Here is the command: gsc -dBATCH -dNOPAUSE -dUseCropBox -sDEVICE=tiffsep -sOutputFile=/tmp/32058-buildtiffs_3 -r300 -c "<< /SeparationColorNames [ /Cyan /Magenta /Yellow /Black (Orange) (Green) (Blue) (DIE) (Varnish) ] /SeparationOrder [ (Varnish) ] >> setpagedevice" -f /storage/archive/20196PCV200.pdf This specifies SeparationColorNames for every separation in the file, and then specifies a lesser number of separations in SeparationOrder in order to selectively suppress the rendering of some channels. In this case, it should render only the Separation named "Varnish". Instead it renders a single blank image and does not output any files named like "s0". The selective suppression of colorants using SeparationOrder does seem to work in some other cases. See comment at gdevtsep.c:1025 for documentation on the behavior.
Created attachment 8550 [details] file referenced in bug
My initial comments were a little off, I do get 32058-buildtiffs_3.s4.tif but it's blank.
My initial comments were a little off, I do get 32058-buildtiffs_3.s4.tif but it's blank. I have tried with /SeparationOrder [ (Magenta) ] and such, and the rendering is completely incorrect, it's a blocky mess. So it looks like the true bug is that if you only specify one separation in SeparationOrder then it's broken.
As a temporary work-around, it seems to work OK if you specify at least two colorants, even if one is blank (as is /Blue in the example file). Since the tiffsep (and psdcmyk) devices and internal handling of spot colors is changing a lot with Michael Vrhel's switch to use planar rendering of spot colorants and to no longer use the compressed color encoding, we will look at this bug after that massive change is committed -- expected RSN.
Fixed with http://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=7b81312d205a2f9b89f40da4b4f6b67bcacd8ef1 Two things to note: 1) Since the tiffsep device now uses a planar memory buffer, it is not limited to 8 colorants. So there is no need to run individual seps any longer. 2) If you do use /SeparationOrder, the composite tif output will only show the combination of the colorants listed in /SeparationOrder.
Michael, Regarding your number 1 point, this is incorrect since I do need to do individual separations for my use case. We have files with a Varnish layer that is sometimes set with a tint and set to overprint the entire image. This is because Varnish is output onto a plate as a separation but is clear in real life. Artists sometimes give it a non-clear tint so they can see it and manipulate it in prepress programs more easily and generators don't always override this tint transform when writing the PDF out. So I do one run with Varnish left out of the SeparationOrder to get a full color composite, unmolested by potential varnish problems, then I do a second run with everything turned off but Varnish in order to get the grayscale Varnish separation. This will not change under your planar patch, unless additional functionality is added that would allow the user to override the tint transform for certain separations, while still outputting the gray version of the separation unmodified. Thank you for your work anyway, it is still a valuable addition, but it does not completely negate the use of multiple passes without the additional feature I mentioned above.
Gigs, I understand then why if someone fails to give you a PDF that has a proper tint transform for the varnish layer then you will have an issue. This issue should be addressed during PDF creation as Acrobat Reader will show the same problem. I don't see us adding any new command options to keep out a separation from the composite while outputting the separation. Michael
They use separation preview in Acrobat (not reader) and toggle off the display of the Varnish. Prepress people usually use full Acrobat and not Reader. It's OK if you don't want to add the feature to suppress colorants in one pass, I just wanted you to be aware that we will be continuing to use multiple passes with SeparationOrder to do so.