Cancel
Showing results for 
Search instead for 
Did you mean: 

Studio Surface: Rebuild-I

Genius
Genius
I'm confused about how Studio Surface > Rebuild works, and generally depends on trial-and-error in using it. I believe that better understanding would definitely lead to more efficient use of the core surfacing tool. Let me present an example.
 
I choose to start from two sections (one degree-3, one degree-2) and two guides (both degree-5). All of them are of single segment.
 
Image.png
 
In Case.1, I don't rebuild sections or guides, but SS automatically rebuilds the sections by adding a knot, possibly to achieve specified G0/G1 tolerance, and (in my understanding) to do its best to achieve G2 (target tolerance unknown).
 
In Case.2, I rebuild the sections only (Degree and Tolerance, degree=3). To ask SS to forget G2, I set G2 tolerance = 10. As expected, SS doesn't care about G2.
 
Q: With loose tolerance, why does Case.2 still add so many knots to the sections?
 
Image.png
Image.pngCase.1    Image.pngCase.2
Image.png
 
As in some of my other questions, here I'm not trying to fix the model, but hope to understand how the command works. Thank you!
16 REPLIES

Re: Studio Surface: Rebuild-I

Phenom
Phenom

@surfactant,

 

First thing I have to ask is what do you expect as far as knots are concerned when the 2 surfaces between which you're making a 3rd surface have unequal parameters (look at the poles)? Do you feel there is any reason why you can't go above 3 degrees on either of the first 2 surfaces (ends)? Once you do that, is there even a need to Rebuild the middle? I thought we discussed this already - good idea to use Rebuild on surfaces which you're not trying to match continuity (end surfaces); bad idea to use Rebuild on surfaces which you ARE trying to match continuity.

 

Look at the images below - does any of this make sense to you?Capture.pngYour choice on the 2 ends. Look at how the poles are misaligned.

 I changed the weight on the 2 ends to match what the minimum result was with Rebuild using AutoFit(4 degrees max, 1 segment max).  Since the sectional shape changed on the ends due to the use of Rebuild, instead of using the curves you made for the sections, I opted for the edges of the 2 Rebuilt surfaces.  Now I don't have to Rebuild the middle surface and look how good the poles look - they even match up on top of each other.Capture.pngMy preferred use of Rebuild - on the ends rather than in the middle.

You get the high knot count because you're not controlling that in your choice of Rebuild (you're using Degree and Tolerance instead of AutoFit) and because you're using low degree surfaces on the ends - the middle surface has no choice to do what it's doing.

-Tim

Re: Studio Surface: Rebuild-I

Siemens Genius Siemens Genius
Siemens Genius

I think youre spending too much time using a discussion group as a training exercise.

send me an email neil.Shand@Siemens.com and we can get training to you.

 

Rebuild typcally means:

 

Algorithm: Spend more time to create a simplified result of THIS order.

You have chosen degree 3

So: SS looks at the initial edge (degree whatever..)  breaks this down into degree 3 segments (as you have specified) THIS BREAK DOWN is driven by your G0 Tolerance

 

Setting a G2 tolerance of 10 seems to me pointless. If you really want to ignore G2 THEN dont select G2 continuity.

 

- neil

 

Re: Studio Surface: Rebuild-I

Genius
Genius

Hi @neil_shand   Thank you for your offer of help! Yes, I had some trainings and they were systematic and efficient way to to learn all the commands. And here in this community I got deep understanding on some key issues.

 

Your description of Rebuild is ok and exactly my understanding. What confuses me is like this: 

 

  • please look at my Case.1 with Rebuild disabled, which gives a good surface.
  • In my understanding, the Case.2 surface should not be more complex than the Case.1 surface, because the tolerance of Case.2 is not tighter than the tolerance in Case.1.

In other words, if it's ok for Case.1 to add only 1 additional knot to the sections to achieve a decent surface, why should Case.2 add so many knots?

Re: Studio Surface: Rebuild-I

Siemens Genius Siemens Genius
Siemens Genius

My training comment might have sounded a bit harsh.. (maybe i sound a bit guilty)

 

But looking at these threads over the last few days seemed to be going around in circles.....

nibbling around the issue but not getting to the real deal.

 

What  seems to be needed is A-Class modelling in its "purest" form..  To avoid all the compromises...

AClass.JPG

Maybe my training comment wasn't harsh at all.... (relief)

 

- Neil

Re: Studio Surface: Rebuild-I

Siemens Genius Siemens Genius
Siemens Genius

Case.2 so many knots. I've already answered =  G0 tolerance and you forcing a degree 3

 

a 3rd order curve CANNOT fit to a higher order definition (see one of my recent replies)

- hence the break-down = added knot complexity

 

If you think this is wrong: submit a PR.  If you think this could be improved: submit an ER.

Start with an IR to Gtac.

 

- Neil

Re: Studio Surface: Rebuild-I

Genius
Genius

Hi @neil_shand   Any kind of comments is ok for me. Never worry about that! Thank you for the very helpful teachings in your past posts!

Re: Studio Surface: Rebuild-I

Genius
Genius

@neil_shand wrote:

Case.2 so many knots. I've already answered =  G0 tolerance and you forcing a degree 3

 

a 3rd order curve CANNOT fit to a higher order definition (see one of my recent replies)

- hence the break-down = added knot complexity


Yes, degree 3 is too low. But, Case.1 shows that degree 3 is good enough to give a decent surface, with only one knot added.

 


If you think this is wrong: submit a PR.  If you think this could be improved: submit an ER.

Start with an IR to Gtac.


No, I NEVER think it's wrong or should be improved. I believe that the command has its logic. I'm a new guy to NX and far from skillful enough to challenge the software that has been developed for decades. I just want to understand and then utilize it better.

 

Thank you!

 

Image.pngCase.1 shows that, degree 3 is good enough to give a decent surface, with only one knot added.

Re: Studio Surface: Rebuild-I

Phenom
Phenom
@surfactant,

According to the NX docs, the fundamental use for Rebuild is to output a surface with a better (smoother) hull (pole) structure using your choice of Degree & Tolerance OR AutoFit where you pick the degree and number of patches (segments/knots - same thing). What can cause knots or surfaces that are not smooth? Mixed degrees (of curves or resulting surfaces), curves with knots already existing, allowing NX to use tolerance over patches. If you mix and match these causes of segmentation, you're bound to end up with surfaces that are segmented, particularly if you're not telling NX I desire X number of segments.

Part of Class A is getting light surfaces with better than average hull structure (pole structure and pole alignment). Uneven hulls result in lower quality surfaces and/or transitions between surfaces (could also be construed as flow). The better the hulls are aligned and distributed, the better the surfaces will likely be when analyzed with the proper tools.

Rebuild will pretty much do one of two things in most cases - it will either output a fairly heavy surface and possibly match the tolerances (it will tell you how much it's off and where) OR it will output a lighter surface that has more than likely changed from the original intended shape (meaning it doesn't follow the curves or ignores the tolerances if it must to generate a shape). If you are attempting to hold any sort of continuity above G0, it's been my experience that Rebuild is going to not meet G1 or G2 - it might if the curve network is a bit more balanced than in this example.

My post above was intended to demonstrate how Rebuild could be used more effectively and output all 3 surfaces without any knots while only slightly raising the degree of the 2 end surfaces to get a quality result that is often desired in Class A. If you don't mind the patches/segments/knots then don't use Rebuild at all. The key thing to remember is that you're more than likely going to end up with some kind of shape change unless your starting curves are of better balance in both the U & V directions.

I don't feel you have to have a doctorate in math to understand the basics of good surface modeling - if you mix and match your curves (degree & segments), your surfaces are going to have the same level of imbalance. Rebuild is there to help improve these situations, but I feel one needs to understand that if you use Rebuild, a change in shape is likely to occur compared to the original curve network.
-Tim

Re: Studio Surface: Rebuild-I

Genius
Genius

Hi @TimF  Thank you for your explanation! And your comments on hull structure is ok for me.

 

My confusion is about the logic of Rebuild. Currently there seems to be too many assumptions and guesses about it. This makes its use often trial and error.

 

In my understanding, with Rebuild by "Degree and Tolerance", I set the degree and tolerance, and the Rebuild option adds no more than ENOUGH knots to meet my tolerance. It's a simple logic.

 

In my example, Case.1 showed that degree 3 with one added knot is enough to give a decent surface. So in my understanding, Case.2 should not add more than 1 knot.