Why are Rhino 7 blend edge fillets so much pointier than in Rhino 6?
In Rhino 6 blend edge fillets equated quite nicely to a 0.85 weighted surface blend, whereas in Rhino 7 they are closer to 1.0.
The problem with this is that on acute angled edges the fillet is really pointy which makes it impossible to apply even a half size smaller blend around the resultant edge. It is also (in my opinion - and many of my clients) aesthetically inferior as it no longer visually approximates a round fillet but rather a much more conical one. Is there a way we could control blend weight within the command - or at least alter the default weight in the options somewhere?
I agree - I sometimes miss the old corners as well. I think they both have relevance, and it would be nice to have the option of switching between them
Cheers Jakob. I agree with you on the old 3-sided corners. I know they’re not always the prettiest, but on small corners I don’t really need the surface complexity a setback corner creates.
The issue of fillets in Rhino has always been problematic, one step forward and two steps back.
In the wip (Rhino 8) it would take something more robust and convincing, a cut with the past.
We would need some tool (a plug-in developed ad-hoc would also be good) that improves the fillet/ chamfer.
Architects will not use them much, but many other professionals who work in the field of design in general maybe yes!
Hi Paul - the control point spacing was changed to be evenly distributed on the chord of the fillet surface -
this tends to allow more progressive and consistent curvature , with one acceleration and deceleration, compared to the old way:
But, yes, that chordal spacing should be adjustable on the fly. My guess is the old way was attempting to make as round a fillet as possible while maintaining G2, i.e. as much like a round fillet as possible. Now there is more difference between fillets and blends…
If you want round fillets why don’t you use the option to make round fillets???
People complain that round fillets are bad and then when they make them not round people complain that they are not round.
it just boggles the mind reading this nonsense.
Thanks Pascal for your explanation, it’s interesting to see how you’re calculating the new edge blends differently. And I agree that the Rhino 6 blends edges where not perfect, however I’m not sure this new approach is an improvement.
From my point of view as an industrial designer with over 15 years of professional experience and 20 years using Rhino, the blend edge command is only really of use to me if it approximates a rounded fillet while adding G2 curvature continuity at it’s ‘tangent’ edges. When developing high quality A-surfaces for plastic injection moulding (or indeed metal pressing or die casting), I would usually manually model surface blends if I need a more pointy fillet, but more often than not, when I use the blend edge tool, it’s because it’s small enough (<2-3mm) that the exact profile doesn’t matter as much, but large enough that G2 continuity is still preferable and the quantity of edges are numerous enough that individual modelling would be too time consuming.
Looking into the Rhino 7 method a bit further, it appears that dividing equaly across the chord of the fillet has the effect of stretching the blend rather a lot the more acute the angle becomes, whereas the flatter the angle the closer it gets to a ‘round’ fillet. Would it not be better to distribute the points more equidistant from each other? (i.e. each point is evenly distributed from each other after tangency/curvature positioning is taken into consideration?) If you look at point distribution more like the points of an isosceles trapezium, then you can calculate the distance of the points relative to each other, the chord and the angle.
If it’s of interest to you - I’ve built a few test cases in grasshopper to illustrate what I think might work a bit better (see links and screenshot below):
Both of these examples start with equidistant placement of points - either just of the curvature points (with tangent points added inbetween), or of both curvature and tangent points, but then tweeked slightly with a multiplier to give the best compromise across a more complete anglular range.
All of my examples are based on a ‘distance from edge’ blend - which I tend to use the most, although the ‘distance between rails’ version is more preferable if the blend angle changes dramatically along the edge being ‘filletted’ as it gives a more consistent highlight.