stevem@hpvcfs1.HP.COM (Steve Miller) (01/05/90)
I recently purchased the color scanning package Chroma-32 from Discovery Software. This package supposably lets you get color scans using your greyscale scanner by scanning the image three times with three different color filters (plastic 8.5X11 sheets). This package runs on a Macintosh on which I have an HP ScanJet+ 256 gray level scanner hooked to. The process works like this: You tape the sheet to be scanned to the top of the scanner bed. You do a pre-scan and define the area you want to scan. Then you slip each color filter under the sheet and scan it once for each filter (red, blue, and green). The result is a color scan displayed on the screen. Various controls over resolution, contrast, brightness are available and the results can be saved in a couple of different file formats. Surprizingly, the registration achieved between scans is pretty good. The Problem: The color scans don't have good color and I think I know why: The color filters suck! The red filter doesn't block all of the blue or green. The green filter doesn't block all of the red or blue light etc.. The result is muddy looking (under saturated) colors. An easy test demonstrating that the filters aren't very good is taking a picture with lots of color and placing one of the filters over it (say the red filter). In theory you should only see red color coming through: whites and reds should be a bright red and all other colors should be darker shades of red or even black. However, this is not what you see. The red filter allows fairly large amounts of green and blue to come through. Likewise the green and blue filters allow other colors to come through. I played around with some red plastic sheets that I found around the office and came up with better colors in my scans, but still not great. It's as if the people who did Chroma-32 just went down to K-Mart and picked out the first colored plastic sheets they saw. This is a $595 package and I was able to find a more effective red filter in our office supplies at work! I've tried to call the product assist line several times with no answer so I'm wondering if anyone else has had experience with color scanning from black and white scanners. Any suggestions as to where more accurate color filters can be attained would be greatly appreciated. I think you probably don't need perfect color filters if you have the properties of the filters and the scanner characterized. You can then compensate for the inaccuracies by using a system of equations for each pixel. Thank you in advance, Steven Miller ...!hplabs!hpvcfs1!stevem Vancouver Division Hewlett Packard
legrady@ug.cs.dal.ca (Tom Legrady) (01/06/90)
In article <720004@hpvcfs1.HP.COM> stevem@hpvcfs1.HP.COM (Steve Miller) writes:
I recently purchased the color scanning package Chroma-32 from Discovery
Software. This package supposably lets you get color scans using your
greyscale scanner by scanning the image three times with three different
color filters (plastic 8.5X11 sheets). This package runs on a Macintosh
on which I have an HP ScanJet+ 256 gray level scanner hooked to.
The Problem:
The color scans don't have good color and I think I know why: The color
filters suck! The red filter doesn't block all of the blue or green. The
green filter doesn't block all of the red or blue light etc.. The result
is muddy looking (under saturated) colors. An easy test demonstrating that
the filters aren't very good is taking a picture with lots of color and
placing one of the filters over it (say the red filter). In theory you
should only see red color coming through: whites and reds should be a
bright red and all other colors should be darker shades of red or even black.
However, this is not what you see. The red filter allows fairly large
amounts of green and blue to come through. Likewise the green and blue
filters allow other colors to come through.
----------------------------------------------
Roscolux produces colour filters for theatre, video and cinematic use. They
make a large assortment of colours, and the sample books include a graph
of the absorption spectrum ( very approximate! ). Check "Theatrical Equip-
ment and Supplies" in your Yellow Pages (tm).
Note that some of these may be overkill. Their darkest blue passes only
about 1%.
It may be better to use some filter that passes a wider range of colours
and do some processing on the resulting data.
Tom Legrady
legrady@ug.cs.dal.cahitchner@hydra.riacs.edu (Lew Hitchner) (01/06/90)
You are right about the color filters being the problem. Colored cellophane
is a horrible choice for color separation filters. You must use "real"
filters,
i.e., something (glass or plastic) that has a clearly defined stop band and
pass band for the spectral range needed for each of the red, green, and blue.
The filters you should buy are called wratten filters. These are not available
at most retail photo shops, but should be available from photo supply
companies,
direct from Kodak or another photo industry co., or possibly at larger printing
or copying businesses. Wratten filters are named by a number that seems
to have
nothing to do with the wavelengths of the filter. The filters numbers
are integers
in the range 0 to 100 (or therabouts), some with an "a" appended, such as 23a.
I don't know the optimal filter numbers for r,g,b separation. You'll
have to ask
for a catalog or ask an expert. However, I do know that filter numbers
22 thru 29
pass red (approx. 560-700 nanometers) and block the rest, and that
filter numbers
57a to 59 pass green (approx. 500-550 nm.) and stop the rest. Sorry, I
don't know
the filter numbers for blue filters. But, I hope this helps get you
started in the
right direction.
Lew Hitchnerlittauer@uts.amdahl.com (Tom Littauer) (01/06/90)
In article <1990Jan5.190700.28712@ug.cs.dal.ca> legrady@ug.cs.dal.ca.UUCP (Tom Legrady) writes: > >Roscolux produces colour filters for theatre, video and cinematic use. They >make a large assortment of colours, and the sample books include a graph >of the absorption spectrum ( very approximate! ). Check "Theatrical Equip- >ment and Supplies" in your Yellow Pages (tm). > >Note that some of these may be overkill. Their darkest blue passes only >about 1%. > >It may be better to use some filter that passes a wider range of colours >and do some processing on the resulting data. I have actually done this (although with B/W cameras) and it works just fine. Also note the the sheets are ~ 20" X 24" and ~ $5 each. Lee makes equivalent stuff for the UK readers. -- UUCP: littauer@amdahl.amdahl.com or: {sun,decwrl,hplabs,pyramid,ames,uunet}!amdahl!littauer DDD: (408) 737-5056 USPS: Amdahl Corp. M/S 278, 1250 E. Arques Av, Sunnyvale, CA 94086 I'll tell you when I'm giving you the party line. The rest of the time it's my very own ravings (accept no substitutes).
cygnus@vax1.acs.udel.EDU (Marc Cygnus) (01/07/90)
In article <1941@hydra.riacs.edu> hitchner@same (Lew Hitchner) writes: > <etc. deleted> > The filters you should buy are called wratten filters. These are not > available at most retail photo shops, but should be available from photo > supply companies, direct from Kodak or another photo industry co., or > possibly at larger printing or copying businesses. Wratten filters are... > <etc. deleted> > > Lew Hitchner Quite correct, Lew. Perhaps the following information would be of use to those of you wishing to investigate colour digitisation: The following filters are excellent for any work involving colour imaging and/or projection. They are classified as 'colour separation filters': Kodak Wratten gelatine filters (you can get glass, but I chose the gels) No. 29 (red) #149 5621 \ No. 61 (green) #149 5894 +-- Kodak catalogue #s for 75mm x 75mm No. 47 (blue) #149 5787 / I have had consistently good results using these gels. Also, Voss Pro- fessional makes a nifty gelatine holder that fits any lens assembly from 50mm on down (and has two adjustable barn doors, too :-). Two rubber - padded leaf springs part under finger pressure, then you place over the lens and let go... easy to insert/remove gels (there's a plate that slides into the assembly; push it out from the bottom and it brings the gel out with no fingerprints). Aaaaaaaanyway... Now, if I could only get my hands on a good CHEAP semi-hiresolution CCD element camera with RGB (or composite) output... -marcus- -- ----------------------------------------------------------------------------- "Opinions expressed above are not necessarily those of anyone in particular." `...but do YOU own a | ARPA: cygnus@vax1.acs.udel.edu homemade 6ft Tesla?' | UUCP: {yourpick}!cfg!udel!udccvax1!cygnus
cygnus@vax1.acs.udel.EDU (Marc Cygnus) (01/07/90)
I just noticed the subject line :-)
I don't recall if this has been said before, but it's probably a good
idea if you not waste your time trying to force a BW scanner to do
colour. Besides the fact that the lightsource is probably not a good
wideband source, and alignment is _ridiculous_ (unless you tape your
source image to the underside of the scanner's bedcover... yuck), you
should keep in mind that the light will have to pass through the filter
_twice_.
When doing camera work with colour separation filters, it's suggested
that 2 f-stops of compensation be used with the filter (and that's when
the light only passes through once).
-marcus-
--
-----------------------------------------------------------------------------
"Opinions expressed above are not necessarily those of anyone in particular."
`...but do YOU own a | ARPA: cygnus@vax1.acs.udel.edu
homemade 6ft Tesla?' | UUCP: {yourpick}!cfg!udel!udccvax1!cygnusmcdonald@aries.scs.uiuc.edu (Doug McDonald) (01/07/90)
In article <5443@udccvax1.acs.udel.EDU> cygnus@vax1.acs.udel.EDU (Marc Cygnus) writes: > >I just noticed the subject line :-) > >I don't recall if this has been said before, but it's probably a good >idea if you not waste your time trying to force a BW scanner to do >colour. Besides the fact that the lightsource is probably not a good >wideband source, and alignment is _ridiculous_ (unless you tape your >source image to the underside of the scanner's bedcover... yuck), you >should keep in mind that the light will have to pass through the filter >_twice_. > >When doing camera work with colour separation filters, it's suggested >that 2 f-stops of compensation be used with the filter (and that's when >the light only passes through once). > I actually tried scanning my MacBeth Color Checker on an Apple scanner. Results: The color response of the scanner is apparently designed to give "correct" gray scale values of color material (i.e. blue darker than red darker than green), because it does do so, quite well. It has enough response in the blue and red to be useful. But of course it is only a 16 level (4 bit) scanner, so would be useful mostly for scanning colored line art. You probably could get some sort of tri-grayscale image by varying the contrast and brighness, but it wouldn't look very pretty. The Wratten filters mentioned previously are certainly NOT right for using in such a scanner. (In fact, they are probably not right for use in front of a camera, for photographing general subjects. They are TOO narrow. They are for PRINTING use, or for doing color separations from three or four color printing or color photos.) For that you need bigger filters, and ones with higher transmission, due to the double pass nature. The theatrical gels are indeed available in suitable colors, and are available in big sizes. With the proper filters you should be able to get excellent scans of colored line art. Of course, with a Mac you are going to have fun figuring out what to DO with color separations! Doug McDonald
cygnus@vax1.acs.udel.EDU (Marc Cygnus) (01/08/90)
In article <1990Jan7.155917.29080@ux1.cso.uiuc.edu> mcdonald@aries.scs.uiuc.edu (Doug McDonald) writes: <regarding my posting about Kodak Wrattens> > >The Wratten filters mentioned previously are certainly NOT right >for using in such a scanner. (In fact, they are probably not right >for use in front of a camera, for photographing general subjects. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ >They are TOO narrow. They are for PRINTING use, or for doing color >separations from three or four color printing or color photos.) >For that you need bigger filters, and ones with higher transmission, >due to the double pass nature. The theatrical gels are indeed >available in suitable colors, and are available in big sizes. > >Doug McDonald Don't use the mentioned filters with scanners - you're right, as I noted in my second posting (I wasn't thinking about scanning when I posted the first article - I was thinking about imaging from a video source). But these filters _are_ correct for use "in front of a camera", or rather in front of a digitiser. I'm speaking from experience here. I don't usually invest large amounts of money in things that don't work. Let me reiterate that the filters I named *are excellent for digitising colour* if you're using a three-pass approach and a videocamera that behaves properly (either a colour cam or a b/w cam that's sensitive more or less equally to the primaries). -marcus- -- ----------------------------------------------------------------------------- "Opinions expressed above are not necessarily those of anyone in particular." `...but do YOU own a | ARPA: cygnus@vax1.acs.udel.edu homemade 6ft Tesla?' | UUCP: {yourpick}!cfg!udel!udccvax1!cygnus
halam2@umn-d-ub.D.UMN.EDU (Haseen DSP Alam) (01/13/90)
I actually have a question for you. Does the package include any software that superimposes all three filtered scans? Once you scan it, in what format do you save all the different scans before superimposing? I was thinking of doing this manually with report covers as my filters and some paint/draw/image processing program, that I have access to in our graphics lab. Thanks. Haseen.
stevem@hpvcfs1.HP.COM (Steve Miller) (01/16/90)
>I actually have a question for you. Does the package include any software >that superimposes all three filtered scans? ... >Haseen. Yes, the one nice thing about chroma-32 is that it will integrate all three color scans together and display the results. You can then play with the contrast and brightness settings. Then you save it as TIF-RGB or as an 8,16, or 32 bit PICT. Steven Miller ...hplabs!hpvcfs1!stevem Vancouver Division Hewlett Packard