Further improvement for Bevel tool suggestion

Hi there,

Blender’s bevel tool really nicely matured lately. Together with the bridge command both provide some serious modeling additions for object builders.

There is just one tiny feature / option missing to really complete the tool.

In the screenshot below you can see on the left the standard edge rounding in Blender.
On the right side you see a more complex but aesthetically also much more sophisticated edge rounding.

This type currently can only be hand modeled which is very complicated when you are serious about it.
The image below is a Fusion 360 screenshot and just with a checkbox you switch between the different ways
how the edges are blended or cut.

Would this be possible to include in Blender’s bevel tool?


1 Like

It looks like it would allow for all-quad surfaces and better smoothing, but it would be more be a bit more difficult to code considering that it already took a lot of work to get even the basic corners nice and even for most cases.

I would think Howard would likely make it a higher priority to fix most of the remaining issues in the knife tool and bevel tool first (knife for example still has several cases where cuts don’t work), and then add bells and whistles like what’s shown here. It does sound like a nice idea though.

Yes, the bugs need to be fixed before add other features !

Great idea by the way cekuhnen :wink:

For me, the biggest bug is that.


But the big Ngon produced on side faces would destroy the chance to loopcut (?)

That doesn’t matter if you’re using it as a modifier. I’m curious though what the actual mesh beneath that smooth shaded corner is, and what it would look like for corners that are not perfectly perpendicular, as well as for arbitrary n-edged singularities.

True, but still do not see clearly the benefits in poly modeling. I’ve seen that more in nurbs surfacing and such CADs.

Marco

what the benefit is? I guess you never used it. It allows for much smoother designs also in poly models.

With the rise of 3D printing also from Blender this is a must have feature.

Similar to the other mentioned bevel problems like having issues with concave and convex forms in one model

I did. Imho If you’re looking for precise surfacing for design you better bet on nurbs modeling and not approximated polygons in the first place (for class-A surfacing etc, or T-spline modeling)

On the other hand if you prefer design with polymodeling/subdivisions - hopefully coming - OpenSubD and their creases tool will be enough (from what i’ve seen around), good balance between polygon meshes “freedom” compared to nurbs, but at the same time much smoother results and surfaces compared to current subdivision. That sounds like the best of both worlds, at least i hope so.

…going offtopic.

Cheers

What do you mean much smoother results? AFAIK the results should look the same as our current subsurfs, except it works faster and so allows you to add more levels.

Someone with more credits on the matter can explain better about what’s under the hood, but it’s not just about speed boost. With semi-sharp creases you can keep the topology really clean without additional supporting edge loops, you just crease the edge you want sharper and the system will do the smooth transition for you at a given value, as current edge crease but right now it doesn’t add polygons, it just “stretch” the one you have and give poor results in most cases and with high values.

Better to link an deeper overview. link there are also couple of example in more design-oriented models from Modo and Harald Belker, more inline with the OP.

Cekuhnen, is there a wireframe pattern underlying the Fusion 360 example that you could show (I’m curious what it would look like); or is it based on NURBS? At any rate, it is possible I could figure out how to add this option to bevel at some point.
Pitiwazou, it seems like your picture is supposed to be reporting a bug, but I can’t tell from the picture what it is. If you think there is a bug there, please make a tracker bug report, explain the problem, and supply the .blend that you apply bevel on.

Here is a screenshot of how the NURBS data would look like incl. a mesh preview in MOI on the far right.


Here is a Blender mesh model that could work


Here is a blend file: blend

Thanks, Cekuhnen. I can see how to adapt Blender’s bevel code to produce that pattern. But it is not trivial and will have to wait post 2.70, I think.

If we are in that…

There are vertex groups… however when u have Face and u want just 2 oposite edges… it select whole 4… so it would be super cool if there is Edges groups with can be assaign in bevele modyficator…

Hi ceKuhnen. I did not understand your sample file exactly so I have probably missed something.
Five objects. 1. subsurf only. 2. 1 segment bevel and subsurf. 3. and 4. seem identical, 5 segment bevel no subsurf. 5. Your object, apparently a 4 segment bevel with extra features and 2 level subsurf. Not sure what that edgesplit is doing in there.

Odd number of segment bevels usually have a point where the original vertex was. So to even things out I made object 4 a 4 segment bevel, which smoothed out that discontinuity. Still I can see how your object has a very nice wider smoother area at the vertex! But the 7-pole created on the side and the very irregular subdivided mesh is hair raising! As expected, the area of the 7pole does not smooth properly, as you can see there is an indentation that does not go away even with more SS thrown at it.
But that corner is just perfect and round, no matter what I can’t get the bevel modifier with or without SS to produce that roundness.

A little added geometry and no more ngons or wild subsurf… This would be nice to have if only as a parametric beveled cube primitive. Hexagonal beveling for hexahedrons is appropriate.


Dru as a cube data this is fine, when you have a product that needs edge rounding it is a much more complicated process to model all this by hand.

Howardt, I am glad that this is helpful to you. Thank you a lot for everything.

Oh I wasn’t suggesting hand modeling! I think Howard has endless tricks up his sleeve when it comes to spiffing up the bevel tool, and this might be an incorporatable algorithm.

The point I was suggesting was that any 3pole vertex is better beveled this way than with a triangle in the middle. Whether that’s true across the board or not is another story, but it would be nice to have this option available. In addition, always offering both odd and even bevel segment numbers is a problem too.
I would have no problem with the bevel tool offering 1 segment-chamfer and thereafter even numbers of bevel segments only.