manifold, transparency, and color
Remember that % does not mean "transparent gray". % means "do not
include in a render", and is by default shown as a transparent gray.
color() might affect the presentation, but the key thing to remember is
that a %-modified subassembly will not be included in a render.
CGAL vs Manifold should not affect rotate/pan/zoom performance, or at
least not much. Once the rendering is done, it's working with a pile of
triangles for either variation. (There might be a probably-small
difference in rotate/pan/zoom performance if the two produce radically
different numbers of triangles for the same model.)
Generally where you see rotate/pan/zoom performance issues is in
preview[*]. Performance is based on the total number of triangles, and
by interactions with difference() and intersection(). I don't know how
to characterize the performance, but complex models can get slow.
Inserting render() at key points may make rotate/pan/zoom performance
better.
[*] Which is (absent render(), resize(), minkowski(), hull())
independent of CGAL vs Manifold. Again, even when there's rendering
happening, CGAL vs Manifold won't affect rotate/pan/zoom performance.
File/Examples/Old/example024 is a good one to play with this sort of
performance on. On my system preview performance is pretty good with
the default n=3, but when you bump it to 4 render performance starts to
lag. At 5, it's something like 3-4s per frame. If you render (which
took 4m16s with Manifold!) subsequent rotate/pan/zoom is at full speed.
That's a good demonstration of the preview vs render performance
difference. For n=5, preview took 0.8s and Manifold took 4m16s, but
subsequent rotate/pan/zoom was unusable for preview and just fine for
Manifold.
interleaved:
On 2/2/2026 1:33 PM, Jordan Brown via Discuss wrote:
Remember that % does not mean "transparent gray". % means "do not
include in a render", and is by default shown as a transparent gray.
color() might affect the presentation, but the key thing to remember
is that a %-modified subassembly will not be included in a render.
Non inclusion is fine. I'm trying to get positioning right and looking
for mechanical clearances. Strictly a layout use.
The nice thing for me, is that instead of a uniform gray, I get a
transparent colored object. Much easier to find.
CGAL vs Manifold should not affect rotate/pan/zoom performance, or at
least not much. Once the rendering is done, it's working with a pile
of triangles for either variation. (There might be a probably-small
difference in rotate/pan/zoom performance if the two produce radically
different numbers of triangles for the same model.)
Generally where you see rotate/pan/zoom performance issues is in
preview[*]. Performance is based on the total number of triangles,
and by interactions with difference() and intersection(). I don't
know how to characterize the performance, but complex models can get
slow. Inserting render() at key points may make rotate/pan/zoom
performance better.
[*] Which is (absent render(), resize(), minkowski(), hull())
independent of CGAL vs Manifold. Again, even when there's
rendering happening, CGAL vs Manifold won't affect rotate/pan/zoom
performance.
File/Examples/Old/example024 is a good one to play with this sort of
performance on. On my system preview performance is pretty good with
the default n=3, but when you bump it to 4 render performance starts
to lag. At 5, it's something like 3-4s per frame. If you render
(which took 4m16s with Manifold!) subsequent rotate/pan/zoom is at
full speed.
That's a good demonstration of the preview vs render performance
difference. For n=5, preview took 0.8s and Manifold took 4m16s, but
subsequent rotate/pan/zoom was unusable for preview and just fine for
Manifold.
Thanks,
Harvey