LDraw.org Discussion Forums

Full Version: [LDPartEditor] First Alpha Version News (until 2014-09-19)
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2 3 4 5 6 7 8 9 10 11
Two things. First, it appears that the spec doesn't specify the units for the angle. However, since it is designed to be easy to use for part authors, I feel that it should be degrees.

Secondly, Phil's suggestion of what the sphere should look like may be correct, but I'm not sure. I'm not having any luck wrapping my head around the description for the spherical projection.
[Image: sphere_problem.png]
A little defect is located at the polar caps (theta: ? phi: .5pi or -.5pi).
edit: This defect can't be solved.
'Little thing' Smile

For the outer top and bottom ones all horizontal angles are correct, so the only way to get a perfect textured sphere you have to take into account the other points of the triangle/quad you're working on (creating multiple vertices for the same top/bottom xyz).

The same problem with -180/180 coordinates. Don't you have this problem at the backside of the sphere/cylinder?

On the other hand I'm wondering if the spec was ment to be used on enclosed objects, it being projection orientated and all.

I'm also curious... how does it look when you map a texture on something when the outer edges of the texture don't align with ldraw points? e.g. a -60..60 deg on a 16 facet cylinder (texture should starts and end in the ~middle of a segment?)
Roland Melkert Wrote:On the other hand I'm wondering if the spec was ment to be used on enclosed objects, it being projection orientated and all.

One of the few places off the top of my head that I can see a use for spherical mapping would be to put texture onto a LEGO ball (like the Hero Factory mixed color shooter balls).

Roland Melkert Wrote:I'm also curious... how does it look when you map a texture on something when the outer edges of the texture don't align with ldraw points? e.g. a -60..60 deg on a 16 facet cylinder (texture should starts and end in the ~middle of a segment?)

The angle that the part author puts into the file is expected to match the angle covered by the textured geometry, so your question seems moot. Remember, textures are always applied to whole facets in the LDraw !TEXMAP extension. It's up to the author to specify correct !TEXMAP parameters for the geometry the texture is being applied to.
Travis Cobbs Wrote:One of the few places off the top of my head that I can see a use for spherical mapping would be to put texture onto a LEGO ball (like the Hero Factory mixed color shooter balls).

Maybe use two texture commands (on two half sphere's) for that (project one on to the top and one on to the bottom) to avoid the whole multiple correct uv coordinates for an LDraw point problem.


Travis Cobbs Wrote:The angle that the part author puts into the file is expected to match the angle covered by the textured geometry, so your question seems moot.

so it's up to the authors to pad the texture sources, I can live with that Smile
I took some backup from my university to solve this problem (a physicist and a professor in mathematics). It turns out, that imperfect polar caps are the effect of the mathematical definition of the perfect sphere and the limited resolution of our LDraw spheres. Ray tracers do not have this problem. The image on the wikipedia article, which was mentioned by Roland, was generated by a raytracer. An alternative solution would be to cut off the polar caps and use a planar projection only on the (nearly planar) polar caps.

The problem is not solvable mathematically under this circumstances and I do not intend to find a workaround which might expand the current TEXMAP spec.

This night, I will upload a video and more screenshots with different angle configurations to better visualise the capabilities of the TEXMAP meta command.
Would you be willing to share your proof? I'm curious to read it.

Tim
Hi Nils,

Would you mind posting the test texture (the checkerboard texture) you used? I would like to try an experiment with it to see if I can get spherical UV mapping without a polar artifact.

Thanks!
Ben
Is there a set of (unofficial) parts to use for testing? Preferable ones using all three kinds of mapping and some sort of nesting etc in order to test all texture usages in one go.

This thread made me want to give it a go to in LDCad too Smile, most work will go into setting up the support stuff though (and keeping normal rendering fast), but I'm hoping to get something going this weekend.
Pages: 1 2 3 4 5 6 7 8 9 10 11