Comparing scans at different resolutions

[Peter De Smidt]

There's been some talk about various resolutions and scanning in other threads. My question is how should one go about comparing scans at different resolutions? They will give different image sizes in Photoshop. Should one downsample the higher res scan to match the lower one? I realize that the thing to do would be make prints from various files at the largest size that one might make, but in my case I currently only have a 13x19" printer, and I hope to have a much larger one down the road. Suggestions?

[Daniel_Buck]

What I've always done is find my target print size (you want larger than 19" wide, so maybe shoot for 30" wide as your 'test' resolution?) Then compare your scans sized (up or down) to your target print size. That makes the most practical sense to me.

[Bruce Watson]

There's been some talk about various resolutions and scanning in other threads. My question is how should one go about comparing scans at different resolutions?

It depends of course on what you are trying to find out. I'm guessing that you might be trying to decide how much scanner resolution is "enough" for your purposes. In that case, what I would do is to decide what print size will be the biggest you ever print. Then, resize without resampling the various scan files to that print size. This will result in files that will give the same print size but with different pixel densities (say, 360 ppi, 300 ppi, 240 ppi, etc.)

Print sections out of these files (for example, an 18 x 12 inch section out of a 40 x 30 inch print size). Put the resulting prints up on a wall and compare them side by side under the same lighting. Look at them close and from a "normal viewing distance" (if you can decide what that might really be). From this you should be able to decide on how much scanner resolution is enough for what you want.

But make the prints. It's nearly impossible to figure this out by just looking at the monitor. Monitor and prints are just too different.

[Ted Harris]

What Bruce says. Plus, generally, it is best to do all your scanning at the highest resolution your scanner can handle (~2400 for consumer scanners) or, when using a high end scanner, the resolution that produces the largest file size your computer can comfortably handle. Do this and you have a master file you can use as you please without ever having to scan again, no matter what your future print requirements.

[Peter De Smidt]

I appreciate all of the suggestions. I'll certainly take Bruce's advice and make some prints. With 35mm, I scan at the top 5300 spi that my scanner can do. With medium format, though, I scan at 3650 spi, which is close to the highest that my scanner can do without having to scan negatives in two strips and then putting the two together in Photoshop. This would enable a 360 dpi file to print at 20" on the short end. 4x5 would require three separate scans at 5300 spi, and printing a 360 dpi file would produce a print of over 58 inches on the short end. While I could print something that big at work, I doubt I ever would. In the darkroom I usually made 11x14s from my 4x5 negs I expect that something in the 30x40" range would be as big as I'd ever go.

[Gordon Moat]

Generally agree that a print is the only valid comparison. However, if you are simply curious, there are a few steps that allow you to preview, even though it is a poor simulation. PhotoShop is biased towards showing images in 100% increments. If you use Image Size to set the final print output (i.e. 8" by 12" at 300 dpi), then view at 100% you will get about as close to a preview as possible.

Unfortunately this creates a few problems. First is that 100% view often goes beyond the screen dimensions, meaning you have to scroll around the image (or use the hand/grabber) to see all the image. Next is that you are viewing very coarse squares limited by your monitor, which will be much worse than a same sized print. Last is that you are viewing colour differently than it appears from a print, and all monitors clip CMYK to some extent, and most barely show NTSC or sRGB colour space.

A better way to test possible results would be to print a highly detail section of a larger image. That way you get a small print showing at least some of the detail. Then if you want a larger print later on, you have a nice test for a basis.

Ciao!

Gordon Moat Photography