View Full Version : What is with Rounder?

02-26-2013, 03:29 PM
Rounder is so flakey. I'm just trying to round the two corners with he selected points, and I get the error "A selected point is not inside he selected surface". What the hell is it talking about, I have two points selected on a rectangle. What is so special about that?111975

02-26-2013, 03:31 PM
Try to do a merge (shortcut "m") and a unify polygons (shortcut Shift+I) just in case.

02-26-2013, 03:36 PM
I wish the new 11.5 chamfer tool did rounded corners. I'd put that 's$*#' on everything!! :)

But yes. Rounder is flakey. I tend to get crashes more than results.

02-26-2013, 06:40 PM
When you used it the previous time, you used the 'OpenGL poly' setting right? That feature is a little bugged, because it doesn't fix the points to the mesh. By using it it's made the 2 ngons disconnected. If you go into edge mode, you'll see you have some 1 poly edges. hit 'w' for statistics, and hit the + next to 1 poly edges, my guess is you'll have 12 of them :) The simplest 'fix' for you is to delete the 2 ngons, lasso select each open end, in edge mode, and hit 'edges 2 poly' (hidden in the create>polygon>more>edges to poly) I use this tool so much it lives on my custom menu :)

The easier option is to just not use that 'opengl poly' option, and pre-edit the mesh to have that built in, using the knife tool, or multishift first etc. More trouble than it's worth! :)

02-26-2013, 09:29 PM
More trouble than it's worth! :)
That phrase applies pretty well to Rounder itself in many scenarios. If Rounder were a horse, Ikea would have served it up long ago.

Bad Rounder! Bad! No!

02-27-2013, 03:06 AM
If Chamfer did multi-segment chamfer's I'd not be using Rounder at all. sadly, it doesn't... Maybe LW 12....

02-27-2013, 04:22 AM
Try applying Rounder to points...

02-27-2013, 10:15 AM
Rounder is useless as dead!
It should not even be in the lw tool set, doesn't worth it :P

Waves of light
02-27-2013, 10:23 AM
I think I stated in a previous thread that Rounder is very flaky. It needs clean geometry (no 2 point polys, etc.) and you can only do it whilst viewing one layer, so no background layers visible.... and even then it can cause issues.

Chamfer and Mass Round (LWCAD) are the way to go. Also, check out Jetto Bevel and Bevel++.

02-27-2013, 02:35 PM
Rounder is useless as dead!
It should not even be in the lw tool set, doesn't worth it :P

I disagree, what other LW tool does this... select point loop, expand points to quads? You could just select a face loop, bevel and shift but it isn't the same.

02-27-2013, 05:03 PM
I use rounder all the time. and yes...no 2pt polys, broken edges, etc... (though I often have background laters visible).

My only complaint is that it breaks UV's.

03-17-2013, 05:32 PM
I was going to post a question on rounder, but I finally got it to work, but so not to waste a screen grab:

I Booleaned 3 slots in a box, and was trying to use Rounder to ....fillet?.... chamfer?... the edges.

All sorts of issues came up, especially the "A point is not...b.effinb." error.

Long story short, I finally gave up on EDGES and polys and several permutations of both and just selected the outside points of the slots, and WALLAGH! Rounder worked like a very annoying charm.


So, we got that goin' for us. ;D

BTW, the Snipping Tool in W7 is changing my life. Check it out, screen grabbers! (And thanks to Celshader for pointing it out!)

03-17-2013, 05:38 PM
Yeah, for rounder to work right on edges, you would need to slice that up and connect the corner of the insets to the outside. I like to do that, cause it makes it easier to make sure I keep the polys to 3 or 4 pts....though with the new slice tool its easy to fix that stuff.

03-17-2013, 06:14 PM
Yup: it works great on ONE, beveled-in slot, but 3 booleaned slots, ehhhhhhh, not so much.

"SLICE" ..... tarnation, so many new tools to remember. The killer is I can almost always remember what they LOOK like, but not the name. Grrrrrrr.

03-17-2013, 06:18 PM
hmmmm, sorry Jeric. Still seems like there's a problem with the result. At least you got a result instead of an error message. :)

03-17-2013, 06:20 PM
A couple of quick knife cuts (I'm still in 9.6) and it shouldn't be a problem. You just need to "interrupt" the two edges connecting the three boxes to each other.

03-17-2013, 06:39 PM
Hmmm, I was doing the thing w/an extraneous point, but it was causing more trouble than it solved....

Good eye, joe.

Adding "stopping points" or cuts just makes it fail again. :cry:

I'm attaching an object before trying anything (remotely) tricky, so everybody's on the same page.


03-17-2013, 07:47 PM
My first knifing (on just the two little connections) was giving some really sharp angles. It was was rounding, but not well. So I just went with a simple cut down the middle to start, to get away from the angles....... you'll see. Added just a wee bit to the cleanup at the end.

03-17-2013, 08:56 PM
I sliced between each inset, and again twice near the ends..rounded the verticals, then selected the edges around the opening, rounded them, then just a lil slice time to make all 3 or 4 pts


03-17-2013, 11:05 PM
tHANKS guys. Very educational.

and now..... ARGHGHGHGH. This reeeeeealy makes it difficult to use (I know I know, it's been said before) as a GENERAL tool, if you have to jump thru a dozen hoops to make it work the way we all seem to expect it to work, i.e. semi-painlessly.

In this case, all I wanted to do was round/chamfer/fillet some 'ventilation slots' in a housing. Now it gets muy laborioso.

BUT, thanks for the workshoppin'! :thumbsup:

03-18-2013, 02:39 AM
Hehe....Its never as simple as you think

03-18-2013, 05:48 AM
Ok 2 separate issues in here, which are both WELL doccumented and solvable

1) Rounder ignores edges. If you select edges, all Rounder sees is points, so any selected points have rounding interpolation. Which is precisely what has happened in the first issue. Simplest fix is to select the front face and inset it, with bevel or multishift or similar, then perform the rounding operation.

2) this is *NOT* a rounder error, but a classic opengl/shading error caused by openGL tessellation. Basically put you have a shading issue happening on the flat face, because it is a complex ngon (or series of them) which joins to a face which is curving away from it. Because openGL basically tessellates ngons into triangles for display, it guesses how to subdivide the polygon, then attempts to smooth between neighbouring triangles. The result is a mess. To mitigate this you need to create a flat neighbouring polygon region between the rounded corner, and the flat ngon face (NB, manually tessellating the Ngon into triangles doesn't solve the issue). Rounder has an option to 'solve' this, called 'opengl poly' which adds an additional flat loop either side of the rounded edge. This option however causes later problems with Rounder (because of the open edges it can cause) so it's best to manually do it yourself, before the rounding operation. Note also I say 'OpenGL' error, but it will cause similar artefacts at render time too.

03-18-2013, 09:25 AM
Ok 2 separate issues in here, which are both WELL doccumented and solvable

1) Rounder ignores edges. If you select edges, all Rounder sees is points, so any selected points have rounding interpolation.
Even when I try selecting just one set of four of the points a the top of the insets in the sample mesh, I get an error out of Rounder.

I'm not sure what you mean when you say "simplest fix", as that makes it extremely difficult to select the points in question.

03-18-2013, 02:48 PM
If Chamfer did multi-segment chamfer's I'd not be using Rounder at all. sadly, it doesn't... Maybe LW 12....

Yes, we wanted to try and get that in, but ran out of time.

03-18-2013, 03:29 PM
In these kinds of operations, it seems concavity is the spice that makes everything difficult (like Beveling tight angles).

It's only a type of concavity though: this is hard to explain, but: once you slice the object up in a way that would allow the parts to fall apart revealing the edges to be Rounded to be notionally external, then Rounder works fine.

But the amount of prep-work you must put in to achieve this state makes the candle scarcely worth the game. (Or whatever that saying is.)

03-18-2013, 03:39 PM
I had forgotten about Rounder,

just tried it, made me frownder...

03-18-2013, 03:56 PM
I had forgotten about Rounder,

just tried it, made me frownder...
Hah! Maybe all complaints, versus valid bug reports, should HAVE to be made in the form of a limerick. ;)

Shrox, do you have LWCad? That seems to be the real solution. (Not to don the tinfoil, but perhaps Rounder is a favor from NewTek to the LWCad guys.....)

03-18-2013, 04:28 PM
I have LWCad 2.1....havent had the funds to get upgraded :(

Ivan D. Young
03-18-2013, 06:07 PM
I did some examples some years ago, probably the LW 8ish years that Rounder seems to get its wonkiness from vert normals or something like that. so when you have been heavily chopping up a mesh or modifying it. The verts now are no longer completely associated with the direction of the face. If you take that face off into a new layer, rebuild the face and paste it back in. You can get much better results. Rounder seems to want to round based on an average of the sides associated with that vert. normally, it is a value like 45 degrees or something similar. you find if you start checking that some of the verts seem to pick up a direction other than the average of the two adjacent faces. Rounder is actually working as expected. you just didn't know that the averages of normals it is using have been mutilated. Plus in LW there is no way to fix vert normal wackiness. So you have to rebuild those faces and you should be good to go. Hope that helps.
I would work up some examples, but I did post them like years ago. I currently do not have LW installed but If I get the itch I will do some new examples that would be more self explanatory. I know someone at work who still does LW and has it installed. maybe I can get him to do some practical examples.