Multiblend in Rhino, how to do?


Greetings to everyone.
I propose an interesting video of a well-known modeler.
If someone wants to try, if you have time, I’d like to understand how you can perform with Rhinoceros and any plug-ins (VSR, for example): I was not able to do it!

Chamfer of a three sided corner. Is it possible?
Problems making Objects watertight
(Mary Ann Fugier) #2

Hi David,
We would recommend trying the Patch command.
My settings are attached. If you need it., I can make a video and show you how.

Let me copy Kyle Houchens @theoutside and see if he has any additional suggestions.
Mary Fugier


Hi Davide,
FYI, there is a similar thread in the Rhino Mac Forum. I’m fiddling around with making a manual solution for a setback fillet, but I’m not having much luck. In another past post it looks like there was something like a solution but I couldn’t quite make out the author’s methodology. It looks like his surface continuity was really nice.

Do a Google search for setback fillet nurbs and you get all sorts of people like us discussing the problem. So far I haven’t come across any fantastic solutions. I have the VSR Shape Modeling plugin, but I haven’t played with it yet enough to make the solution work optimally (using default settings it flattens out the corner which doesn’t look good). It’s fast and pretty good, but I want as close to perfect as I can get.

I’ll post again if I can come up with anything meaningful. I’m thinking a solution looks something vaguely like this (work in progress, and it doesn’t work for tangency yet):

I’ll pick it up again tomorrow for a while.

EDIT: I fiddled around with it this morning for a while longer and I was able to get G2 continuity manually, but the surfaces were too complex and have unpleasant surface waviness. I think it’s time to finally break down and watch all the VSR youtube videos.


The best solution I’ve found for this problem is using the iCapp RhinoReverse plug in patch command. It has a couple of limitations. The model tolerance should be set at .01 mm. All the input topology should be G1. The resulting patch will join all the input surfaces without any naked edges (unlike Rhino’s patch). It’s fast and easy and the results are very impressive as long as you keep things to G1.


Thanks to all of you. The problem is the following (see video to two minutes): in this case the solution to be found is more complex. Has been trimmed the upper flat surface…
Thank you for the interest shown.

(Kyle Houchens) #6

Mary’s solution is the fastest “and free” solution. In the case where the patch does not join, you can always pipe trim the un-joined edges and make a blend surface to fill them in. (a pipe trim is where you use edge curves and run a pipe down them, then use the pipe as a trim object to make a clean parallel hole that can be easily blended closed.

Jason’s solution using rhino reverse would be my next go to, but requires a purchase.

Vsr has a tool specifically for doing this (multiblend) but again requires a purchase.


I did a little test and this was not terribly challenging to do with t-splines, btw.


I’ve tried VSR’s multi blend for conditions like this and the results were disappointing. At the point of intersection of the patches there is always some continuity problem. That’s why I prefer one single patch.

The two min mark of the vid shows topology that I can’t understand why anyone would use. I would really question the results of that patch. It looked a little lumpy in spots. I’ve watched vids that make solidworks surface patch look amazing, so I tend to be skeptical when I see stuff like this.

T-splines will initially look smooth. Once you convert it to a polysurface you will see problems. There are also inherent modeling accuracy issues with sub-d. It’s great for small organic forms like jewelry and quick concept models if you are skilled in sub-d.

In the end I’ve found it’s better to just avoid topology like this altogether. Make compromises and keep things simple. Using G2-G3 blends will usually disguise these choices.


Yes, I tried some times ago with TS and you can get a nice result easly, but not clean (single span) surface.


Jason, I agree with everything you said(!) In the end I think you’re right when you said:

Having said that I fiddled around with it a bit more and I came up with a less than optimal, manual solution (way to sell it). I thought it might be of interest to a few people. Just like the VSR solution there are obvious issues with tangency but unlike the native Rhino patch solution it is at least created with untrimmed surfaces.

cornerBlend3.3dm (246.3 KB)

FYI In the video I had my mesh settings turned down which is why it appeared there were some gaps in the render mesh along the border edges of the area that I was working on. Afterwards I joined the surfaces into one polysurface; there were no naked edges and no non-manifold edges.


Problems making Objects watertight


I rebuilt your blend surfaces to G2 and had a friend patch in the 3 sided hole with VSR. The results are really impressive. It looks like reducing the number of input surfaces to 3 from 6 made a big difference. This appears like a solidworks setback fillet but with much cleaner results. I tried using a patch and few other methods all had surface continuity problems.

It’s probably a big ask if Rhino would add a setback fillet option.


Sorry if I reply to you all; these your examples are very interesting and well done, nothing to say, but look at the best video, the fitting is performed with a trimming of the upper face done in reverse. How do I achieve it with Rhino?
Thank you.


Ah, good work! nice fix :slight_smile:


I really like having control over how large the soft corner is relative to the fillets. Yes, having a setback fillet option would be a nice addition to Rhino. Compared to using a script or a tool, creating the blend manually is time consuming and prone to operator error. Also it makes iterating quite slow. Another advantage of this workflow is being able to control the topology/ edge flow of the geometry. Being able to use single surfaces whenever possible is a great advantage to having control over tangency. For example, I was able to take the same model in my previous post with a few edits and no trims:

How about this?

cornerBlend4.3dm (267.4 KB)


EDIT: Creating the surface (at 3:52 in the video) next to the trim I created at the beginning of the video seems to get better tangency if you use a surface from network of curves rather than a two rail sweep.
cornerBlend5.3dm (275.0 KB)


I managed to do it with RhinoReverse patch. I used two helper curves (magenta) to guide the shape. The final result is good as it appears smooth. However there a few continuity issues that prevent the joined polysurface from offsetting to a solid or shelling.

I would wager that this result is as good or better than the one in the video. Adding the trim to the top only adds continuity problems. Seeing how the video failed to show any surface analysis of the patch. That’s usually a pretty good indicator of the quality of the surface.


I appreciate the modeling work done, but I ask you: <<You seem a good solution? by the performance of zebras does not seem a good solution, or am I wrong? >>.


I appreciate the modeling work done, but I ask you: “you seem a good solution? Zebras by the trend does not seem a good solution, or am I wrong?”.


This my example would be a way to solve some issues maybe in the next version of Rhino (improve patch, add fillet type of setback …). It is not possible to make a decent shape you should make all those operations, otherwise what’s the software? Rhino should facilitate the task, do not exasperate him.
it is clear that something is not working; other software try to solve the problem, Rhino remains in the stone age if we go on like this!


@Stratosfear The RhinoReverse patch looks great(!) From what I can see it looks super-smooth. I wish it wasn’t $1290 USD! Based on the intended purpose I’d mostly just use only the patch function. I think I’d feel I’d have to make everything out of patches to justify the cost :wink:

For the light lines/ zebra stripes, no you’re not wrong at least for what I made in the video which was a relatively quick version (doing it manually just isn’t fast, period). But with a bit of practice and little bit of manual tweaking, I’ve been able to get it pretty close:
cornerBlend7.3dm (234.4 KB)

The light lines/ zebra lines aren’t perfect (especially in the very centre of the corner), but it’s not bad. The three surfaces meeting in the corner do not have G2 continuity. My goal is always to use the least number of control points and single span surfaces where possible. In this case I did not achieve this but the results are pretty o.k. considering there are no extra plugins needed. And all the blend surfaces are untrimmed. It’s taken me a while to figure out how to get it working fairly consistently. The devil is in the details and one needs to understand the underlying continuity issues and play around with the software for a while to figure it out. I don’t think there is a magic workflow solution for this problem, but a decent theoretical understanding of continuity issues doesn’t hurt.

As much as possible, pre-planning the topology helps (or at least not painting one’s self into a corner, like what Jason said about compromising). The problem with trying to find a good solution to the problem posed in the video is that one is starting with a trimmed edge(s). It’s not a good starting place. A bit like fillets, if at all possible I like to leave trims until last or avoid them altogether because it just adds spans/ control points and can make the surfaces unnecessarily complex. Using the patch solution isn’t ideal topolgically and it’s blending a trimmed surface with another trimmed surface. It’s not so bad if it’s the final modeling step, but if the surface requires further work, it could create problems.

I agree with you that it would be ideal if there was a native software solution that could solve setback fillets with untrimmed surfaces while maintaining G2 continuity. At the moment I don’t think there’s an [obvious] ideal, native solution in Rhino to this modeling problem. Being relatively new to Rhino, the whole process of figuring out how to do it manually really helped my understanding of maintaining continuity (which is why I bothered, I started learning Rhino in February), but I won’t use the manual method very often in a production setting (it’s just not time/ cost effective).

I would bet that coding a solution to this issue isn’t exactly straightforward. I’ll keep my fingers crossed that it ends up on the task list along with some of the other things on my wishlist.



This thread reminds me of the video tutorial series on conics & spine curves in NX.