At 19:32 4/18/01, Acer asked:
>flatbed. Last step after adjusting contrast and color balance from the
>scan to bring it back to the transparency is a combination of sharpening
>using an "unsharp mask" algorithm, resizing to 639x426 pixels (if it's
>still a 3:2 aspect ratio) using a smoothing algorithm, and then sharpening
>again using an "unsharp mask." The level of sharpening done the second
>time depends on how sharp the image is following the resizing.
is there a reason for the smoothing in the middle?
/Acer V
Yes,
There are two algorithm options . . . one is the "smoothing" which uses a
weighted average of its neighbors and the other is a "nearest neighbor"
which simply uses one of the colors. The "nearest neighbor" method can
result in a loss of color gradation during image size reduction. The
"smoothing" tends to reduce the effect of the unsharp mask as well . . .
that's why another unsharp mask is usually done afterward. It took some
experimentation with whether or not applying an unsharp mask before the
size reduction improved results. Found that it did so I do it twice. The
final unsharp mask can be as little as 250r as high as 100% . . . I
preview 100 0.000000irst and back it down until aliasing is no longer visible at
4X normal image size.
-- John
< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >
|