PDA

View Full Version : Problems with Rounder



jrandom
07-17-2010, 03:55 PM
I often get crazy behavior with Rounder on simple meshes. As far as I can tell, all points are merged and there are no duplicate polys hiding anywhere.

How can I avoid this particular glitchyness?

Myagi
07-17-2010, 03:59 PM
Have you also checked that there aren't any 2-point polys?

jrandom
07-17-2010, 04:09 PM
Yep. All three and four point polys. In this particular case, band-gluing the two long middle sections made the problem go away, but I have other mesh objects where this is still a problem. I keep simplifying the meshes and moving points around until I can eventually get a good rounder applied but it's time-consuming.

jrandom
07-17-2010, 04:58 PM
On a related note, any idea of the best way to track down the cause of a "point not inside polygon" error? (See attached.)

I'm assuming this means I've got some bad geometry somewhere but I'll be darned if I can find it.

gerry_g
07-19-2010, 12:59 PM
N for numerics, zero point polly's (under polly stats) delete them, look for unwealded two polly's and merge them, remove any two point polly lines and check for three part edges under edge stats and deal with them. Look at this panel constantly after every operation like a religion, like your life depended on it, well at least if you're as crap at modelling as I am ;-) ....Oh yes and don't use rounder it's crap use Jetto Bevel and or Fillet instead.

jrandom
07-19-2010, 01:42 PM
N for numerics, zero point polly's (under polly stats) delete them, look for unwealded two polly's and merge them, remove any two point polly lines and check for three part edges under edge stats and deal with them. Look at this panel constantly after every operation like a religion, like your life depended on it, well at least if you're as crap at modelling as I am ;-) ....Oh yes and don't use rounder it's crap use Jetto Bevel and or Fillet instead.

I didn't know about the edge stats but I've been checking the rest constantly. I will look into Jetto and Fillet, thank you!

It seems a lot of the issues I've been having are where a convex edge meets a concave edge. Rounder doesn't handle those too well.

Tobian
07-19-2010, 03:26 PM
The thing you have to remember about Rounder is while it will bevel edges just fine, it will treat them as 'points' - i.e. any 2 points which connect together by an edge will be rounded, so yes, the edges you have selected will NOT work because they are all connected together in ways you didn't intend.

Either 'break' the connection, by adding more geometry (such as by cut/bandsaw/connect etc) or do alternating rows of rounding, or one row at a time etc.

Rounder can also fall over polygons which have 'open' sides. Ways to check for this include going to statistics and checking for 1 or 2 polygon points, or going into edge model and checking for any edges with only 1 or >2 polygons as those can both be causing major headaches for rounder.

Rounder isn't completely broken, it just teaches you a lot about having to keep clean geometry! :D

jrandom
07-19-2010, 04:06 PM
I didn't have any 2-point or two-edge polys so I still don't fully know what caused the issues. I've attached a render of my current work in progress (no small-scale detail, very few materials/surfaces).

The four white wall brackets were the worst offenders, having quite a few badly-rounded corners that I had to patch up by hand. (Lots of breaks and polygons that shot hundreds of meters away that had to be deleted.)

jrandom
07-19-2010, 05:26 PM
if you have LWCAD just use mass round and dump rounder.

rounder is borked and its not getting fixed :(

I've been putting off getting LWCAD for awhile now. Looks like it's high time I finally buy it. (It works with Lightwave HC, too! Yay!)

Tobian
07-19-2010, 05:32 PM
Oh dear, well it looks nice now which is the main thing! Could use some more detail :D

Sometimes with things like that it's just a little more forward planning on how you construct it.. aka some ways create more headaches with the tools than others, and that's hard to 'get'. Quite often I end up making it harder for myself too, and I know I am, but it's just the way I chose to do it hehe :)

jrandom
07-19-2010, 07:30 PM
Oh dear, well it looks nice now which is the main thing! Could use some more detail :D

Sometimes with things like that it's just a little more forward planning on how you construct it.. aka some ways create more headaches with the tools than others, and that's hard to 'get'. Quite often I end up making it harder for myself too, and I know I am, but it's just the way I chose to do it hehe :)

Yes, more detail is the plan. I just have to get the large-scale detail in there first and will work my way down to the smaller stuff. Better/more materials are in order, too. I'm also nearly done with the UV texture for that shuttle (which has taken me far longer than I anticipated).

jrandom
07-19-2010, 07:57 PM
*buys LWCAD*
*installs LWCAD*
*activates LWCAD*

OH. MY. GOD.

jrandom
07-19-2010, 10:26 PM
I understand now. LWCAD is one of those plugins that fundamentally alters the way in which one uses Lightwave. There's no going back.

*cackles with glee*
*cackles, I tell you!

colkai
07-20-2010, 05:12 AM
Amen, welcome to the fold! :D

Try to watch as many example videos of the tools as possible, the 'manual' is highly informative.

jrandom
07-20-2010, 09:12 AM
Amen, welcome to the fold! :D

Try to watch as many example videos of the tools as possible, the 'manual' is highly informative.

The manual is a little painful since on Mac OS X the videos don't launch from the Manual flash app. I have to manually launch each video from disk.

RonGC
07-20-2010, 02:33 PM
Strange, they launch quite well on my Mac 10.6

Pop in to the Wtools site and download the latest Beta of the plugin, your gonna love it!

Ron

jrandom
07-20-2010, 02:50 PM
Strange, they launch quite well on my Mac 10.6

Pop in to the Wtools site and download the latest Beta of the plugin, your gonna love it!

Ron

3.6? It's all stable and whatnot? I shall dooo eeet!

So I found that even LWCAD's rounder has problems if you've started with a nice, roundy closed nurbs curve extruded into 3D. I tried making a wall bracket for a pipe by adding two small circles near the edge of one big one, then subtracting out a circle in the middle for where the pipe would go, converted curve to poly, added thickness, then tried to round the sharp edges for the outside and the hole in the middle and that didn't work at all.

Drat.

LWCAD still totally worth the money. It fills a huge gap in Lightwave's native modeling toolset. Just wish I had a good solution for rounding edges.

RonGC
07-20-2010, 04:23 PM
Yeah its pretty darn stable. It has some big improvements to tools. Mass round has been worked on as well.

Ron

RonGC
07-20-2010, 04:47 PM
wasn't sure of your clamp design so i did a quick one with LWCad3.6.

All LWcad cylinders with some boolean add and subtracts.
Then ran the new and improved massround in LWCad.

It works very well.

Ron

RonGC
07-20-2010, 06:06 PM
So much fun had to do another quick speed model with LWCad. Second clamp design with functional opening closing action. This time used LWCad tubes and cylinders with add and subtract. Plus an additional step of a box subtract to get one half of the clamp for a mirror to create top and bottom, then mass rounded.

By using LWCads Boolean add and subtract, mass rounding gives both fillets and rounding, very nice.

Total time, just over 1 minute :)

Ron

jrandom
07-20-2010, 07:26 PM
Nice! I can't seem to find 3.6. With 3.5 I get the below results:

Trying again starting with curves, it actually worked! The downside is that I get some material smoothing weirdness (not shown as LW crashed and I hadn't saved my work) on the flat top side which I was unable to get rid of.

http://www.newtek.com/forums/attachment.php?attachmentid=86552&stc=1&d=1279675362


I tried a new approach using boolean cylinders...

http://www.newtek.com/forums/attachment.php?attachmentid=86553&stc=1&d=1279675368


... and this is the result of trying to Mass Round (both Fillet and Chamfer have the same effect):

http://www.newtek.com/forums/attachment.php?attachmentid=86554&stc=1&d=1279675374

I verified that there were no 1-point and 2-point polygons. Also tried the FixPoly2 tool, to no avail. What I want is to have the chamfer and then a fillet along the new edges created by the chamfer.

jrandom
07-20-2010, 09:36 PM
Finally got it! The trick was to split the shape polygon into two halves before extrusion so that no single polygon contained a fully-enclosed hole. To keep the top surface flat w/out any smoothing artifacts, I multishift insetted it by a tiny amount and excluded it from the Mass Round tool.

http://www.newtek.com/forums/attachment.php?attachmentid=86557&stc=1&d=1279683360

RonGC
07-20-2010, 09:55 PM
Nice work, now i understand what your design concept was. Mass round works very well there are a few idiosyncrasies you need to be aware of but with the ability to select various points mass round gives you a lot of control.

The current beta is 3.592 so it is basically 3.6 or will be soon.

Ron

Snosrap
07-21-2010, 08:58 AM
I've been following this thread and was fascinated with the simplistic shape that was being worked on so I decided to recreate the shape myself and test LWCad’s Mass Round tool for myself. Its a great tool and vastly improved in 3.6. I went overboard and ended up with a knob.

jrandom
07-21-2010, 09:15 AM
I'm still having trouble with sharp corners (I just had to go and make my shape more complex) so I'll be trying out the 3.59whatnot beta and see if that works better.

(I'm at work so I can't post any screenshots yet.)

When it works it's freakin' fantastic.

Matt
07-21-2010, 10:21 AM
Something else to remember when using Rounder.

If you activate OGL Polys (which creates an small smoothing buffer inset to fix the look of rounded polys in OpenGL and sometimes rendering).

Rounder will fail on any subsequent Rounder operations. Turning it off fixes that.

Myagi
07-21-2010, 11:07 AM
If you activate OGL Polys (which creates an small smoothing buffer inset to fix the look of rounded polys in OpenGL and sometimes rendering).

Rounder will fail on any subsequent Rounder operations. Turning it off fixes that.

Because the bastard adds the points (for that extra segment) only to the directly affected poly, leaving an edges to adjacent polys. I whined about this in some thread a while ago, but it never hurts to point it out to NT again :D

colkai
07-22-2010, 03:45 AM
Yup, one huge reason I dearly hope Newtek don't show any interest in acquiring LWCAD. They totally killed / failed to develop Rounder, as it is LWCAD runs rings around native stuff and I'd be less than amazed if Viktor continued to do the same with the CORE version when that comes out. :)

JeffrySG
07-24-2010, 01:09 PM
I'm sure the last thing they would want to do is spend time on rounder rather than work on a better rounding/bevel/chamfer tool for core.

colkai
07-26-2010, 02:22 AM
Wouldn't be difficult to understand that, after all, they spend sod all time on Rounder as it was :p

jrandom
07-26-2010, 09:06 AM
ill assume you will want NT to work on core 2 instead of fix the core 1 tools too if they are broken? :P where do you draw the line?

If they fix rounder for Modeler, the same "correct" technique can be used for CORE, so honestly? I'd like a tool that works now. :)

colkai
07-26-2010, 11:19 AM
To be honest, I'm not bothered about CORE, as far as I can tell it's accepted now that CORE is still in very early days.
I'm more interested in LW, but accept that now, if I want any fixes apart from show stopping bug fixes, I'd have to buy into LW10 with CORE.

So as a result, I'll stick with LWCAD, Viktor tends to be all over features, for LWCAD, even for little ole LW9.6 :)

My point mainly was, nothing ever happened with Rounder, beats me why the heck it was "bought into" LW in the first place, why not just give it as part of an upgrade as they did with LWCAD if they had zero intent on fixing it's problems and refining it? :confused:

Thankfully these days, there are more 3rd party tools around for rounding, so at least most of the time, you can avoid frustration by using them, or at least, reduce the frustration. :p

jrandom
07-26-2010, 11:37 AM
So as a result, I'll stick with LWCAD, Viktor tends to be all over features, for LWCAD, even for little ole LW9.6 :)

I also ran into issues with Mass Round as well. I'll have to post some screenshots from what I was attempting and what happened when I tried.

colkai
07-26-2010, 01:28 PM
I also ran into issues with Mass Round as well. I'll have to post some screenshots from what I was attempting and what happened when I tried.

Oh, not to say there aren't issues at times, but I'd trust Viktor to attack 'em. :D
I've sent the odd bug / problem to him in the past, he's always come back with something, top bloke!

I found odd "rounding" with mass round if I only rounded part of an object, it got a tad 'untidy' at the truncation point. Dunno if that's been sorted as life has kept me busy lately so haven't had time to plough through the latest builds of LWCAD. :)

JeffrySG
07-26-2010, 01:37 PM
ill assume you will want NT to work on core 2 instead of fix the core 1 tools too if they are broken? :P where do you draw the line?

Well, I wasn't saying this was the way that 'I' felt, but the way that I would imagine NT feels. ;) But to that point I can understand not wanting to fix a ton of old code that they didn't even write as compared to writing a new tool that will see many many years of use to come. I actually like rounder but do have issues with it sometimes as well.

colkai
07-27-2010, 03:46 AM
Agree 100%, a bad habit of Newtek is this "buy it and forget about it", you'd almost think they wanted to buy it to halt development. :p
When you think of the fixes / changes Pictrix and Viktor kick out for LW plugins and the development pace of 3DCoat, it's pretty scary. :)