Surface continuity analysis

@jim and @Lagom

I guess it’s these kinks in the zebra stripes that makes it not as flowing as one might need?


Tho yes, I get your point, jim, that G1 would be adequite in most cases. I think I can probably live with G1s in most cases,
probably more importantly may be how the surfaces are made and not just the how edges are connected.

At least for me I feel I need to “investigate” the surface and not just turn on stripes and look a bit to see
how the surface transition goes… just a quick tool would be nice. Maybe after few more months I can get a better sense of using it…

Was also thinking if the curvature analysis tool would help…though I guess with G1 connected surfaces it seems you can’t really see the transition well…

It depends on the business sector you’re in, on the type of clients you work for.

Rhino is a NURBS surface modeller, so it should contain tools that offer interactive control features that help the user manipulating the shape and continuity of surfaces accurately.

1 Like

If the curvature discontinuity in this file IsThisG2.3dm (843.2 KB) is so easy to see then why do you ask for better tools to help you find such curvature discontinuity??

BTW, In Rhino the tool that will interactively show you where the curvature jumps are is called CurvatureGraph

That wasn’t the point. Using G2 continuity can make a better looking model but can also make things uglier. The same could be said for G1 continuity.

Sorry for the wrong interpretation…
Also I’ll use curvaturegraph more often with zebra to check any bumps.

Please show me two surfaces, G0 along the common boundary (with CPs shown, ready for interactive matching to G1) and now bring CurvatureGraph into the game - how does it help making one surface G1 to the other? I’m all for learning things I have missed so far. Please find attached an STP file of a typical situation: a corner with three unequal setbacks, later to be executed in ABS to SPI A-2 (Grade #6).

Corner with unequal setbacks.3dm (314.3 KB)

Thanks in advance!

PS: Result

here a thread where this was mildly discussed i believe there are a few more.

curvaturegraph is only visually and a mess of course. but history with curvaturegraphs set on input curves to control the surface works at least visually. probably a mild comfort for a real matching situation, since a dynamic isocurve CurvatureGraph would be more likely what is needed.

there is no numeric solution as far as i believe, but maybe PointDeviation could be enhanced, if it would have the ability to analyse the continuity as an option. it would work numerically at least and would not be very far from what is needed.

1 Like

[quote=“RichardZ, post:13, topic:49621”]
here is no numeric solution as far as i believe
[/quote]Numeric solution for ???

for a surface equivalent to this ??? !!!

Command: GCon
Curve end difference = 0.00 millimeters
Radius of curvature difference = 459.10 millimeters
Curvature direction difference in degrees = 64.05
Tangent difference in degrees = 0.00
Curves are G1.

:face_with_raised_eyebrow:

1 Like

My comment about curvaturegraph revealing where the curvature jumps are was in regard to the model I posted.

I assume you are talking about manually editing surface control points to modify the surfaces so that they are tangent. There isn’t anything designed for that purpose. Personally, I would welcome something that does that, but I think there are very few Rhino users interested in manually matching surfaces for continuity.

1 Like

That’s too bad, really, because the information is surely there somewhere in the bowels; it would just need an interactive display of it, while matching or manually manipulating CPs. Then one could use Rhino for many top-tier clients and reduce SolidWorks or Alias billing hours.

2 Likes

A report on surface continuity along a pair of surface edges could report:

Surface edge difference

Radius of curvature (or curvature) difference in direction normal to the edge

Radius of curvature (or curvature)  difference in direction tangent to the edge

Edge tangent difference normal to the edge in degrees

Edge direction normal to the surfaces in degrees

The math to calculate these quantities is straight forward.

how would you calculate it? without having some computational maths studied i am not sure if i can say the same for myself… if you want to explain that in a tiny nutshell as a starter that would be interesting. i still remember that there are explicit and parametric forms, explicit on simple linear calculations i think like circles spheres or flat surfaces, parametric for more complex geometry which probably would be needed i this case.

i tried some nurbs books a year ago but i havent touched them since they are a bit intense without education in this field. so if you want to explain that be sure its in a crystal clear nutshell with a refractive index of vacuum :upside_down_face:

yes for example, i would add the conclusion if its continuous according to the classification.

Hi, everybody,

every time someone asks how to improve the quality of his model by means of G2 / G3 transitions, people quickly appear who want to talk the interested parties out of curvature continuous modelling because of the too much effort. It is claimed that G2 is exaggerated. For me, however, the statement that G2 is “marketing bullshit” actually marks the hitherto best of the false statements on this subject.

From the design point of view, a G2 transition never has disadvantages compared to a G1 transition. G1 will always leave a visible light edge in the model.
In the production process, however, there may be reasons to build only G1.
For example, if you use a tool that absurdly increases the time required for G2.
Or if your own skills are not sufficient to create the model in the required time.
Or if the design model is rebuilt by engineers who lack tools or skills.

Everyone using normal tools in Rhino should be able to tell the insufficient quality of the model surface provided in “IsThisG2.3dm”. Even the simple matte white shader shows the problems. This model actually shows pretty good, what G1 actually gives you.These “features” would appear in the end product for sure. It does not matter whether the surface will be high-gloss or matt.

I know I used hard words here. But I think I’ve stayed objective. Please do not regard my words as an attack, but as a request to look at the subject again from the points of view outlined here.

But pictures are known to say more than a thousand words, therefore the analysis of surfaces in “IsThisG2.3dm” as images.

First of all: a real-life example of the painful, widespread result of ignored curvature continuous edge processing.

Now the model. Check how the faults appear exactly at the G1 surface transitions…






Hi @digiformer thank you for your feedback.
When I make like a sweep2 with say, 3 sweep2 coming together for multiblends, I can’t get a good continuity on the surfaces. Like a letter K but a curvy in all axis

Is it that I need to start with curve more than degree3?
Or degree does not matter and I need crosssections of the sweep2 to have three edge points to allign on a line?

There is no insufficient quality in that model that’s just your biased opinion. The model has G1 seams, but if those seams were on a manufactured part nobody would be able to see where the seams are located. I base that opinion on many years of experience creating lots of models with seams like that have been manufactured. What do you base your opinion on?

How do you know that there is a G1 seam there?. That could just as easily be one continuous surface (there’s no seam there at all)

Yeah there’s no abrupt “G1” transition there. Is that a phone? Those things are brute-force point-sculpted beyond all reason, they’re not going to be examples of “ignoring” anything.

If those who have to resort to juvenile language work for uncaring clients in uncaring industries, G1 is perfectly ok. To dismiss that there is a wealth of quality conscious clients from FMCG, medical, sports equipment and electric goods sectors, let alone aircraft, marine and public transportation sectors, is nothing but absurd.

I think these kinks are caused by the tessellation of the model is too low. Turn on the preview of render polygons and i believe you will find polygone edges right at the kinks.
Evaluation tools using shaded or rendered views will never tell you the whole truth. How close you get to the truth depends from the fineness of the polygon mesh the tool is forced to use.
To avoid misunderstandings: Zebra and friends are of course a very good method for assessing the visual quality of the model.

Go to the custom polygon settings of your model, put every value to zero and play around with the “maximum angle” value. Go below 0.5. But be careful. This is quite resource demanding. You have to try out what your system can handle.

Uhh, you guys ar

Uh, you guys are new here, right? Accusing lowercase jim of being lazy or ‘uncaring’? Hah, oh, wow. No, he’s more obsessed with the quality of Rhino geometry than probably anyone

1 Like

Hmm I think I need Help in learning creating curvature continuous surfaces.
For example, I’d like to make three way surface coming together and connect with surface continuous segments.

I was thinking extracting Isocurve, then blendcrv(getting the curve in blue)
then sweep2 to get the surface.
But, looks like it’s not working so well.

Is it the cross sections I need to go back to?
Is it the degree 3 curves which are not suitable?

Any tutorial/links etc… is appreciated.
I’d like to learn from scratch of setting up good curves -> making good surfaces -> making good polysurface.

Curvature Continuous.3dm (319.6 KB)