scanning large format

**Leonard Evens** · 2-Jul-2004, 07:16

Donal,

What you say is consistent with my (limited) experience and reading. It should be added however that some people (notably Kennedy McEwen) claim that the staggered array reduces aliasing. As you say, in some ways it is better and in other ways it is worse. In any event, for the price, the Epson 3200 (and the new 4870) give very good results, particularly for 4 x 5.

**Frank Petronio** · 2-Jul-2004, 08:53

Back in the old days of the 90s, there used to be a popular argument for scanning only at the resolution of the final output for optimal results. Downsizing a huge file means that the computer has to replace multiple pixels with one averaged one, and depending on the calculation it can produce worse results than simply scanning at the size you need to end up with.

Of course I don't always know what my final output will be when I do the scan, but I do try to scale my images by a factor of two, rather than some random fraction, in the hopes of averaging "whole" pixels rather than 1.37849 pixels.

**Greg Miller** · 2-Jul-2004, 15:58

"Back in the old days of the 90s, there used to be a popular argument for scanning only at the resolution of the final output for optimal results. Downsizing a huge file means that the computer has to replace multiple pixels with one averaged one, and depending on the calculation it can produce worse results than simply scanning at the size you need to end up with. "

Maybe I don;t get it. If you scan at output size doesn't the scanner have to do the exact same thing? Aren't you just switching the timing of the act of averaging?

Wouldn't you also have to consider any differences in the averaging algorithm of the scanner software with the editor (Photoshop?)?

**Frank Petronio** · 2-Jul-2004, 18:48

Yes, either the scanner or the software has to do it. The idea is to do it with less steps. That's why trying to do the gross optimizing with the scanner software rather than correcting it in Photoshop is better as well.

I think the common wisdom now is to scan at a multiple of two of the native optical resolution of the scanner. However, determining what exactly that is (see above) is often the problem.