Fin Surface (Bug) / imprecise

_fin
produces imprecise results on untrimmed surface edges. - especially regarding its height / distance from surface.
the workaround is to use _offsetSrf (solid = yes) and _explode and only keep the needed surface

example shows a “20mm” fin in a document with 0.001 tolerance.
the marked position is 20.014mm

fin_imprecise_bug.3dm (64.1 KB)

EDIT:
Maybe a solution:
Loose option
and explicit tolerance (which should be document tolerance by default)

kind regards -tom

1 Like

Hi Tom - I see that - the OffsetSrf result is also much denser - I suppose that will be one consequence of a fix here…
Note duping the edge as a curve and ‘finning’ that is cleaner still, but even les precise. There was a bug track item about this sometime in the last year or two, I will look for it.

Yeah, in V8, the fin from an untrimmed edge is the simple thing… Often this will be better but it is not precise as it does not add any knots. I hate new options… I am not sure what should happen here.

-Pascal

sorry @pascal
;- :roll_eyes:

I think my preference, long term and more generally for this type of tool, is to provide some interactive input with deviation info along the lines of V8’s refit trim. It is another step, and yes more options but being able to get the right result the first time rather than needing to check and redo, seems like a good thing in the long run.

-Pascal

Dear @pascal - yes i see this concept - kind of preflight / review / fine-tuning state before finishing a command.
And I agree that it makes sense for some advanced tools, where there is an individual compromise between precision and the structure (CV, count etc…) of the result.
_rebuild is a good example
_arc / _circle deformable = yes would be fantastic to see the deviation to a real arc/circle …
and for sure other commands.

But if I already know in advanced, that I need 0.0001 tolerance - or that I love the initial nurbs structure (loose) - why should I enter this in a second step ?
I would love to see similar options for similar commands (offset, offsetSrf, fin, offsetCrvOnSrf, …)

kind regards -tom

Yeah, agree - I was just thinking over lunch that it does not fit everywhere - in this case you might need either loose or accurate but there is probably not a case for kinda loose very often.

-Pascal

loose is always nice if a surface is the start for some CV-modifications + sculpting…
(Edit:) and also if it is used for a later _matchSrf

_fin
ChangeDegree 2,3 to have one additional CV-row to keep rectangular …
start moving the 3rd row…for style.

same why i like simple sweep and others…
And i believe as users will do more sculpting - because of subD - they will also start to do this with nurbs-surfaces. some very old-school tools like moveUVN and others will regain love in the next years ;-D

The default behavior of Fin should be to return accurate (within absolute tolerance) results, with added knots if necessary.

in general accurate results should be standard. Devation from accurate results should be an option, not the default or only possibility.

1 Like

just discovered a similar topic…
and a youtrack RH-67883

As this is part of the new in V8… i hope this aspect (fin should be accurate) is also solved …
looks like the current (not satisfying) update only focuses on CV-Count / Knot -Style - not on precision.

thanks