On Thu, Aug 17, 2023 at 11:56:36PM +0000, Jordan Brown wrote:

STL is just one output format that happens to not-support color.
There are other formats that do.

That is the essential issue, I think.

I would think that a "preview-render" would be enough to find
real-world coordinates for things.

Suppose we tell OpenGL offscreen 1024x1024 framebuffer orthogonal
projection from TOP, then we render everything, and the "pin" we
render just like the pink see-through objects-tagged-with-# : On top
of everything. Now we can look at the framebuffer, and find the one
pixel that has become pink. Oh.... we can tell openGL to not-render
all normal objects (100% transparent). If you use say a 1024x1024
buffer, you can "zoom in" by a factor of 512 to render an originally
2x2 pixel area again in a 1024x1024 frame buffer and get 9 more bits
of accuracy. The new framebuffer should be centered on the center of
the pixel that was orginally pink. It might be possible to zoom by up
to 1024, but zooming by "only" 512 is for certain going to be useful
to get the "center on the center of the pixel" correct while debugging
the code.

Do this for one or two(*) more axes and you've got all three coordinates,
using the preview-engine.

There might be tools (API) inside of openGL to not have to search the
whole 1Mpixel framebuffer. People want to add 2D tags to objects in
the rendered 3D space. Think of the place names or pins in a 3D render
in google earth.

This might be the hint:
https://stackoverflow.com/questions/3792481/how-to-get-screen-coordinates-from-a-3d-point-opengl

Roger. 

(*) If you do two more views, you'll get all coordinates twice. Might
be good for verification or a little extra accuracy.

--
** R.E.Wolff@BitWizard.nl ** https://www.BitWizard.nl/ ** +31-15-2049110 **
**    Delftechpark 11 2628 XJ  Delft, The Netherlands.  KVK: 27239233    **
f equals m times a. When your f is steady, and your m is going down
your a is going up.  -- Chris Hadfield about flying up the space shuttle.

Yes, I was wondering if one couldn't hack the existing GUI to do something
based on color. As one can display different objects in different colors in
the UI

On Fri, Aug 18, 2023 at 10:28 AM Rogier Wolff R.E.Wolff@bitwizard.nl
wrote:

[ Let's try to fix the subject...]

On 8/18/2023 1:28 AM, Rogier Wolff wrote:

Maybe.  Such a scheme would be outside my knowledge, without a big
learning curve.

But if one was satisfied with a pin with its coordinates reported in the
console, and was willing to live with the various restrictions, it seems
like the "carry the transforms down through the execution context"
scheme would be straightforward enough.  It's just hard to see how to
turn it into anything like a decent measurement user interface.

One could imagine making the parent transformation matrix available via
a function or $ variable.  Reset it to identity inside an extrusion,
minkowski second arg, et cetera.  But again, the "black hole" nature of
modules, that there's no way to get information out of one, would make
the value very limited.

To use join_prism() from BOSL2 you would need to join the side section onto
a cylindrical portion and move the taper somewhere else.  That is,
join_prism() cannot have the base be a cone, only a true prism of some kind
(or a plane or sphere).

On Wed, Aug 16, 2023 at 4:52 PM Revar Desmera revarbat@gmail.com wrote: