How do you do something like Radiate Surface - but in 3d?

It would be so much easier to explain if I could post a picture of what problems I'm having and what I'd like to do...

So why don't you take a screen shot and upload your pic on ImageShack

and then post a link.

Muggs

doh!

ok here's the first image that shows trying to offset surfaces (and failing)

the second image shows why it failed and the third image shows what I would want it to do I've even tried Offset Entities in a sketch but that messes up on splines as well.

I hope it's a bit clearer now. So like in picture 3 but in all 3 dimensions, not just 2. Thank you!

Look into "Ruled Surface" feature. I think it will do what you need.

nope, that doesn't do what I need. think offset surface, but "better", as in "radiate surface" kind of better.

The curvature being pointed to in the third image isn't why it's failing. What you point to is an outside radius and when offset is getting bigger. Failures are caused by the radius offsetting smaller past zero. I would be willing to bet the failure has to do with that feather edge to the right of the arrow. When SW has to offset a very shallow angle, it has to extend edges, and when it extends spline edges, they get unpredictable and may not intersect the other side. In the first image, you can see the offset faces are almost parallel in that area, and the extension to make them intersect is going to be long. In the second image, you make a little arc out of the edge which is great, but SW won't do that for you automatically.

Also, its not clear at all where the 3D curvature in your part is or what you are wanting to do in 3D. You shouldn't be so fast to dismiss Ruled surface functions, they are the improvement on radiate.

I suspect that this is not the part you are actually trying to offset surfaces from. If it were, then the solution would be a piece of cake - use radiate (or ruled surface - at the core they are the same thing), convert the outer edge, extrude, and you have the offset surface you want.

The reason that offset surface doesn't always work on more complicated samples is something you probably know intuitively from your experience mowing your lawn. When you mow your lawn, you are in essence offsetting the perimeter of your lawn with each pass of the lawnmower. At first you have a single curve around the perimeter, but with each pass the inside corners get tighter and the outside corners get rounder. At some point, the inital smooth curve on the inside corners turns into a sharp corner - offset surfaces (and shells - same thing when you get down to it) have to deal with that when the minimum radius of curvature on an inside curve exceeds the shell/offset distance. That is why the minimum radius of curvature tool is only partly useful

- if that minimum radius is an 'outside curve' it gets bigger and bigger and becomes less of a problem as the offset/shell thickness becomes larger. The only minimum radius of curvature you have to worry about are inside curves (for instance, rounds instead of fillets).

In general. SWx isn't as smart as you are when you mow your lawn. It doesn't generally like to split a single face into two faces (the sharp corner you create with your lawn mower after a few passes around an inside curve). However, SHELL has had some special exceptions coded into it that will allow this sort of thing to happen - often accompanied by a yellow exclamation point on the SHELL feature that won't go away (see part two of my Curvy Stuff presentations at last years SWx World - at

To get through this, sometimes we have to take the bull by the horns.

In the sample you showed, you could offset those faces that you can, then use your superior brain to make the faces that SWx CAN'T figure out itself and trim them together. Its not fun, but it gets the job done. Just last week I had to build multiple shells on a job by hand - its actually quite easy when you get the hang of it (same goes with offset surfaces - after all, shells are just offset surfaces). There are also samples of that on my website from old SWx World presentations, though I don't give any warranty for those old files working in the current SWx. However, the principle is still valid - offset what you can, make what you can't.

Ed

BTW - radiated surfaces are ruled surfaces, and are simple - just imagine the spines sticking out of a porcupine. That's what a ruled surface is - a bunch of spines sticking out of the porcupine that get linked toeghether into a single surface .I am not sure how SWx deals with the spines sticking out from an inside curve where they would overlap - that's kind of cool, but it is a 2D (sort of) problem. Turning it into a 3D problem (offset surface) is wayyyy more complicated and currently seems to take a human intelligence to solve - that's why we make them ourselves.

I'll try and describe what I'm trying to do. Look at the preview in the first picture. That's what offset faces would do and that's what I'm trying to achieve - i.e. offset all faces by 3mm. But offset fails because of the face pointed to by the arrow. Yes it's to do with the inside radius on one of the points on the spline getting past 0 where the spline begins to intersect with itself. Now what I'd want it to do is following: if a face is going to fail because of that then create a face that follows the edge as created per radiate surface.

Here is another set of images to help you visualise what I'm after

First image shows the starting body

Second image shows the resulting body And the third image shows the differences (original body highlighted in green) Clearer now?

You have some really good points on the need for a different or optional spline and surface offset function which uses similar or another set of algorithms that radiate is using.

I don't know how you access it thru the API.

Anyhow, you got me thinking about this because it's something I do need every once in a while so,.. the need for this maybe warranted? For instance, (since this is a fresh but indirect issue) a Blister package around a soft product which has complex topology.

.. (since you made me think about this so, I approximated your model.. for anyone interested..)

What I guess I didn't convey very well is that the problem exists in every 2D section through a 3D surface. Take an angled slice, parallel slices, kattywompass slices - in every slice, it still applies. The net of all those slices is your 3D surface. Just like the point of every spine on the porcupine makes a 3D shape, even when those spines overlap or intertwine. Porcupines can do this easily - SW, not so much.

Fortunately, every sample you've posted is a simple planar problem. Paul's post works, but again because it was a simple planar problem. The important thing to note is that even in a simple planar problem the route to a solution is to build the offset ssrface yourself (like Paul did by converting the edge of the radiated surface and extruding it). You can do this in 3D when necessary, usually throuhg more construction, but it can be done and a lot of times it takes less time than you would ever suspect. Usually, you only have to make a patch or two. My rule is offset what you can, build what you must.

To your point, why can't SWx figure it out? I dunno. I lack the math skills - its pretty complex stuff. I just know from years of building plastic models as a kid that the inside was often sort of 'muddy' or 'soft' looking, where the toolmaker couldn't make a pure offset of the outside (presumably with a pantograph) so they just sort of filled in material (by grinding away the tool walls manually) while keeping that wall section as consistent as possible. They did it by hand, I do it by hand (my hand, of course, on a mouse).