"filleting" the "corner" of a surface (first "non-trivial" surfacing in onshape)

kyle_altendorf

https://cad.onshape.com/documents/5a38fb744042fa8c2982151a/v/595b8b3caca1ce5234c1e6e0/e/e61578f4c763bc663c1070c6?renderMode=0&uiState=67bbb1de8b0f840d790ca7a4

The main surface there is a guided loft. The 'rolled' edge is a simple sweep. You can see a hint of a bridging curve there that I made while trying to experiment around rounding out that sharp corner of the loft.

What is a proper way to create a surface like this with a 'filleted' corner? I know I could create some plane, sketch a fillet, extrude to a surface, then split the loft… seems pretty long-winded and I expect it would not give proper curvature control etc. So, what is the/a 'right' way?

Sorry for all the 'air quotes' here. I'm trying to acknowledge my lack of proper knowledge around terminology. Thanks for any help.

Cheers,

-kyle

MDesign

Simple variable fillet work for you?

MDesign

Or split your loft face with your bridging curve. delete the little face making the corner and redifine the sweep path to include the new edge of the loft.

jelte_steur814

this is your seat with 'tangent edges removed'

do you see how some of the rounded edge is still a black line along the length of the seat? meaning it's not tangent in those area's.

probably not what you want. I think the sweep approach is not the best way to go if surface quality matters.

moveover there are some 'semi' fillets here in this area which are the result of the isoclines in the loft that are a rather arbitrary result of the 2nd and 3rd guide which will be difficult to control or get right.
I'd suggest a different approach:

1. create overbuilt surfaces for the back, seat, and side,
2. create an overbuilt contour surface to cut with
3. create a bunch of face fillets to transition between these.

you can add more and more control and precision to the design by tuning the first surfaces.
this approach will result in cleaner surfaces and an easier to edit result, especially using the new 'edit curves' feature on a few strategic curves.

especially the side surface shape will have a big impact..

https://cad.onshape.com/documents/ce2f1d9affb355246c852948/w/6f1bf4725859cd6fafad3757/e/127dade981c433389c9940e4?renderMode=0&tangentEdgeStyle=2&uiState=67bc493aa92568487a5b4f08

you'll get something like this.

kyle_altendorf

@MDesign, your first attempt I had already done but I did indeed want to eliminate the corner on the first surface, not just that edge on the round over. So, your second effort looks like what I wanted. I know I tried exactly that and… I guess I gave up much too quickly. I got it working now. Thanks for pushing me back to that.

kyle_altendorf

@jelte_steur814, thank you for the detailed response and example. I was aware that the quickly hacked up example had poor (just terrible :]) surface properties. I was specifically going a 'non-planar' direction with this experiment. I can see how the planes and overbuilt surfaces offer a few distinct benefits and appreciate you bringing that to my attention.

As I created the roundover I did notice I was not specifying constraints to the surface and knew that wouldn't be good. That was another point I was intending to explore. I just redid that roundover sketch with an intersection of the seat surface, instead of using the 'center edge' at first. The lack of tangency (let alone any higher continuity) is still present on the sides. Locking the sweep profile to the face of the initial loft seems to have addressed the tangency at least, but I know that's a bare minimum requirement in surfacing. When thickened, some tangency issues remain on the added face. You're concerns about the stability of my initial approach remain valid. :]

Thanks again for presenting the alternative approach.

kyle_altendorf

https://cad.onshape.com/documents/5a38fb744042fa8c2982151a/w/c2e66a2445f0668f88188944/e/e61578f4c763bc663c1070c6?renderMode=0&tangentEdgeStyle=2&uiState=67bc79c17709ad15ce6b358c here's my adjusted model, for the record

MDesign

https://forum.onshape.com/discussion/comment/114361#Comment_114361

awesome glad you got it. I always go for simple then progress. Never know the knowledge base of the inquirer.

kyle_altendorf

When splitting based on the bridging curve, the curve must be projected onto the surface. I selected normal to target, but I think this concern will apply regardless. Doesn't this allow for the curvature "cleanliness" of the bridging curve to get "damaged" as it is projected onto the surface? In this exact context the bridging curve is only minimally away from the surface so it could be argued that the effect would be negligible, but I'm also trying to make sure I understand "correct" surfacing.

This may just be another point in favor of @jelte_steur814's concerns about this approach.

MDesign

If your worried about surface quality. you might think of the sweep as an intermediate step to the final product. use the sweep to build the edges of the the next surface. once that surface is built delete the sweep and apply a proper blended,tangent surface. That just one way to look at it or attack it. if the adjactent surfaces are known or built seperately you wouldn't use a sweep in that manner and opt for some other surfacing tool like lofts, of boundary or others.

jelte_steur814

Hi kyle,

try both and evaluate the curves using the curvature combs.

you'll learn that projections will indeed mess up the input for the next surface. I expect the combs will look a lot less clean. that's why projections in general are to be avoided.

be aware that the quality of your surface is very much dependent on the quality of the edges/curves it's made with.

so garbage curvature edges in, garbage surfaces out.

A fillet corner for surfaces feature would be a very nice addition here for Onshape. but i don't think the required operations are there yet to featurescript something for that… (I'll think about it though)