At 05:18 PM 11/29/99 +0000, Brian Windrim wrote:
>Hi All,
>
>I have some comments and questions regarding Garth's suggestions:
>
>>> *Input* DPI, on the other hand, should be as high as possible (you can
>always do dimensional reduction of the image later).
>
>I assume you mean the natural scanning resolution of the device.
>Overscanning (interpolation) wouldn't get you anything if you're reducing
>later.
Correct. My apologies -- I always assume that everyone scans at the highest
purely mechanical resolution available. Natural bias on my part.
>>> This gives you more raw information to work with for any other operations
>such as image cropping, sharpening, unsharp masking, etc.
>>> Then reduce to target screen size as a last step.
>
>What I've read on the subject strongly suggests that size reduction should
>be the second-to-last step, with either sharpening or unsharp masking being
>the last.
I have tried it both ways. My comments are meant to be somewhat generalized,
not necessarily "gospel" (not that anybody in their right mind ever listens to
me anyways... ;-) I'm unable to detect a real difference in some images,
whereas with others, yes, your sources are correct. Still others seem to work
best using dimensional reduction as the final step. I assume the differences
in every case are attributable to the statistical distribution of the image
information in the file -- I've never been curious enough (or had enough time,
which for me is much the same thing) to investigate further.
>>> When reducing your image dimensions, always use bi-cubic resampling (it's
>the best quality overall).
>
>I sometimes use the Photoshop Crop Tool to do resizing, as this allows
>resizing and rotation in a single pass. Unlike the Image Size dialog box,
>the Crop Tool settings box doesn't have a choice of resampling methods. Does
>it always use bi-cubic?
No idea. RTFM, I suppose.
>>> Remember that, if people are going to be viewing your work in a Web
>browser, you have to subtract some pixels from each dimension (you'll have
>to figure out what you're comfortable with), since the browser's window is
>not the same aspect ratio as any of the standard screen dimensions.
>
>Surely it's the *size* of the usable part of the browser's window (width and
>height, in pixels) that matters, not the aspect ratio?
Depending on how you crop or initially scan, it could be either that's your
limiting factor. Everyone thinks a little differently about how to reach the
goal, and proceeds along those lines. If the aspect ratio of a browser screen
(with your monitor at 640 x 480, for example) is different than your monitor,
then by definition you're losing pixels in at least one dimension. That's
really all I was implying. It's complicated a bit by the fact that you never
really know anything about the final output device for your images, or the
browser's settings -- there could be more control bars showing, or less, for
example, and this changes the aspect ratio as well. Ultimately, I just crop
and downsample to what I *want*, and if your browser can't deal with it
properly, oh well, that's the beauty of the Web.
>I'm new to this stuff so apologies if I seem a bit slow.
As the Aussies on this list might reply, "No worries, mate."
To everybody: Again, I sayeth unto you, go to
www.scantips.com
and read up, for yea and verily, that Website is a mighty work, and righteous
before the Almighty Algorithm.
Garth
"A bad day doing photography is better
than a good day doing just about
anything else."
The Unofficial Olympus Web Photo Gallery at:
http://www.taiga.ca/~gallery/
< 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 >
|