Modeling Corruption on Revolve.

[MACKTEK]

Hello,

I am getting a modeling corruption when attempting to Revolve intersect or Revolve cut a simple repeating pattern of Hexagons on an Arc.

I am using both Inventor 2008 and also I tried this with 2014 30 day trial.

The steps are pretty simple:

1. Sketch a series of arcs that will cut thru the part later.

2. Sketch a pair of Hollow Hexagons with a wall thickness and extrude them.

2. Pattern the hexagon many times.

3. Revolve by intesect to exclude the center of the hexagona pattern (or you can revolve cut, and choose different arcs... either result is the same).

After the revolve, there appears faces on the hexagonal surfaces where there should only be "air".

Image:

[conklinjm]

Not exactly sure why the model you created shows those corrupted elements; however there are several things I did notice.

1) Your sketches were not fully constrained

2) You had many extra lines and arcs which didn't seem necessary for your final mesh

- - - -

Attached is a version created in Inventor 2010 which (I hope) maintains your design intent without those corrupted elements.

If you need it in Inventor 2008, open it with Inventor 2014 to look at how it was constructed and use it as reference when recreating a similar part back in Inventor 2008.

HTH

[MACKTEK]

Thanks. Your part does not seem to suffer from the corruption issue. However, I can't really explain what specifically you are doing that prevents the corruption. Are you suggestion that a sketch that is not fully constrained is subject to modeling corruption? Also, the original part was oriented to a different axis... I am not sure if that has anything to do with it. I do appreciate your efforts. But, It would be helpful if some good reason for the corruption of my part could be explained so I can learn how not to make the same mistake. Thanks. (Still trying to get some reason as to why the part is corrupt).

[JDMather]

---

@MACKTEK wrote:
 It would be helpful if some good reason for the corruption of my part could be explained so I can learn how not to make the same mistake. Thanks. (Still trying to get some reason as to why the part is corrupt).

---

1. If I add a dimension to your part the next dimension number is d=946.  There is no logical reason to be at d=946 in the file.

2. Everything that is in the file is poorly done from established "best practices" experiential knowledge.

In analysis,

#1 would seem to indicate that a lot of prior effort was expended in this file for which there is no longer evidence in the feature history tree that would explain the logic of 945 prior dimensions.  When I see something like this I am reminded an old term that was taught in computer programming classes - the GIGO Principle, that is Garbage-In-Garbage-Out.  In other words you should expect poor results from poor initial imputs.

#2 is evidence of GIGO Principle, and would seem to indicate that the missing information cited in #1 would be the same, if avialable.

I recommend starting with simple geometry - fully constraining all geometry and then do it over from scratch using what you learned from the first attempt (Steve Wozniak's book iWoz is a good guide).  Walt Jaquith says something similar here - Post #8

I recommend that you go through these before doing any othe work

http://home.pct.edu/~jmather/SkillsUSA%20University.pdf
http://inventortrenches.blogspot.com/p/inventor-tutorials.html
http://wikihelp.autodesk.com/enu?adskContextId=HELP_TUTORIALS&language=ENU&release=2014&product=Inve...

Then post each file you do for a while to get input on how it might have been done better.

[MACKTEK]

Ok, it will take some time to go thru all of the suggested areas. (Lots of links there. I did go thru 9 pages of the forum link you posted, and visit all the others)

I think the high # for the dimensions must be because I tried so many different ways to prevent the corruption of the model, and the many iterations I went thru. Each time I copied the Arch sketch to another fresh sketch, I think Inventor iterates the #. Other than that, I can't really explain why the dimension number is so high.

So, in your opinion, constraining the sketch properly seems to be critical to preventing this type of error. (And this is the essential problem?)

You mentioned that "everything" in the file is poorly done.

I get GIGO...Other than the constraint issue, what else would you say is a poor practice? 

Thanks again.

[JDMather]

---

@MACKTEK wrote:
 So, in your opinion, constraining the sketch properly seems to be critical to preventing this type of error.

---

No, not really.  There are plenty of CAD softwares that don't even have constraints/parametric dimensions.

But there are "best practice" techniques that are learned over time, and in my experience those who are the fastest and create the most robust models always fully constrain their sketches and don't create a lot (or any) extraneous geometry on the road to a solution.  (this doesn't mean they don't throw out a file and start from scratch)

Your unconstrained sketches indicate the difficulty you are having generating the desired geometry.  I don't really understand your design intent.  It would be even more difficult to try to figure out your design intent without the exisiting sketches - kind of like someone handing over the final answer to a math question and asking, "Is my answer right?"  The step-by-step work to achieve the answer might be more important than the final number.  Those steps might be a general process to solve many problems.

I will try to take a more in-depth look and offer some constructive examples later today.
Any additional information you could supply on design intent would be helpful.

[MACKTEK]

I am trying to model a hexagonal dome structure for a 3d virtual reality engine. (Which probably is an affront to your mechanical side... but I think its a great tool for me) Since I started using Inventor 2008 for CAD a long time ago, I prefer the engineering aspect of it, and the exactness of the dimensions, which is why I chose to model this in Inventor.

However, because its "virtual" it does not need to be made to exacting specification. Its mostly designed for "looks" and it needed to follow some simple rules:

1. It needs to be dome shaped, and be relatively low poly.

2. It needs to have a gap to fit a virtual sheet of glass.

The "Gap" is where that modeling corruption always seems to occur. However, sometimes the corruption does not show up until later, after some "split" ops.

So, the idea is to make a Hexagonal Dome, with a Space for the glass. Optimally, I would have liked to figure out a way to do this by laying out Hexagons on the surface of the Arch / Hemisphere and extruding the pattern such that the center of each hexagon was always perpendicular to the center of the sphere and so that each hexagon lay tangent to the curved surface...

BUT, I cannot seem to find any tutorials on how to place a sketch on a curved surface in that way... nor any tutorials to lay a pattern on a sphere or hemi-sphere in that way. So, I DID find tutorials to lay circle pattern on a cylinder, but that does not work for a sphere.

In an attempt to get "close" to the look and feel of a hexagonal dome, I resolved to just Extruding the Hexagonal Grid, and Revolve by Intersect to exclude everything above and below the arch sketch (and also the "gap" for the glass. That of course, led to the corruption of the model.

I have looked at the replacement provided by conklinjm (Very helpful). I have to figure out how to achieve the same thing (adapt my method) and try to get the sketch to be properly constrained. And hope that will fix the problem after I re-orient the axis.

[-niels-]

This is not related to your problem, but after seeing your shape and reading what you're using it for i wanted to share it anyway.

It's just a bit of messing around, seeing how unfold/refold would handle such a pattern.

The result surprised me, it looks pretty nice, but it's not what i'd hoped would happen.

Hope you find it just as interesting and that you find the solution to your posted problem.

[conklinjm]

---

@MACKTEK wrote:
However, I can't really explain what specifically you are doing that prevents the corruption.

---

I took another look at both your version and mine and found some interesting things ...

1) If I redefine the sketch planes in my model to match yours, mine too becomes corrupted.

2) If I redefine the sketch planes in your model to match mine, yours too becomes uncorrupted.

3) If I redefine/create the extrusion in two opperations (extruding a solid hex before the revolution and cutting out the center after the revolution), it still becomes corrupted.

4) Exported the corrupted model out as an .sat file, importing it back into Inventor still showed the corruption _but_ importing it into SolidWorks didn't.

This leads me to the conclusion that perhaps it is a modeling engine/graphics display problem.

HTH

[JavaLodge]

I don't have a lot of experience in game design, but I do have a fair amount of experience in 3D graphics in general, which includes creation of game assets.  I can tell you that this part would not be low poly at all.  This is the kind of thing that would be 'faked' so to speak in a real time graphics engine.  Is there not some way that you can use a dome object with a hexagonal texture mapped to it that controls the transparency?  I could whip something up pretty quickly in Blender 3D that demonstrates this technique.  It's your call in the end, but I wouldn't use Inventor to model this part specifically.  It does seem like an interesting idea to use Inventor to model some of the more mechanical things in games are art, though.

[MACKTEK]

---

@-niels- wrote:

This is not related to your problem, but after seeing your shape and reading what you're using it for i wanted to share it anyway.

It's just a bit of messing around, seeing how unfold/refold would handle such a pattern.

The result surprised me, it looks pretty nice, but it's not what i'd hoped would happen.

Hope you find it just as interesting and that you find the solution to your posted problem.

---

Wow, great job.  Yea, this technique will work great for me!

Thank you very much.

I am not sure if I can do it in 2008, but I will give it a try.

[MACKTEK]

---

@JavaLodge wrote:

I don't have a lot of experience in game design, but I do have a fair amount of experience in 3D graphics in general, which includes creation of game assets.  I can tell you that this part would not be low poly at all.  This is the kind of thing that would be 'faked' so to speak in a real time graphics engine...

---

Agreed, The engine we are using can handle quite a lot of verts but it does not have tesselation.  This particular model is around 20K verts which is on the high end for our models. We can bake this model onto a very low poly model and still get quite a decent visual especially because the player will not be very close to it.

[MACKTEK]

---

@conklinjm wrote:

---

@MACKTEK wrote:
However, I can't really explain what specifically you are doing that prevents the corruption.

---

I took another look at both your version and mine and found some interesting things ...

1) If I redefine the sketch planes in my model to match yours, mine too becomes corrupted.

2) If I redefine the sketch planes in your model to match mine, yours too becomes uncorrupted.

3) If I redefine/create the extrusion in two opperations (extruding a solid hex before the revolution and cutting out the center after the revolution), it still becomes corrupted.

4) Exported the corrupted model out as an .sat file, importing it back into Inventor still showed the corruption _but_ importing it into SolidWorks didn't.

This leads me to the conclusion that perhaps it is a modeling engine/graphics display problem.

HTH

---

Well, that certainly DOES suggest that the problem is with Inventor's engine.  How do I get this to an official Autodesk person so they can possibly fix it?

Ps, (I will still attempt to improve my skills by constraining my sketches better)

[JDMather]

Look into doing that as multi-body solids rather than disjointed solids.

[MACKTEK]

I just realized that I can't do these sheet metal ops in 2008, it does not even have them available!

Argh... I hope 2015 adds the ability to place custom tool bard on both the TOP and SIDES at the same time.

[MACKTEK]

Kudos to Niels btw for a very cool Sheet Metal alternative.