[comp.sys.next] Pattern fills in Display Postscript

chavez@sumex-aim.stanford.edu (R. Martin Chavez) (01/12/90)

My colleagues and I would like to fill PostScript paths with patterns
defined as 8x8 bit matrices.  We need the fill operation to be extremely
fast, and so we'd like to use the halftone mechanism (which would clearly
win over replicated tiling of the region with the pattern).

We're presently using code from the Blue Book that purports to call setscreen
with frequency, angle, and spot-function parameters that accurately represent
predefined bit patterns.  Unfortunately, the results look terrible (and wrong)
on the MegaPixel display, with its 4 gray levels.  Instead of setting pixels
to either black or white, DPS seems to want to choose intermediate values,
with unacceptable results.

We know that FrameMaker has some mechanism (presently unknown to us) for 
filling regions.  Can anyone out there point us to PostScript code that does
the job?

Thanks,
R. Martin Chavez
Stanford University School of Medicine

glc@c3po.UUCP (Greg Cockroft) (01/16/90)

Checkout  FrameMaker.app/.fminit2.0/next/makerbegin.ps.

search for pattern.

The only way to do fast patterns in NeXTStep, or NeWS is
to tile on offscreen window, and then later clip & do
a window to window copy operation.

The halftoning mechanism method only works if you have
a 1 bit display, because all you get to specify is 
the order to set the dots, not the actual values.

Notice how patterns in FrameMaker documents look 
horrible when previewed, this is because the halftoning
mechanism is being used on printout.


NeWS never implemented halftoning. Not in the PostScript
sense anyway.