The answer to the mail follows, but to reformulate my question slightly:

How can I make shading inside metapost(play) to produce exactly what's in the attached picture (smooth shades). Or more generally: how can I "hardcode" some PS/PDF code in .mp, not (yet) supported by metapost(play)?

(PDF supports even 3D shapes (already shaded properly) since version 1.5 or 1.6, but I suppose that's another story anyway.)

Here's the corresponding PostScript code:

%!PS-Adobe-3.0 Resource-ProcSet
%%Title: Hsbwheel.ps
%%Creator: Henry McGilton, Adobe Systems Incorporated (modified!)
%%EndComments

200 dup scale

/X3  1.0 3 div def
/X6  2.0 3 div def
/Y3  1.0 3 div def
/Y6  2.0 3 div def

/middlecolor  [ 1    1   1 ] def
/topcolor     [ 1    0   0 ] def
/rightcolor   [ 0.5  0   1 ] def

newpath
<<
    /ShadingType 6
    /ColorSpace /DeviceRGB
    /DataSource [
        0
          0.0 0.0    X3 0.0     X6  0.0   1.0 0.0
          1.0 .552   .552 1.0   0.0 1.0
          0.0 Y6     0.0 Y3     0.0 0.0   0.0 0.0    0.0 0.0
          middlecolor aload pop    rightcolor  aload pop
          topcolor   aload pop     middlecolor aload pop
    ]
>> shfill

showpage

************************************************************************

Laurence Finston wrote:

Mojca Miklavec wrote:


There's a wonderful tutorial on drawing 3D graphics in PostScript on http://www.math.ubc.ca/~cass/graphics/manual/, specifically
http://www.math.ubc.ca/~cass/graphics/manual/code/doughnut.eps as an example.


Thanks for the reference, it looks very interesting.


In metapost, "perfect" spheres can be drawn, like on http://www.gnu.org/software/3dldf/cdsmpls.html. But how can I add > colour?


The spheres in that example are really just circles.  The "globe font" was
done using Metafont, which doesn't support color at all.  The sphere graphic
at the beginning of that file was done using MetaPost, so the circles could be
drawn in color, but I know this isn't what you mean.

Well - the main question is how to add the proper (smooth shading) color once I calculate the grid points in one way or another.

The glyphs from the globe font (which I plan to improve) could be colored,
e.g., by using the macros defined in the `colordvi.tex' file supplied with
`dvips'.  Of course, a given glyph could only be printed or displayed in a
single color.

GNU 3DLDF does not yet possess a `sphere' type, but I plan to add one in the
not-too-distant future.  I also plan to add shading, but that is in the more
distant future.  The shading methods I'm (vaguely) familiar with operate on
raster data and I haven't quite decided how to handle rasterization.  I'm
leaning toward using the PNG format rather than PostScript for this.  I
believe PostScript can be used for both vector and raster data, but my memory
of the details is hazy.

Well, PostScript can be used for raster images, but that doesn't bring anything. The examples on Cass' page are actually done using *vector* graphics (colors are really calculated only in grid points, but then the color is interpolated by postscript interpreter): the perfect sphere coloring is thus really limited to "floating point accuracy" and printer capabilities.

The spheres drawn directly in PostScript as in the examples above still have straight lines (well, I agree that with some mathematics it would be possible to draw them with proper splines as well).



If you mean my spheres, that's an artefact of the PNG graphics.  The
PostScript versions look nicer.  I believe that on commonly available devices,
 curves are ultimately drawn as short straight-line segments, though.  In the
code in that file, the circles have 64 points, and I thought the results were
good enough.  Raising it to some value x such that x % 4 == 0 and x > 64 might
improve them, I haven't tried it.  32 wasn't enough.

No, I wasn't talking about the PNGs. I was talking about the PostScript pictures. http://www.math.ubc.ca/~cass/graphics/manual/code/doughnut.eps has a really *perfect* shading, but its edges are drawn using straight segments instead of drawing them with splines which damages slightly the whole impression of its perfectness. Some additional calculation for spline drawing would help, but I still find it easier to calculate the curves in metapost than in PostScript.

Those "straight-line segments" on printers are just as straight as they are on any laser print-out of texts if I'm not wrong - i.e. not worth considering.

Thank you,
        Mojca


Attachment: hsbwheel.pdf
Description: Adobe PDF document

_______________________________________________
ntg-context mailing list
ntg-context@ntg.nl
http://www.ntg.nl/mailman/listinfo/ntg-context

Reply via email to