Re: 16 bit scan vs. 8 bit scan/16 bit conversion
Re: 16 bit scan vs. 8 bit scan/16 bit conversion
- Subject: Re: 16 bit scan vs. 8 bit scan/16 bit conversion
- From: David Scharf <email@hidden>
- Date: Fri, 24 Sep 2004 23:17:36 -0700
David Scharf wrote:
DAVID SCHARF PHOTOGRAPHY
Scanning Electron Microscopy
Phone 323-666-8657
Fax 323-666-0449
2100 Loma Vista Place
Los Angeles, CA 90039
http://www.scharfphoto.com
http://www.soquelec.com/dscharf/images.htm
Hi James,
Here's what I think:
Dave Scharf
_________________________________________________
James Schaefer wrote:
If I scan a 4x5 transparency RGB at 2500 dpi and 8 bit
color, I get a file that is a bit over 300 MB; call it File A.
If I scan the same transparency at the same dpi, but with 16 bit color,
I get a file that is twice the size; call it File B.
If I take the 8-bit File A and convert it in Photoshop to 16 bit color,
it too is twice the size of A; call this File C.
I know that File B has more color information in it than File A has.
What I don't know is whether File C actually has more color information
than A. I suspect not -- the Second Law must come into play somewhere
along the line -- but can anyone confirm one way or t'other?
File C does not have more information.
There is a practical reason for my asking. I shoot panoramas composed
of multiple overlapping 4x5 images that I assemble by hand in PS to
produce a final seamless image. I do the assembly with files that are
300 or 360 dpi and at print-size, up to 22.5 in. high. I print the
final product on my 7600 at up to the 8-foot-wide limit of the print
driver. (I don't have a RIP.)
These panoramas can be up to 10 images wide, which means a lot of image
manipulation before I get to the final color work. When I got my G4
three years ago, its 1.5 GB of RAM seemed endless. It isn't. I'd love
to have a G5 packed with 8 GB, but I don't. Thus, as I assemble these
pans and they get larger, they get increasingly unwieldy. If I
assemble them in 8 bit format, they can approach 800 MB in their
intermediate stages, but if I try to work in 16 bit, they become simply
gargantuan and completely impractical to work with.
Why don't you scan in 16 bit do your processing, then convert to 8 bit.
After you have done this with each image, then join them together. You
must, however do the exact same processing to each image.
Then save your money for a G5 and a terabyte HD.
So here's the question: If I assemble the pans in 8 bit, then convert
the final flattened image into 16 bit for final color work, have I
gained anything? (I then return to 8 bit to print and to archive.) Or
do I retain color headroom only by scanning in 16 bit and keeping the
file that way until the bitter end?
The only thing you may gain by converting your 8 bit to 16 bit is
possibly (depending on if you leave shadow and highlight headroom) some
elbow room for processing so that you don't clip.
Actually, on second thought, you don't really gain anything here except
turning a hard clip into a soft one. There's no added image information.
Jim Schaefer
email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden
|
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden