Fan Cage

This object was the result of playing around with techniques for creating offset
spline based radial symmetry. I had no intention of making a "fan cage".

fancage2.gif


Fan Cage unsub.jpg
 
Thank you Frank, this one was fun.

I was able to get it all in under one Boolean Creator "parent"
with two Boolean Modifiers to pre-cut the pieces.

If only the Ring Modifier had the snap/weld function like the Symmetry Modifier,
I could get to the final object without making the stack editable.

Fan Cage setup.jpg
 
Part of my last statement is wrong, even if the Ring Modifier had
the snap/weld function, the Boolean Creator also creates a Seam.

The Seam will prevent a Subdivision Modifier from smoothing the joint.
So the Boolean Tools would need the added function of turning seams off.
 
Two crazy pilots are doing a plane swap right now, two planes they jump and swap.

One made it and landed safely the other parachuted away.
 
Last edited:
Wow Frank, good eye.

After making it editable I needed to increase the Optimize weld radius.

I'm impressed at how well Cheetah handled that n-gon.

fancage_ngon_gone.gif
 
I'm impressed at how well Cheetah handled that n-gon.

"That n-gon"? In the first pic I spotted 3, in the corrected you can see that there were more on the downside. Catmull-Clark handles these things really well (the same for the 6 star poles I myself usually would try to avoid, but here they do not cause any harm. As I didn't want to nitpick again because of some ngons, I didn't comment at all at first

The form itself is really pleasing, and to create it this way is rather ingenious (I am still not totally sold on booleans for subd-modeling even if I do it from time to time).
 
"That n-gon"? In the first pic I spotted 3, in the corrected you can see that there were more on the downside. Catmull-Clark handles these things really well (the same for the 6 star poles I myself usually would try to avoid, but here they do not cause any harm. As I didn't want to nitpick again because of some ngons, I didn't comment at all at first

The form itself is really pleasing, and to create it this way is rather ingenious (I am still not totally sold on booleans for subd-modeling even if I do it from time to time).
You have to show me where you saw that. I corrected the original
with an Optimize command, so I can't go back to find the mistake.

In the corrected version there are only n-gons on the tube ends that I filled and creased.

This mistake is a product of Cheetah's handling of Booleans where two
points are created very close together where there should be only one.
 
I do understand you're reluctance to use Booleans but in many instances
they can be used simply as a drill or chopsaw with little problems.

I keep changing and refining my technique and to be honest I was very lucky with
the first gang Boolean operation but I was not so lucky with subsequent attempts.

Now I work on making one element and then Ring Modify it.
Here's the one element used after I Boolean mitered and drilled it.

Saw Drill.jpg
 
After deleting the unneeded polygons, and using the select
n-gon command, Cheetah shows there are n-gons.
Even after optimizing at .001, they still are there.

Next I'll optimize at .01.

ngondetction.jpg
 
After optimizing at .01, there are no n-gons now.

Running the test again, I get no n-gons after optimizing at .001. 🤪

I will be using the Select n-gons command more often to check for hidden n-gons.
 
Maybe you counted the ngons as one (Frank certainly did because I'd bet on it that he saw all of them). They are visible at 2 other connections if you start from the one Frank pointed out. The connection above seems correct, the others may or may have not that same ngon. If you look at the second picture you posted, you can see from the moved edges, that you probably had the same problem with some points on the downside. But it doesn't matter in this case, and to be honest, I wasn't even sure if they were not intentional (sometimes a model looks perfectly correct with such a ngon, but doesn't, when it is corrected).

About modeling with booleans in general:

For poly-modeling (non-subd) I use it all the time. For me, that's where it shines.

For subd-modeling I use them from time to time, a bit more drill or stencil (you just "imprint" the conncection from one mesh to the other. It's easier to clean up and can then be bridged or beveled in), always when I do not have a "better" method at hand. For me it's all about being faster, which isn't just a question of technique, but also of exercise, personal preferences, the way you're wired, your knowledge, the tools your software offers. Mostly it's just habit. In "production mode" I try not to think about what I do, I just do till I run into a problem and actually have to think about the best way of solving it.

I also have a "learning mode" for which I look at lots of tutorials, from beginner to professional, just to see how others tackle a problem. And in "experimenting mode" I try to find the methods that are best suited to me, like doing a model (or a difficult part of it) several times over with different methods, often just to see with how many of them I can come up with. In exercise mode I try to incorporate some new learned or found methods with doing the same thing several times over the same way. And you doing your boolean bit certainly made me try out such things a bit more, but very often I just know other ways I'll always be faster.

But do not get me wrong, I find it always interesting how YOU do things. And I'm not above it to steal any method that's fitting me ;)
 
Maybe you counted the ngons as one

I also have a "learning mode" for which I look at lots of tutorials, from beginner to professional, just to see how others tackle a problem. And in "experimenting mode" I try to find the methods that are best suited to me, like doing a model (or a difficult part of it) several times over with different methods, often just to see with how many of them I can come up with. In exercise mode I try to incorporate some new learned or found methods with doing the same thing several times over the same way. And you doing your boolean bit certainly made me try out such things a bit more, but very often I just know other ways I'll always be faster.
Actually it's 4 n-gons repeated radially for a total of 24. ;)

That is word for word what I do. Since I'm retired the learning mode has
been my way, and it's great to have the time to experiment and play.

Here's the thing, forget about sub-d for a minute. I have one eight sided cylinder joining another of the same diameter.
All points are snapped precisely to the raster grid to create quads, but Cheetah fails to weld the top and bottom
points and you end up with the n-gons with no sub dividing.

So the notion of it "shining" with non sub d techniques doesn't shine too bright for me, at least with Cheetah.
 
One more time, I don't think I can pare this down any more than this.
I need one tube with a double mitered end and another with a straight mitered end to be the "drill".

This was one I had thought of trying but I didn't think it would work.
The Boolean Creator only affects the main object and the Modifier affects all produced by the Ring Modifier.

I still have to "drill" the hole.

steelheart.gif
 
Here's the thing, forget about sub-d for a minute. I have one eight sided cylinder joining another of the same diameter.
All points are snapped precisely to the raster grid to create quads, but Cheetah fails to weld the top and bottom
points and you end up with the n-gons with no sub dividing.

So the notion of it "shining" with non sub d techniques doesn't shine too bright for me, at least with Cheetah.

The one command I used in every 3d app the most was some automatic cleanup like optimize because in every app I ever tried sometimes a single vert or two on each other as well as two point polys etc. can originate, most of the time somewhat inexplicable. So I do not see that as a problem in Cheetah or anywhere else. If I had worked on something like this with your methods, I'm not even sure if I would ever have seen the ngons, because optimizing is an ingrained habit for me.

With that automatic cleanup taken into account, the rest doesn't matter much with booleans in simple poly-modeling. I do not care about tris, ngons, poles, whatever and only have to do some cleanup-work on some edges I want to bevel, maybe split up a big ngon that somewhat can break a reflection etc. Like you I do use rather "proper" meshes as parts of the operation and most of the time not much of a problem arises.

More complex shapes I (subd-)box-model anyway most of the time (with some exceptions not possible in cheetah, where you can actually use a special kind of booleans created for subd-modeling. The resulting mesh is not always ideal topology-wise (it's not really meant for further subdiv), but it's fast (and feels like cheating).
 
Back
Top