closest point embedding problems

Having problems with the canvas. With grid = 768 and a subgrid = (3,3,3). These should be stripes, but it all kind of collapses around x=1700.

Changing the subgrid size doesn’t help, its still about 73% of the grid in X. Changing the extension size doesn’t change anything either.

And now render seeds.

Here’s what it looks like if we change the grid size to 1024. The striping goes wrong around 1500 (out of 3072).

Instead of stripes, just make it white.

And now render the seeds.

Stripes are generated from on_surf_cmpt. That may be where things are going wrong.

Another view of grid=768, solid fill and seeds.

Zoomed in a little further.

That’s not a coincidence. It looks like we’re not generating the correct closest points. Those bands are from the grid-space.

Why is it only in X?

Huh. That cut is right in line with the Z-plane. So, it might not only be in X. But there’s no cut line in the Y-plane. Unless we get the strides because its greater than all three planes.

Increasing the g_on_surf_cnt by 1.25 didn’t change anything.

So, it’s not true that the subgrid size doesn’t matter. Changing it changes the spacing after the failure. Here’s grid=1024, with subgrid =2,3 and 4 respectively.

Here’s what it looks like with grid=1024, but viewing the same place as 768. It looks that at the same spot where the stripes start when the grid=768, now it goes to black with no stripes.

That’s weird. When the subgrid is (3,3,3) the spacing for the rings is either 4 or 5.

If I reset on_surf_fulldim, this is what happens. What I was hoping for no stripes, just black.

Nevermind, apparently with these settings that happens. Besides, on_surf_fulldim doesn’t get resized after on_surf_cmt, pts[0] does.

Okay, let’s try copying the values of on_surf_cmpt to the orig_prts list and see what happens.

Here’s a better look: what comes out of on_surf_cmpt is wrong.