leisk , Nov 08, 2008; 02:04 p.m.

Here's an informative review on the Epson 4990:


I'm not clear if you've acquired this scanner or were just trying one. I would think you will get better results with a dedicated film scanner, but the Epson 4990 *may* be good enough. As far as dedicated 35mm film scanners go now, the list starts, and more-or-less ends, with the Nikon Coolscan 5000. The Coolscan V is a cheaper alternative that will do 90% of what the 5000 can do, and will deliver 98% of the quality, but it is discontinued. Currently there *may* be a few new V's still available.

The only caveat with Nikon's: if your slides are Kodachrome, both of the Nikon dedicated 35mm scanners will have problems, with color balance in general, and also with artifacts if you use ICE (infrared cleaning). The 9000 multi- format model is better with KC, but a big beast, very pricey, and somewhat limited in availability.

Regarding "appalled" and "disaster", you need to get more specific and objective in what's bugging you about your scans, LOL. What in particular are you seeing?

The image size a flatbed scanner can deliver is debatable, and dependant on the model as well. Unlike dedicated film scanners, the flatbeds are doing double duty, and there is likely some point, short of the scanner's advertised max dpi, that is the *practical* limit. In other words, you can scan at higher dpi, but will gain little in terms of actual detail. Have a read through the linked article for more info on this.

Also, you make no mention of the software you're using with the scanner, which I'd assume means you are using the Epson supplied software. Again, check out the link for suggestions on how to exploit it. Alternately, look into the Pro version of Vuescan. It is independant software, capable of running a multitude of scanners. It's interface will likely be just as bewildering at the outset, but it is a very capable, full featured tool, capable of getting the most out of your scanner (and many others). A demo with some features locked out and watermarking added to the output images is available for try out.

Gary Demuelenare , Nov 08, 2008; 02:45 p.m.

i use the nikon coolscan v for my slide scanning. scanning ektachromes. the coolscan v makes images files as a tiff that are 131mb each in size. they are, as farv as quality is concerned, undetectable for my dslr or the original projected slides. in other vwords they look great. this is for resolution or IQ or color. but note the coolscan is a 4000dpi scanner. it is capable of picking any detail that is there, it is also a scanner for slides or negatives only, not prints or other images.

when i got into the ideaf slide scanning i had to decide between the flatbed and the film only dedecated scanner. it is simply no contest in IQ between the flat bed a film only scanner. but the film only was $499. the flatbed is a lot cheaper.

Michael Axel , Nov 08, 2008; 11:15 p.m.

Chris, It would help to know what you are scanning. I use a 4490 which is supposed to be the same engine as the 4990, except for smaller formats. I find B&W negatives to be atrocious, but it does okay with 35mm slides. Attached is a quick scan. Also attached is my panel as it looked for this scan. The Dmax is okay so long as you scan in 24 bit or higher. Let us know what your panel looks like or what settings you're using and maybe we can help.

Quick 100% scan and panel

Chris Watts , Nov 09, 2008; 10:22 a.m.

Thanking you for your responses, i will first have a look at those web pages and then get back.


Chris Watts , Nov 09, 2008; 10:44 a.m.

Just had a quick look at the scantips site. I recognise it, was here once before but found it too much waffle in order to have a simple question answered.

Please could someone just tell me the following: I want to scan a 35mm slide. Do I scan it at the original size? then interpolate in Photoshop CS3 in order to get my print? OR do I scan it and make the target size greater, such as 1000% which is what I need. i want an 8x10 inch print. Then I do not have to interpolate in Photoshop. Also, is it 'generally' better to scan slides at 200 or 300ppi? Even one website said that you should Double the ppi for double the size. For example if you were scanning a 5x7 colour film in order to produce a 10x14 print on inkjet then you should scan at double the ppi so therefor 300 doubled is 600ppi. To be honest, Im fed up with all this complexity and just need to do the "Averagely' acceptable where good results are predictable without viewing it all under an electron microscope to be analysed by science geeks.

I appreciate your advice, and will return your messages.



Mendel Leisk , Nov 09, 2008; 02:00 p.m.

Regarding your question, some preamble:

There's a *lot* of superfluous math and ratio used in regards to print parameters. There are *two* concepts in play:

1. Scanning Pixels Per Inch (PPI)

2. Printing Dots Per Inch (DPI)

To further confuse things, the acronym DPI is sometimes employed, when describing scanning resolution, where PPI would be "better". But, this is so wide spread it's maybe best to just go-with-the-flow, accept DPI as being applicable to scanning resolution as well, and just keep in mind if the subject is scanning or printing.

To your question:

First, I would determine: what is the actual, practical, scanning resolution of your scanner. Looking at an Epson brochure, the 4990 is specified as having "optical resolution" of 4800 DPI. (See what I mean? Even Epson is expressing PPI as DPI.)

So, it's advertised as being able to scan at 4800 PPI. But, flatbed scanners in particular are notorious for having *actual* detail resolving limits much lower than their max PPI. How much varies from model to model.

The simplest and most direct way to sort out the maximum *effective* PPI of your scanner is to test it. Set your scanner to scan at 4800 "DPI" and scan a nice sharp, fine-grain slide with some good small detail. Then try lower resolutions, say 3200 and 2400.

Open two output images in Photoshop. Say the 4800 and 2400 outputs. Zoom the 4800 image to 100%, centring the zoom on some distinct detail. Repeat the exercise with the 2400 output, but zoom to 200% (this will "size" the images the same on-screen).

Toggle back-and-forth between the two images [(ctrl) TAB on windows system]. Compare resolved detail. Repeat the exercise with the various PPI output images. There is likely a level, somewhat lower than 48000, that offers the best resolution. Going beyond that level will gain you little beyond large file size.

Now you have to make a decision: do you want to scan at the maximum *practical* PPI limit (say for example this is 2400 PPI), or would you accept some reduction in fine detail in order to reduce output file sizes? And will the reduction still allow you to print the max size you want, at a decent print DPI (finally using that term in it's proper application)?

Ok, so say you've settled on scanning at 2400 PPI, what are your image dimensions, in *pixels*, going to be? Assuming your scanned area has dimensions of 1" by 1.5" (*roughly* the slide dimensions), at 2400 PPI, your scanned output image will have *pixel* dimensions:

2400 pixels by 3600 pixels

How big a print can you make from that depends on the DPI you print at. In all cases:

To determine the dimension of the print, divide the pixel dimension by the Dot-Per-Inch value you will use when printing.

For example with the 2400 pixels by 3600 pixel image:

If you print at *ONE* Dot Per Inch you could print an image 2400" x 3600". If you print at 100 DPI, you could print an image 24" x 36". If you print at a decent practical level like 240 DPI, you could print an image 10" x 15".

Hope this helps ;)

Mendel Leisk , Nov 09, 2008; 02:09 p.m.

Also, in that last example, the 10"x15" 240dpi print, if in your printing software you push your size up to something like 12"x18", say on a sheet of 13'x19" with 1/2" border all around, it will drop your printing DPI, slightly, but it will not be noticeable. Acceptable DPI guidelines are just that, nothing more.

Chris Watts , Nov 10, 2008; 02:17 p.m.

Hallo Mendel,

thankyou for taking the time to explain that. By your explanation then would I be right in assuming that you would prefer not to scan say at 300ppi and use the scanner's tick box which will re-size the image by whatever percentage you fill in. You would prefer to scan the original size as is and make it also the target size, but up the ppi to say 2400? This would then give me a dimension that I am seeking. I just want to know if it is preferable not to allow the scanner to increase the size:

For example, you scan AT 1x1.5 at 2400ppi rather than scanning at 300ppi but make the target size 1000%. OR is there no difference between these techniques?

    1   |   2     Next    Last

Back to top

Notify me of Responses