G3 surface continuity

[rossbau]

Hello together. In order to get a serious understanding of G3 needs amongst surface modelers I want to ask you three questions:

1. How do you/your customers measure G3? I am aware about the smoothness of curvature combs. If there is another method, please answer. I am specifically interested in tolerances and how those tolerances will be calculated.

2. In order to achieve G3 mathematically there must be added a fourth CV row. If this is the case then I would like to know if it won’t hurt you. A smooth curvature plot across two surfaces can be reach even with the proper scaling of the first three CV rows.

3. In which tools do you see G3 surface continuity?

I know this topic is somehow controversial. I need solid and serious answers on those three questions in order to use it for designing our software. I appreciate your input in advance.

Uwe Rossbacher

[c3systems]

Hi Uwe,

Evaluation:

G3 continuity is the matching of the third derivative. In other words, the curvature plots between surfaces/curves should be tangent continuous. G3 continuity is still evaluated qualitatively. Apart from the 2D curvature plot of the cross-section curves, we also look at the amount of "bleed" of curvature diagnostic shading from one surface to the other. It might be computationally expensive to calculate G3 continuity in real time using the derivative functions. It might be good feature if we can get curve definition instead of polylines for curvature plots. There are some patents with Dassault regarding G3 continuity measurements.

How:

Other software packages (Rhino, solidthinking, npower) makes G3 continuous blends, using degree 7 surfaces. If the basic surface is not "tight" enough, this could lead to massive oscillation of the blend. You are right that G3 approximation could be achieved by tangent and/or curvature scaling. It could also loop if the distance between edges are not enough.

Where:

Not clear what you mean by which tool? with reference to Alias or other software?. As of now we can make G3 continuous blend curves in Alias.
Rhino, Solidthinking and npower allows you to make G3 continuous blends.

Wish:

What we need is a "lead-in" "lead-out" tool before we can have g3 continuous fillet. This tool should create a lead-in surfaces to a common intersection. The intersection curve will indicate how "compatible" the basic surfaces are for the proposed fillet.
There should a g3 check/creation for symmetric align

Regards,

rajeev

[Anonymous]

Hi Rossbau, did you mean "Gruetzy mitenand" by your Hello together? Then Gruetzy zurueck....

Back to your question. Not really sure how to begin....

1 First personally I'm not going to take my calculator to be sure that Alias does G3 or not, by I would like to be more than 100% sure that it's G3 when I ask for it. If I only want G2 I want to get it and it should be the same for any other requirement. We use the "Construction Options" settings to be sure that we go for A-Class or B-Class fittings and if the software shows me "Green" it's fine. Where it starts to be a problem is that the most Design studios want directly (and also in the development phase) A-Class surface which means NO SPANS.... When you go for G3 on a 5 deg. Blend Curve then you get directly 3 Spans which you don't want, so you go back and use 7 deg and so on. So you see I don't think that the problem is how to measure it because the visual Comb continuity should be good as long as there are not too much spans on the surface for the quality control.

2. To all your proposals I want to answer that for me the problem is not that the software has some limits but most of the time the user has more limits. If you have some bad surface trims and radiuses, then you will fight with the software a lot to reach G3 with as less Spans as possible, so there the user should delete and start again and there I think the "scaling of the first 3 CV rows" is enough.

3. I use most of the time the visual way, through Cross Section + Curve Curvature+ Diagnostic Shading. But I also know that some people are intensively using the deviation table to check if green is really "green".

Hope I could help you.

Andrei Kramer

Alias Design Manager
Mitsubishi Europe