Re: What image type should I use for storing scanned images?

[Pete Hall <pbhall@xxxxxxxxxxxxx>]

At 09:59 11/11/2004, Art Tostaine wrote:
> I'm scanning BOL's and other handwritten forms.  Most of the scanning
> systems I've seen always use TIFF, which is a pain because it requires
> plugins for IE.
>
> I was using JPEG, and the file sizes were around 400-500K.
>
> I tried TIFF, and they are 3 times that size.
>
> Is there a reason I should use TIFF anyway?

TIFF files are lossless, meaning you will never lose any information that's 
in the image. JPEG allows compression, and is not lossless, unless you're 
using the JPEG 2000 variety, or you don't compress at all, in which case 
you're probably about the same as TIFF.

Here's my understanding of JPEG, which may be a little off, but it's pretty 
colse:

JPEG is great for images that do not have sharply defined edges (abrupt 
color changes), because the image format stores the current pixel color, 
and a number of iterations of that color, until a threshold change value is 
detected. Then it stores the new pixel color. The compression level 
determines the threshold value, and impacts both the size of the image, and 
the amount of loss. JPEG isn't great for text because text images contain 
lots of edges. It can be great for continuous tone photos though, 
especially if they are relatively low contrast. JPEG images that are 
compressed too much exhibit a ghosting effect round the edges of objects in 
the picture. My experience has been that you try to achieve a balance 
between the amount of ghosting and file size in order to arrive at the 
optimal compression level. I've used it mostly in situations where file 
size is important, and some amount of data loss is acceptable. Remember 
we're talking pixels here, not characters.

hth


Pete Hall
http://www.pbhall.us