Camfer on irregular editable mesh objects

[AdamT]

Hey Squid,

No, I don't think there is any good way to do this. It's been a frequent gripe.

[Per-Anders]

just make sure the normals are aligned, then select the edges and use bevel on the edges that need beveling... maybe repeat a couple of times for rounded bevel

or am i missing something?

[chris_b]

.... yup
currently smoothshift is as close to a fillet/rounding tool that we have. Unfortunately, this kind of rounding
is difficult to implement in the absence
of NGons (there would be loads and loads
of messy tris all over the place). See this
thread for some good examples.

http://www.cgtalk.com/showthread.php?s=&threadid=59419&highlight=better+boole

You can also try the bevel tool while in edge mode. Just select the edges you want to round and apply bevel 2 or three times (depending on the degree of roundness you need). The results can be acceptable for relatively simple models, but the 'fillet intersections' (corners) are handled rather badly.

Here is a quick example :

[chris_b]

... and here is an example with intersecting fillets (sorry it is so tiny... 20000 byte limit and my server is down):

[AdamT]

Originally posted by mdme_sadie
just make sure the normals are aligned, then select the edges and use bevel on the edges that need beveling... maybe repeat a couple of times for rounded bevel

or am i missing something?

Missing this:

http://bellsouthpwp.net/A/d/AdamTrachtenberg/Bevel.jpg

http://bellsouthpwp.net/A/d/AdamTrachtenberg/BevelWire.jpg

[squidinc]

Originally posted by AdamT
Missing this:

yeah, I figured that was currently the only way, and if it goes weird on just an extruded cube its gonna go nuts on something large like detailed ship hull or weapon parts :annoyed:

[squidinc]

Originally posted by chris_b
.... yup
currently smoothshift is as close to a fillet/rounding tool that we have. Unfortunately, this kind of rounding
is difficult to implement in the absence
of NGons (there would be loads and loads
of messy tris all over the place). See this
thread for some good examples.

http://www.cgtalk.com/showthread.php?s=&threadid=59419&highlight=better+boole

You can also try the bevel tool while in edge mode. Just select the edges you want to round and apply bevel 2 or three times (depending on the degree of roundness you need). The results can be acceptable for relatively simple models, but the 'fillet intersections' (corners) are handled rather badly.

Here is a quick example :

cool examples :thumbsup:

this ngons issue is really starting to piss me off, it seems like every other package handles ngons in some way or another, hell, even truespace does for crying out loud, I don't care whether some people think its not necessary for modelling "properly", I want to be able to model something the way I want not the way C4d limits me, after all if its for a high res still does it matter if there are ngons all over the place? :shrug: ( rant over :D)

[ThirdEye]

You know where you have to ask for them Mike ;)

[squidinc]

Originally posted by ThirdEye_01
You know where you have to ask for them Mike ;)

yeah I know, buts its like talking to a brick wall :annoyed:

[ThirdEye]

Originally posted by squidinc
yeah I know, buts its like talking to a brick wall :annoyed:


hehe i know what you mean, the problem with ngons is they'd require a rewrite of both the editor and the renderer, right now C4D thinks about a tri as a quad with two points at the same position, introducing 5+sided polygons would require a totally new modeling paradigm (and many months of work :hmm: )

[basilisk]

not an answer to the original question, but using hypernurbs with weighting can achieve a nice selective bevel. As Squidinc points out this substantially raises poly count, as you need to turn up the subdivision count on render to 4 or 5, see attached. But it is possible to get reasonable joins between sharp edges and beveled edges (see attached)
james newman

[mnu]

To contribute a little to the ngon issue. Why do you think it would be so tricky and exhausting to implement ngons ThirdEye? As far as I can imagine a nice new "ngon" object and some modifications to poly modeling tools and the Catmull-Clark algorithm should be enough. On the most basic level all objects are tesselated to triangles, ngons or normal polys, there's no difference. For sure, complete switching to a ngon environment would be a big affort, but just implementing them as another possibility shouldn't be that hard.

Just my two cents, cheers mnu

[AdamT]

Originally posted by squidinc
cool examples :thumbsup:

this ngons issue is really starting to piss me off, it seems like every other package handles ngons in some way or another, hell, even truespace does for crying out loud, I don't care whether some people think its not necessary for modelling "properly", I want to be able to model something the way I want not the way C4d limits me, after all if its for a high res still does it matter if there are ngons all over the place? :shrug: ( rant over :D)
You're preaching to the choir, brother. Unless or until Maxon does something about it I think the best solution is to just add a second modeler like Rhino, EI modeler, FormZ, LW, or even trueSpace or Amapi. To be honest, this is my *only* real issue with Cinema, i.e., the only problem that forces me to turn to another application on (rare) occasion.

[ThirdEye]

Originally posted by mnu
Why do you think it would be so tricky and exhausting to implement ngons ThirdEye?

Because i sadly know it for sure mnu, it would require a rewrite of both the modeler and the renderer. :hmm:

[mnu]

Well, then. I only thought about how difficult it would be to write a ngon plug and came up with this: It defenitely would be possible to write an object plugin that does ngons. The problem is the integration into the C4D quad/tris core, right? But the plug would be a custom polygon object with custom controls, that internally deals with ngons. It would do its own point/edge/poly display. Maybe one has two write custom editing tools for this, too. Then, whenever the object's chache is submitted to the core app the plug would generate a traditional triangulated mesh from the ngons, as e.g. Max does when rendering. With this all Cinema parts would be able to deal. And for the SubDee part: The plug could come with a new HyperNurbs wich uses an evolved Catmull Clark algorithm that is capable of handling ngons as e.g. found on the Siggraph web site. Once again the output of this custom HyperNurbs would be plain quad and no problem to the cinema core.
For sure, this wouldn't be as useful or comfortable than a included ngon system but it would be a start.
The main question is: is all this too much afford to a plugin developer? Maybe next year such a plug could be ready but Maxon decided to integrate it into the core app meanwhile.

Just some thoughts. Or is there anything I missunderstood about the whole thing?

[chris_b]

hmmm.... well they'd better get started on it then.
This is something that will have to be implemented
eventually, and the more they build on a lousy
foundation, the more work there will be later on.

It is actually kind of amazing that there is no
fillet/rounding tool. It pretty much forces you
to use HNurbs and rack up the poly count if you
want acceptable results.

Pretty much any man made/machined object has rounds
somewhere in its geometry, so this really is a serious
problem for anyone doing extensive mechanicall/hard-
surface modeling in Cinema.

mnu: If it were a simple matter, we'd probably have
them by now.
:hmm:

[chris_b]

mnu: I have also wondered whether such a thing might be possible.

Maxon could implement several modeling 'modes' linked together by a very robust mesher that tesselates the geometry to tri's and quads before sending it to Cinema's non-NGon core for textures, animation and rendering.

---
MODELING MODES

• standard polys (the current system - tris and quads)
• winged edge (NGons and true edge loops)
• solid/surface (Nurbs based)
---

FormZ (horrible as it is) works with a similar 'multi-kernel' system. It started out as a solid polygonal modeler, and now has a Nurbs layer built on top of it.

The down side is that a plugin like this (if it is even possible) sounds a lot like the seperate modeling app that is present in Lightwave/Electric Image.

[mnu]

OK, I agree, the form Z multi kernel approach is really horrible. But we're talking about a plugin that helps us out as long as the MAXON guys rewrite their core, no? This won't be like a seperate modeling app but only a little limited in compatibility to the rest of the application. I think of a plugin like wings3d within the Cinema Environment. And hey, this is developed by one person and it's excellent.

cheers mnu

[Per-Anders]

can't say i would be keen on a broken up modular approach like this. i.e. why would maxon invest the time in an n-gons system like wings as a modular Shave like app which is launched and closed from Cinema... when wings already exists and is free.

The core of cinema needs to be re-written to handle some form of n-gons, however n-gons are not a be-all and end-all. Lightwave has n-gons and it's decidedly lacking in chamfering tools. Maya has decent nurbs even, and it's still not got a decent chamfer/fillet tool (at least it does have one, but the results are decidedly dodgy).

In fact outside of thigns like solid thinking, rhino etc personally speaking only, i've not seen very many of these tools succesfully applied. The thing to remember is cinema will never be solid thinking et al, there's no point, others have got there already, done it well, and lets face it cinema wouldn't be able to compete. However cinema can support those third party apps fully and become part of the pipeline, that's the strength of maya, it's a pipeline thing. cinema can support that, and concentrate on being a kickass poly modeller, because enough of a foundation is there already to do this imo.

[ThirdEye]

Please no winged edge system on C4D. I HATE it, it doesn't allow poly by poly modeling, you have to start from a box and grind it till you obtain what you need. The only thing i want in C4D is a half edge system (like in Maya, Max, XSI...).

[ThirdEye]

I forgot: if you really want ngons just fill in the Maxon request form on their website, the more people sign it the higher the possibility to have them in the future, otherwise nobody will care coding them. I can't keep asking for them alone.

[neods]

I think that eventually Maxon has to get ngon's in to cinema. All the other big products and smaller products all ready support them. They will loose the battle if they dont get ngons and a few other features.

:hmm:

[brammelo]

There are two things that are very high on the modeling wishlist, and each of them would require the bigger part of a year to implement: NURBS and n-gons.

So if we would have both NURBS and n-gons, this would mean the Maxon development team would be bussy for a year and a half with only those things. A year and a half is more or less the lifecycle of C4D, so that would be the only improvement in the new version - nothing else could be implemented.

There are other ways, like enlarging the staff, but that would mean paying more for your C4D. Anyone interested in that?

Keep in mind that, although C4D is playing in the same league as the other big guys, Maxon isn't a "large" company. It isn't like A|W, employing around 600 people worldwide (used to be a lot more), let alone Discreet.

So if you really want the development of other things, like for instance a node-based material system or simply the rewrite of SLA, to grind to a halt, then keep on asking for NURBS and n-gons ;-)

Seriously guys, there's only so much a developer can do. We already get a lot from Maxon, don't you think? It could be much worse - or at least, that's how I feel about it.

Cheers,
BaRa

[squidinc]

It seems a shame that maxon didn't think it was worth the effort when they were developing R6

*wanders off to look at 3ds MAX pricing*
hehehe ;)

[brammelo]

Hey Squid,

If you're willing to pay the price difference between the Studio Bundle and 3DMax, then just give the money to Maxon. They'll use it to develop whatever your heart desires ;-).

They'll even rebuild C4D for you till it resembles Max.

O.K., now I'll have to deal with the Wrath of Srek!

Cheers,
BaRa

[squidinc]

They'll even rebuild C4D for you till it resembles Max.

the couple of times I've used max I've found it very easy, maybe c4d max would be good thing, hehe :D :applause:

[AdamT]

Lightwave has n-gons and it's decidedly lacking in chamfering tools.
True, but there is what appears to be a very good plugin that does this for LW--and is only possible because of its ngon support.

BaRa has a very good point regarding the cost in development time of implementing ngons, but I believe it's a price that has to be paid eventually--and better sooner rather than later.

The lack of ngons contributes to several of Cinema's biggest weaknesses, e.g., it's very poor nurbs caps, subpar booleans (even with BB), creation of unwanted trianglues during edge cuts, the aforementioned lack of fillets and rounds, and also the inability to follow in Cinema most of the available modeling tutorials from other applications.

[brammelo]

Hey Adam (and all the other of course too),

You're completely right, but it would not only involve the development of n-gons and adapting the existing toolset, but also improving the toolset to take advantage of the n-gon technology. The estimate of 6 to 9 months was only for the first part, not the improvement of the tools - that might take a whole while longer.

Like anyone else I would like to have n-gons and NURBS (please give me NURBS as well), but we will have to buy Maxon stock to make it happen, I fear.

Of course, that would be an option: paying for the development of our own tools, just like the big studios do ;-). If we team up, it might work. Up to find a couple of thousand willing users!

Cheers,
BaRa

[squidinc]

I'd be a lot happier if maxon did spend the time, I 'm not concerned if it takes 6 months or a year, as adam said, it would improve all of c4d's major modelling flaws

[chris_b]

yes....
and if Maxon comtinues to build on current foundation, it will only take
longer and be more difficult to make
the changes later on.

This really is Cinema's Achille's heel...
it's last major shortcoming.

[ThirdEye]

I don't give a **** about NURBS, even Softimage implemented them just because some users said they were bad, give me ngons, poly+subds all the time! :buttrock:

[brammelo]

Hi ThirdEye,

I'm not talking about a full blown modeler (although that would be nice), I am talking about the possibility to IMPORT Nurbs without messing with them. Sort of like the situation in Max. Nurbs are horrible in Max, but Discreet leaves them as they are just to make sure people can render NURBS-based models in Max. I would like to be able to do the same in C4D. You can do it in Maya, Electric Image Universe and in XSI as well, so there is no reason to have n-gons and not have Nurbs.

Polygons are nice, but once you get beyond the concept-stage, you need very precise and mathematically correct information. You can't do that with polygons, you need NURBS.

Cheers,
BaRa

[chris_b]

... yeah but Cinema is intended for visualization of concepts and DCC for film and games. If you want to make models for manufacturing... you need an MCAD or Industrial Design application like CATIA, Pro E, Rhino, SolidWorks, Studio Tools..... you name it. Nobody uses DCC apps for work that needs to be mathematically precise. You have no control over tolerances even in Maya and Softimage.

That said, I'd love to see some true Nurbs in Cinema (or even just IGES import).

... and true displacement
and a shader tree
and NGons
and hardware rendering

*sigh*

and I'd like to step into a time warp and skip waiting for 5 years of software development. :D

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.