PDA

View Full Version : 11.6.3 Win64 stats (or box tool) acting odd...

jwiede
02-02-2015, 07:45 PM
Can anyone attempt following with LWM 11.6.3 to confirm (and I'd be interested to know if still in 2015)?

1. Create->Box (press 'n' to activate numeric panel)
2. Set to: {5m x 5m x 5m} @ {0,0,0} with 20 segments in X, Y, Z
3. Drop tool so nothing (everything) is selected
4. Shift viewport to wireframe mode
5. Select "Point" mode
6. Open stats panel

Here with LWM 11.6.3 Win64, repeating test from clean/restarted multiple times, LW reliably shows a total of 2402 (?!?!) points in the stats panel.

Would love to know if others see same with LWM 11.6.3 on Win64? How about with LWM 2015.1 Win64?

spherical
02-02-2015, 08:29 PM
In 11.6.2 (never bothered with 11.6.3) I get: 8 3-poly and 2394 4-poly = 2402 points on 2400 polys. Same in 2015.1.

Each panel can be regarded as having 361 points unique only to it. 19 * 19 = 361
The 12 edges that are made up of points common to two panels or three panels total 236 points.

361 * 6 = 2,166
2,166 + 236 = 2,402

jwiede
02-02-2015, 09:05 PM
In 11.6.2 (never bothered with 11.6.3) I get: 8 3-poly and 2394 4-poly = 2402 points on 2400 polys. Same in 2015.1.

Each panel can be regarded as having 361 points unique only to it. 19 * 19 = 361
The 12 edges that are made up of points common to two panels or three panels total 236 points.

361 * 6 = 2,166
2,166 + 236 = 2,402

Umm...

A. You were supposed to enter 20 "segments" into X, Y, Z. How are you using 19 points to create 20 segments inclusive? I count 21 points to create 20 connected, inclusive segments -- First segment requires 2 points, each subsequent attached segment requires 1 additional, so 2+19 = 21, thus yielding a 21^3 matrix. Even if LW cheats and only creates 19 segments (using 20 points), it should still yield a 20^3 matrix.

B. Where do you envision points being shared when iterating through the matrix as "panels"/slices along an axis? Can you plz describe your counting algo?

More testing...

2x2x2 segments (= 3x3x3 points = 27 total / counted) yet stats shows 26 total points.
3x3x3 segments (= 4x4x4 points = 64 total / counted) yet stats shows 56 total points.

Time to file, from the sounds of it, but kinda scary. Wonder how far back it does that?

spherical
02-02-2015, 09:39 PM
Did exactly as you instructed. 20 segments per axis. I'm working up a second way to describe the geometry. BRB.

OK, here's another way to analyze it.

Select a 20 x 20 grid of points on one side face:
126835 This totals 400 points: 20 * 20 = 400
Do this on the remaining three sides:
126836 this totals 1,600 points. 400 * 4 = 1,600
The top only has a 19 x 19 grid of unselected points:
126837 19 * 19 = 361
The bottom has zero points selected.
126838 21 * 21 = 441

1,600 + 361 + 441 = 2,402

OR you could select all of the points on all four sides. This is a 20 * 21 matrix = 420 points.
420 * 4 = 1,680

This leaves both the top and bottom with a 19 x 19 grid of unselected points. 361 * 2 = 722

1,680 + 722 = 2,402

jwiede
02-02-2015, 10:34 PM
Ugh, nvm, I think I'm having some kind of driver issue. I was seeing what looked to be interior points, as if it were constructing the full matrix (incl. interior points) instead of just a shell, but full reset seems to have resolved the issue. Very, very odd, never seen it do that before. Mea culpa.

spherical
02-02-2015, 10:57 PM
Which would be meaningful if the points were only created on the outer surfaces, but they're not.

And what inner surface is there?

OK, consider this. There are, in a cube, always two more points than polys; no matter how many regular divisions you have.

Try it yourself on a simple cube, having zero divisions (1 face, 1 poly per side):

6 polys
8 points

It just propagates outward from there.

spherical
02-02-2015, 11:15 PM
Ugh, nvm, I think I'm having some kind of driver issue. I was seeing what looked to be interior points, as if it were constructing the full matrix (incl. interior points) instead of just a shell, but full reset seems to have resolved the issue. Very, very odd, never seen it do that before.

That would have increased the point count... by a lot; way more than two. You can see in my screen shots that the back-facing surface selected points are visible. That doesn't mean that they are "extra". Seeing those "interior points" doesn't translate to there being more of them. They are the same points, only on the opposite side of the view.

jwiede
02-03-2015, 09:25 PM
Seeing those "interior points" doesn't translate to there being more of them. They are the same points, only on the opposite side of the view.

Nope, not talking about the points on the opposite side of the shell, by "interior" I do indeed mean interior, as if it were generating a full matrix. That's why I talked about expecting 21^3 points total instead of the 2402 points shown in the stats window, hence the thread. It was momentarily plausible to me (in my defense, it had been a long day) that it normally generated interior points, and following that premise, a 20seg x 20seg x 20seg box should generate a 21^3 matrix of points (instead of the actual shell produced which lacks interior points).

In any case, it was a total brain poot on my part, and I apologize for wasting your time.

spherical
02-03-2015, 11:25 PM
No problem. It was a good exercise, cuz' it seemed weird to me at first blush, too. Then I analyzed it a bit more in the back of my mind while doing other stuff and then took a closer look to prove to myself, one way or the other, what was really what. As it says in my Bio, any day I learn something new is a good day. So is any day I wake up. :D