Message-Id: <mailto:199409131530.KAA23173@library.wustl.edu> Date: Tue, 13 Sep 1994 14:08:23 +1000 From: Brett Powley <mailto:brett@PROCESS.OZ.AU> Subject: Re: 8-bit vs. 24-bit color To: Multiple recipients of list IMAGELIB <mailto:IMAGELIB@ARIZVM1.BITNET>
mailto:"mailto:Walter_Gilbert@UMAIL.UMD.EDU--Maryland" <WALT@UMDD.BITNET> writes:>8-bit vs. 24-bit color.
>I have a garden-variety Dell comuter with a 17" monitor and graphics card
>capable of displaying 24-bit color. I delight in displaying an 8-bit color
>and a 24-bit color version of the same image side by side and having the
>viewer tell the difference. (I did this at a conference and most people
>guessed it wrong.)
>Of course there is a difference, and if you get up close and look at areas
>of slowly changing color (sky is a good example) you can see the difference.
>But, for everyday use at normal viewing distances, the factor of 3 savings in
>size is worth while.
Ummmm.....no.
It, of course, depends on the image. If your image is of blue sky with a couple of pink flamingoes, then 256 colours might be enough to give you all the blues and pinks you need. If you image has a lot of different colours (and shades of colour), 256 normally isn't enough. How good your 8-bit image looks also depends on whether it's dithered or not (giving the illusion of more colours).
There are other factors too. Humans are not very good at distinguishing shades of blue, but much better at distinguishing shades of green. So, displaying a mostly blue image in 8-bits might not _look_ much worse, but displaying an image with lots of greens would be much more noticeable.
In a real application, using remotely sensed data, 8-bit RGB composites often simply don't have enough definition (not enough different colours) to distinguish between features, even without "getting up close".
Not that I would want to disillusion you about something you "delight in" so much.... :)
Brett Powley Process Software Wollongong, NSW, Australia