USER FORUM
(you are viewing a thread; or go back to list of threads)
fillet workflow via line styles (by cmpxchg)
I hear many people complaining about the lack of fillets and chamfers in solvespace. I was wondering if a the following workflow via line styles would work:
A new application-provided line-style defined in the reference group could be used to mark a line that borders two faces as a candidate for fillet or chamfer, and indicate it with a new color (variant of white) for this purpose.
In a post-processing step these lines can then generate extruded difference-solids, that are aligned to that line and the adjacent faces.
This post-processing could be an integral part of the in-application rendering, and switched off for performance, nurbs/triangle reduction purposes etc.
Fillet parameters such as radius could be a document-specific- or a group-provided parameter, or overridden by a line property.
The fillet/chamfer line-style-property assigned to a line could reference the faces as a property or pick them automatically.
This looks simple, but could get complex pretty soon - is there any literature that I should read before trying such approach ?
A new application-provided line-style defined in the reference group could be used to mark a line that borders two faces as a candidate for fillet or chamfer, and indicate it with a new color (variant of white) for this purpose.
In a post-processing step these lines can then generate extruded difference-solids, that are aligned to that line and the adjacent faces.
This post-processing could be an integral part of the in-application rendering, and switched off for performance, nurbs/triangle reduction purposes etc.
Fillet parameters such as radius could be a document-specific- or a group-provided parameter, or overridden by a line property.
The fillet/chamfer line-style-property assigned to a line could reference the faces as a property or pick them automatically.
This looks simple, but could get complex pretty soon - is there any literature that I should read before trying such approach ?
(no subject) (by Daniel Engineering Solutions)
I like the idea a lot! I don't have any recommendations for literature but I've heard that the main thing keeping a fillet integration into SolveSpace is that the calculations required are pretty intense and the solver isn't built well for it. That could all be wrong as I'm not a computer scientist so I'm not exactly sure what is required. I definitely think its worth digging into though! Best of luck!
(no subject) (by cmpxchg)
perhaps the operations can be inserted on a per-group basis, in a subgroup that follows the group, via a checkbox changes its group, maintaining the order of groups. Perhaps the constraint-solver should be avoided/disabled for these sub-groups - no new user-definable connection/relationships for the new objects and new lines/surfaces appearing in the sub-groups. The faces themselves will become smaller due to the fillets and chamfers. Lines get split into two lines, and a curved, and therefore unselectable 'face' (mesh,or nurbs?) in between them.
Primitive translations for generated fillets:
lines: arc+2 lines, linear extrusion along line, diff or union (or assemble)
arcs,circles,: arc+2 lines, lathe along arc
bezier: problematic, since there is no sweep-extrude along bezier support right now in solvespace. Perhaps add this first before going for fillet feature
fonts: no comment... Just skip fillet feature.
one things to mind: the group-to-group entity managment needs to be perfect and handle insertion and removal of a subgroup in a seamless way - might not be trivial.
Primitive translations for generated fillets:
lines: arc+2 lines, linear extrusion along line, diff or union (or assemble)
arcs,circles,: arc+2 lines, lathe along arc
bezier: problematic, since there is no sweep-extrude along bezier support right now in solvespace. Perhaps add this first before going for fillet feature
fonts: no comment... Just skip fillet feature.
one things to mind: the group-to-group entity managment needs to be perfect and handle insertion and removal of a subgroup in a seamless way - might not be trivial.
sample chamfer object (by cmpxchg)
Find attached a experiment with chamfers added on about all edges.
Using constraints I tried to create a well-defined sample, so that an automated system could chamfer it in a similar way.
I used a new workplane in the middle, and perpendicular to each to-be-chamfered edge.
A two-sided difference extrusion with an arc and tangent- and perpendicular lines defines the radius and direction of the chamfer.
Perhaps I should approach it differently: Extrude the exact same chamfer 2D workplane template twice, using one-sided extrusion, in both directions.
The extrusion lenghts on each side only happen to be similar/same, resulting in either overconstrained two-sided length OR round-off errors on the opposite side.
In an early group, I already had to enable 'force nurbs to triangle mesh'.
It renders fine on screen (altough a bit slow), but exporting STL creates many small errors with floating objects.
Would there be a way to modify or redesign this object chamfers, so that it is well-defined and can exports a perfect STL file ?
Using constraints I tried to create a well-defined sample, so that an automated system could chamfer it in a similar way.
I used a new workplane in the middle, and perpendicular to each to-be-chamfered edge.
A two-sided difference extrusion with an arc and tangent- and perpendicular lines defines the radius and direction of the chamfer.
Perhaps I should approach it differently: Extrude the exact same chamfer 2D workplane template twice, using one-sided extrusion, in both directions.
The extrusion lenghts on each side only happen to be similar/same, resulting in either overconstrained two-sided length OR round-off errors on the opposite side.
In an early group, I already had to enable 'force nurbs to triangle mesh'.
It renders fine on screen (altough a bit slow), but exporting STL creates many small errors with floating objects.
Would there be a way to modify or redesign this object chamfers, so that it is well-defined and can exports a perfect STL file ?
Post a reply to this comment: