I am running into problems that I'm sure are just due to my being a novice at this type of modeling. Any help understanding the concepts would be very much appreciated, thanks.

1) I am modeling a knife blade, just getting a feel for the workflow of building multiple surfaces to create a complex smooth and curvy surface. In the image below I have built a blade using multiple surfaces. The blade's sides have been joined together and now I want to join the top to it and create some small fillets. Am I going about the wrong? The Top will not join with the sides, and even if it did I suspect I would have problems with the fillets because of the triangle shaped surface at the tip.




2) When viewing my blade I noticed that the surfaces do not match up nicely to the curves I used to create them. The surfaces were generated using sweeps. I used as few points as possible for the rails and the same number of points for each rail. I was under the impression that fewer points would create a smoother surface. Do I actually need to add more points to create a smoother edge here?

Hi Joe, can you please post the model so I can take a closer look at it?

Your feeling of potential trouble with the fillet is pretty correct, the more edges running into each other at a point increases the difficulty of the fillet and it can fail to calculate a result in these situations.

In some cases you can get a better result for a complex area by filleting the corners of the initial curves that you are sweeping instead of trying to fillet the results of the sweep.

But joining should work unless the surfaces involved do not touch each other close enough.


For #2 your surfaces are probably fine there, if you post them I will double check them though. What you are seeing there is the display mesh. To display a surface, MoI will break it up into a triangle mesh and what you are seeing here are those triangles. You can increase the display mesh density under Options, but the default density is already fairly high so that is probably not a good idea. But try checking the "add detail to inflections" checkbox - that will boost triangulation in specific areas to help give a smoother result, but that probably won't help out your particular example shown though.

- Michael

Yes you were right, when I zoomed in closely the points were touching, but the curves had a gap in them. I was expecting the command to work like Maya's "stitch" and merge the edges together and didn't realize how much of a role the distance between edges played in determining if they would join. I suspect when you get to the documentation aspect there may be some feedback added into the Command Options area when commands fail.

In the end the gap helped me out because I just blended the surfaces together which gave me the effect I was after, as well as opened my mind to and additional technique.

Thanks for pointing out that #2 is just the way the surfaces are displayed. So now I know to trust the curves for the best representation of the surfaces I'm creating.

Just in case I am attaching the 3dm of the model in the images.

Hi Michael,

Having had my share of filleting frustration
in Rhino, I learned, (as you suggest here) that
it's often easier to build soft edges right into the
sweep by drawing construction curves that have soft corners..

When I saw Joe's model, I was going to ask if
you planned to put "sweep to a point" in MoI,
but as I've discovered, it does it already!

Also, I never knew what the " pointy end "
option in MoI's sweep was for...MoI continues
to impress me!

-Jesse

Thanks for posting that model! It's amazing to me how such a simple solution exists. My fault for following a Maya tutorial where the modeler decided to model each side separately then stitch them all together...

I'm getting better results now, but am still running into problems getting these two surfaces to join, sides and the top. Even thought I used the same rails for the sweeps there is a slight gap that must be preventing the join. Since I used the same rails, I'm not sure how to get the curves closer together to lessen the gap.

Hi Joe, I think I see what you're doing there - I was a little confused at first, but now I see that you've got one set of rails slightly to the inside of the side surfaces edges.

Don't use those rails - instead use the actual edge curves of the surface as the rail curves for your top sweep piece. That's what I did for the attached version, which will now join.

It's good idea in general to set up the rails so that they are attached right to the end point of a profile, that way when you sweep using those rails the resulting surface will coast right along that rail edge.

It is ok to use rails that are not exactly touching the endpoints of the profiles for certain situations, but having them exactly touching will guarantee more accuracy. When the endpoint is off of the rail it will cause the ends to wiggle just a slight amount more as the profile is swept along the rails. As you have seen it doesn't take very much wiggling to stop things from joining.

Here you are seeing some of the finicky parts of NURBS modeling which can take some time to get used to - it is sort of a package deal - NURBS modeling provides for more accuracy in a lot of things (like being able to cut accurate holes in things), but also it sort of depends on accurate results for stitching parts together as well.

- Michael

> Thanks for pointing out that #2 is just the way the surfaces are
> displayed. So now I know to trust the curves for the best representation
> of the surfaces I'm creating.

Yeah, the curves are drawn to a high degree of accuracy, they are dynamically adjusted to fit smoothly to your current view. So that's exactly right, you can really trust the curve display (including edge curves) a lot.

But this is prohibitively expensive to do for surface shading, so surfaces get just one static display mesh that is used at all zoom levels. This causes these kinds of artifacts which you kind of learn to ignore over time.

However, sometimes there are mesh shading artifacts that can be an indication of problems, this tends to be sort of undulations in the shading in an area that you thought should be nice and smooth, sometimes this will show you that there are unwanted ripples in a surface. But this type of artifact looks different than the jaggedy edge type.

Later on you can export your model using a high mesh density to get rid of these jagged edges in your final output model. In fact it is probably not a bad idea to every once in a while do a test mesh export, and set the mesh export display to shaded, so you can view what your higher density final mesh is going to look like and if it looks nice and clean or not.

One other note - it is good to make sure surfaces join together because joined surfaces makes for shared edges between surfaces, and shared edges between surfaces are necessary to get "watertight" matching meshing for each different surface along that common edge. If you try to export separate non-joined surfaces they can have slightly different mesh structures along their sides which causes little tiny cracks between them. But shared, joined edges have extra processing done on them to avoid this.

- Michael

Doh! Yeah I have actually done what you suggest on a few models, but forgot to try it this time. It makes complete sense to me. These "gotchas" don't ever frustrate me anymore, I'm used to it after learning polygon edgeloop modeling for animation deformation. NURBS just never felt like a creative process to me before MoI so I avoided them like the plague, but right now I am having a lot of fun learning how to use them.

Thanks for taking time out to help me with these basic things, I know you must be busy on the next beta, which I can't wait to get my hands on!

> it is good to make sure surfaces join together because joined surfaces makes for shared edges between surfaces, and shared edges between surfaces are necessary to get "watertight" matching meshing for each different surface along that common edge.

Yeah I am used to this concept from learning low-poly modeling during a game prototyping course. We had to keep things "watertight" to avoid getting visible seams during real-time rendering. It has kind of stuck with me and has carried over to MoI, which is why I keep trying to make sure all the surfaces will join together. Thanks for all the tips.

> Thanks for taking time out to help me with these basic things,

No problem, helping people out also helps me to understand which things could use more tune-ups in the future.

I also don't mind especially spending time answering questions here in the forum, because it makes it easier for others to read it and benefit from it as well.


> I know you must be busy on the next beta, which I can't wait to get my hands on!

Well, it's been going a little slow so far, nothing really spectacular set up yet, I've just been nailing down a couple of bugs. But it looks like I've tracked down the one really big major remaining crash bug that I know of, so that is good.

- Michael

Hi Jesse,

> When I saw Joe's model, I was going to ask if
> you planned to put "sweep to a point" in MoI,
> but as I've discovered, it does it already!
>
> Also, I never knew what the " pointy end "
> option in MoI's sweep was for...MoI continues
> to impress me!

:) Just to clarify for others - to do a 2-rail sweep to a point in MoI you just make the rails end in a point.

To do make a 1-rail sweep to a point you use the "pointy end" or "pointy start" options.

- Michael

Sometimes moi just refuses to join 2 surfaces for a couple of reasons.

1) Try saving and then reloading the mesh. Often joins/unions/etc will work after this

2) Examine the profile curves you used for the sweep, and check for overlapping points ( Click show points ). Overlapping points will show up as a red box. Just select and delete one. These overlapping points will generate sweep curves just fine, but cause problems with joining the resulting surfaces if they occur at the edge which you are trying to join another curve to.

Perhaps moi should pop up a notice when sweeping a curve containing overlapping points? This has bitten me a few times. You can join the swept curve to other swept curves at the edges that does not contain a overlapping point.

Also, I notice trimming circles/ellipses close to the quad points leads to overlapping points in a lot of cases.

Hi Joe,

I hope my model wasn't confusing...I drew it by eye before you posted your 3dm model, so mine is a rather simplified version
in comparison.

I did the whole thing all at once as a 2 rail sweep, but because the long side of your top surface has a shape
that is uniquely particular to both sides of that surface along it's length, you might get more predictable results by sweeping the surfaces separately,
as you have done....if you swept the top surface first and then used the edge curve of the top surface as one of the drive rails of the side
surface, the two surfaces will join without gaps.. (does that make any sense)?


Michael explained it much more clearly that I am...:-) , using commonly shared drive curves in a sweep will better insure that the resulting surfaces
join up into a closed polysurface without any gaps.



-Jesse

Yeah your model helped because I realized I should include the rounded edge in my profile curve instead of trying a fillet. Thanks to help I got from you guys I'm starting get a hang of the basics. After I got the blade together I was able to put together a guard and handle very quickly. Tonight I plan to start on a more detailed knife.

I finally got some time to play with MoI more and put this sword together tonight. I am really loving the "sketchy" workflow. I didn't even draw any concept art, just played around with shapes. Now that I am starting to understand the toolset better I am getting really excited about this app. I was able to get my model into Max and throw on a couple of shaders for a quick render.

Added to the special thread Gallery :)

Hey thanks!

Very nice!

Nice job!

-jdk-