+1
I would definitely want the sides to be constructed from one surface – preferably an untrimmed one.
I’ve come across this issue numerous times and have had to rebuild the sides. Imagine trying to edit that edge with the little slivers there!
+1
I would definitely want the sides to be constructed from one surface – preferably an untrimmed one.
I’ve come across this issue numerous times and have had to rebuild the sides. Imagine trying to edit that edge with the little slivers there!
Great, thanks! I opened an issue here. A bit more complicated question but when an entire segment of the extruded curve needs trimmed away what should the result be? I imagine both could be desired depending on the circumstance.
ExtrudeCrvTapered_Option.3dm (159.7 KB)
In this particular case, I think the surface which diminishes to the point – the triangular one – would be a trimmed one. The top surface, trimmed also (has to be!). But the side rectangular ones could be untrimmed I imagine.
I think the most important thing is that the resulting surfaces consist of one element and are not broken down into two parts, especially with those narrow slivers.
That was the particular issue that Davor raised.
My issue with ExtrudeCrvTapered is that it produces an output that I cannot use for any curve input that is not a planar curve composed of tangent lines and arcs.
In the Davor’s example file above, the problem I have is not just the slivers in the corner but the curved part is also garbage. I demonstrated in my file how I would avoid the garbage surfaces with hundreds of control points and dozens of fully multiple knots. That’s not the only way to work around that problem.
What I would like to see is a tool that is designed to make good quality surfaces for creating draft for molds and molded parts. One of the unique things that defines a good quality drafted surface is that if you project a line that is perpendicular to the input curve, that projected line should always coincide with a vertical Isocurve of the result drafted surface.
In the examples you gave the B option is the correct result IMO. The A option does not have the requested draft angle on all surfaces and the flow lines of the surface are not oriented with the direction of draft, but giving users a choice would be great.
The question of whether planar surfaces should be made as trimmed or untrimmed surface is not something that I think should be dealt with inside the extrude tapered command. Just as users can choose whether to Use Extrusions there should be a global option to use trimmed planes or untrimmed planes (when that’s possible).
Thanks for typing this up. This is all really helpful.
Yes, I understand what you mean. I created an issue here for this.
I’m going to add this a comment to RH-81764 since this is where I’m seeing this issue. If you have another example where this occurs outside a corner I’d like it.
I opened the issue here. This is probably going to be substantial amount of work to get right.
I opened a master issue here to track all the subtasks to get ExtrudeCrvTapered working well. There’s a lot of devil in the details with this but it’s something I’m hoping to work on.
Any user who wanted it done right has been doing that work for the last 25+ years. Its not that complicated. Its just annoying that the users are expected to compensate for the developer’s laziness.
I often complain that one of the main problems with McNeel’s development approach and culture is that you guys make a new feature/tool about ~10-30% right. Then you leave it like that forever, or you might improve it another ~10-20% in the following decade. And then you all decide: “this should be good enough for most people!”
Then developers’ attention switches to making more commands that are mostly addressing workarounds for the lack of proper execution and functionality of your existing tools.
These issues here with ExtrudeCrvTapered are a perfect example of that problem. And this is a core tool for anyone doing drafter parts for production, not a niche or esoteric tool.
Thanks @jim for pointing this out. I just wished McNeel could audit all their existing tools internally and say: “we have to really make all this stuff sooo much better”, before even thinking about making new tools.
G
The classic problem of the “Swiss Army knife” approach. A bit of this, and a bit of that, and now also this, and now maybe that, and then again something snazzy and shiny, woo-hoo.
Before coming out with ever new half-baked tools for any imaginable creative profession, I’d like McNeel to first return to its original claim “NURBS modelling for Windows” (and macOS of course) and get the NURBS modelling tools right.