Try this:
collink -v -G AppleRGB.icc JapanColor2002Newspaper.icc apple_to_jNP_photo.icc
collink -v -f AppleRGB.icc JapanColor2002Newspaper.icc apple_to_jNP_neutrals.icc
control.txt:
Image_RGB apple_to_jNP_photo.icc 0 1 0
Graphic_RGB apple_to_jNP_neutrals.icc 0 1 0
Text_RGB apple_to_jNP_neutrals.icc 0 1 0
and
gswin32c -q -sDEVICE=pdfwrite -o out.pdf -sColorConversionStrategy=CMYK -sSourceObjectICC=control.txt in.pdf
Then the DeviceRGB
in source PDF is converted to DeviceCMYK
, and RGB 0/0/0
becomes (as I'm checking now) the DeviceGray
0
, which should be OK (and all other neutral RGB shades are mapped to true grayscale, too).
The reason we are using different DL-profiles for different objects, is because, though saturated colors (far from neutrals) will be converted to the same CMYK through both profiles, nevertheless you probably don't want color suddenly switch to 0/0/0/n
in continuous tone photographs, if color happens to be near neutral -- it'll look terrible on the press.
If your "images" are e.g. rasterized graphics (diagrams, etc.) with 0/0/0
RGB, then you can consider using apple_to_jNP_neutrals.icc
for these images too.
If your page has a mix of both real images and rasterized graphics (text) - bad luck, you'll have to compromise.
The reason we use -G
instead of fast and simple Simple Mode
, is because -f
(for second profile) implies the "Gamut Mapping Mode using inverse outprofile A2B", and we want 2 profiles to produce the result (for saturated colors) as close to each other as possible.
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…