Scanner Magnification Issues

[Kevin Convery]

I need help finding information on magnification issues from scanning negative from 35mm all the way up to 8x10. The final output for these scans will be to scan the negative, do some light photoshopping then printing them as 16"x20" digital negatives to be contact printed in platinum. The equipment used will possibly be Epson 4990 or V750 with a 3800 printer. So what types of problems will I run into when scanning original negs and enlarging them to 16x20 negs? I heard magnification was an issue and I've done from searching but am getting nothing. I was hoping someone had info or at least point me in the right direction.

Thank you.

[Ted Harris]

The problem you will run into is the resolution and shadow detail of the enlarged negatives. From 3x5 hey will be acceptable, barely. From smaller sized negatives they aren't likely to be what you are after unless your image will look good soft and had little or no shadow detail that you want preserve.

You can do it, you just can't do it well with a consumer scanner.

[Leonard Evens]

Let me first summarize the rules as they might apply to your setup. Divide the scanning resolution in ppi by the enlargement factor needed to go from the negative to the print. If that number is larger than 360 ppi, you are probably okay with respect to the printer. Also, divide 33 lp/mm by the enlargement factor. If that is greater than 5 lp/mm, then you are probably not going to lose significant detail in the print, independent of printing details.

Let me explain in greater depth where these numbers come from.

There are two ways magnification affects what you want to do.

First let me address the issue of pixelization in the print. You need to send the printer engine sufficiently many pixels per unit length in the print. If this number is small enough, you will be able to see individual pixels in the print, but even for higher numbers there may be subtle effects. How small this number can be depends on factors such as the distance from which the print will be viewed and how discerning viewers will be. The lowest figure I've seen recommended is 240 pixels per inch, but most people argue for a somewhat larger value such as 360 ppi. Some people claim to be able to discern differences up to 600 ppi.

To calculate the ppi you are sending the printer, you divide the horizontal and vertical dimensions of the image in pixels by the corresponding dimensions of the print. For example, suppose your image is 8000 x 10,000 pixels and you want to produce a 16 x 20 inch print. 8000/16 = 10,000/20 = 500 ppi, which is well above 360 ppi.

You can also do the calculation in the other direction by multiplying the print dimensions by the ppi you want to send to the printer. If you think 360 ppi is adequate for the printer and you want a 16 x 20 print, then you need (360 x 16) x (360 x 20) = 5760 x 7200 pixels in the image.

Note that this doesn't seem to depend on the magnification, but it does enter implicitly. A digital image has no resolution; it just consists of pixels, but when you scan and read the image into your photoeditor, it will probably report a resolution, which assumes some reference length. That may simply be the assumed monitor resolution, in which case it isn't of any use, but it may be the resolution of the scanner. In that case, the physical dimensions that are relevant are those of the negative. For, example, if you use the 4990 to scan a 4 x 5 inch negative at 2400 ppi, you are going to get (2400 x 4) x (2400 x 5) = 9600 x 12,000 pixels. (Actually it will be somewhat smaller because a 4 x 5 negative has a somewhat smaller size, about 96 x 120 mm.) If you concentrate on the image size in pixels, that is all you need to know. On the other hand, you can do a quick calculation. Knowing that you scanned at 2400 ppi and that you are going to enlarge about 4 times to produce a 16 x 20 print, you simply divide 2400 by 4 to obtain 600 ppi, which will give you an estimate of the ppi you will send the printer. Similarly if you scanned a 35 mm negative which is very close to 1 x 1.5 inches at 4800 ppi and you wanted to enlarge 16 times to produce a 16 x 20 print, then you would be sending the printer 4800/16 = 300 ppi, which many people would find substandard.

There is one other way resolution enters the picture. If you scan at 4800 ppi, digital sampling theory says you can't do better than half that or 2400 line pairs per inch in the resolution of fine detail. Divide this by 25.4 to obtain resolution in lp/mm, in this case, 2400/25.4 ~ 94 lp/mm. However, no real printer will achieve this theoretical limit, and a printer like the 4990 is more likely to yield at most 33 lp/mm. This is an issue distinct from pixelization. If you scan at a lower resolution than 4800, up to a point, you won't reduce this number, so that is also what you would get, for example, if you scanned at 2400 ppi. If you plan to enlarge by a factor of 4, as when going from 4 x 5 to 16 x 20, that reduces the resolution as seen in the print to 8-9 lp/mm, which is probably adequate. But if you enlarged by a factor of 16 (as for 35 mm) you would be reducing the resolution in the print to just about 2 lp/mm, which would be obvious in the print.

Note that the conclusion is that your setup is probably adequate for producing 16 x 20 prints starting with 4 x 5 but not staring with 35 mm. Of course this makes certain assumptions, the most important of which is that viewers are going to get up close---10 to 12 inches---to view the print, which is likely with a 16 x 20 print. On the other hand, the normal viewing distance for such a print might be considered to be closer two feet, in which case, you could relax the criteria. Viewing a 16 x 20 print at 24 inches is equivalent to viewing an 8 x 10 print at 12 inches. So you could instead do your calculations based on the smaller print. For 4 x5, the enlargement factor would be two and for 35 mm it would be eight. You would be producing enough pixels for the printer in either case, and as far as resolution of fine detail is concerned, for 35 mm, you would be just under an acceptable 5 lp/mm in the print, 33/8 = 4.125.

Finally, note that some of the reference numbers are only estimates. Your scanner may do better or worse than 33 lp/mm. You may not find 5 lp/mm acceptable. Similarly, you may find something less than or greater than 360 ppi adequate for avoiding pixleization. You can only settle such matters for yourself through testing your equipment. You can do that systematically or simply by trying different things and seeing what you generally find acceptable. Also, keep in mind that what is acceptable may vary considerable with the subject matter.