Leaky Boolean Operations in Rhino 4.x

[pyronaught]

Can anyone verify if Rhino 5 fixed the aggravating flaws in its Boolean operations that results leaky geometry and flawed STL conversions? I'm wasting tons of time redrawing the same model over and over different ways until I find the magical combination that doesn't result in non-manifold edges, but don't want to spend the money on version 5 if it is still just as screwed up as version 4. I've been spending 8 hours creating models that should take no more than 30 minutes due to these non-manifold edge problems. Sometimes it is the STL conversion breaking the models, other times the Boolean operations leave tiny holes and you get triangular planes shooting across the model. Sometimes the model looks perfectly fine in Rhino and then the triangle spillage only shows up once imported in to Simplify 3D.

[Jules]

Nope. Wait. It still does it.

[pyronaught]

I guess I'm stuck forever trying to discover the magic secret to the random ways of Rhino. I'm pretty sure the sun and the moon have to be in specific positions relative to each other for certain operations to work. I swear I've assembled the same model the same way using the same sequence and one time it works, the other time it doesn't. I did discover one trick for "patching" holes in spherical or elliptical objects where the leaks always occur at the axis points, with the triangle of death shooting from one axis to the other. If you embed a small rectangular plate at the axis on each end, it helps patch the hole by keeping a million tiny triangles from trying to all terminate at the same point.

[jimc]

naaaa. something isnt right. i have never used rhino 4 but in 5 i can say i have never had a non manifold edge.....ever. i have had naked edges but anytime i have stuff like that i can go back and attribute it to workflow or sloppy modeling on my part. also, if i have a model and when its done i do an edge analysis on it and it has no non manifold and no naked edges then when i export it, its a perfect watertight model. i have never once had rhino export to an stl from a good model and not had the stl be flawless. booleans dont always work. i have had that issue plenty of times. its usually due to something very tiny that i have to zoom way in on and catch, readjust the placement on my 2 parts and try again. if the boolean works then it should be a perfect mate. there shouldnt be any errors. just thinking, have you checked your tolerances in your settings? if your tolerances are really loose and your stl conversion is too tight then i can see that causing issues.

[pyronaught]

Maybe they fixed it then because in 4.0 there is definitely something buggy going on. I never had a problem until now, but these rounded hollow models are really giving me a lot of trouble. It is not even a complicated model, and I'm stepping through each Boolean operation one at a time and checking it in Simplify 3D to find out which step is breaking it. In the example below, it is the swept edge contour that is causing the leak once it gets merged with the shell containing the outer elliptical housing, floor and two inner spherical shells. The defect shows up better in Simplify 3D, as I can not even see that triangle shooting across the left opening in the Rhino model. I've tried every imaginable sequence of adding these parts together and it always screws up in a similar way. It just should not be this difficult. I've also built this exact same model in a different size using the same method and it worked fine, so it doesn't even consistently happen.

I could send you the unassembled model and see if you get the same glitch after welding the parts together as a way to see if there is a difference between Rhino 4 and 5. It would not even matter what order you join them together, you should get a similar defect no matter how you do it.

defect2.jpg (161.19 KiB) Viewed 10092 times

defect1.jpg (225.42 KiB) Viewed 10092 times

[jimc]

go ahead and email me the model so i can check it out. if you dont have my email then send me a pm. it would be the best way. i can tell you rhino 4 is old and even rhino 5 is approaching the end of its life so there have been 1000's of fixes over the years and just plain old improvements to the tools.

[pyronaught]

The absolute tolerance is .001mm and the stl export settings are below. Setting "Maximum distance, edge to surface" from .01 to .02 healed the holes in some cases, but not for this current model. The defect always occurs along the same axis as shown

stl_settings.jpg (129.19 KiB) Viewed 10077 times

In the STL preview here you can see the defect sprouting from the point on the inner sphere where all the triangles come together at the axis. This is always where these leaks occur. Note that it doesn't happen on the other sphere, which was created in the exact same way. Also note that the sweep edge isn't included in the model this time and yet it is still happening. This is the random nature of it.

defect3.jpg (271.68 KiB) Viewed 10077 times

[jimc]

yeah i can see it there. just email me the 3dm and ill give it a try.

[pyronaught]

So I exported only the parts needed to build that model to a separate .3dm file to send you, then closed Rhino and opened that file to test it myself and the damn thing works now. I assembled the whole thing and did a cross section check in print preview mode and there are no glitches. I still get a non-manifold edge error in S3D, but it at least prints now. Same exact model, different file. That's how crazy this is.

[jimc]

i have found that you can build many things but its how they are built thats important. not saying anything was wring with how you did it. maybe there is a bug in it. i can almost guarantee you will see less of these strange anomalies in the current rhino 5 version just because its so mature and its years and years ahead of what you are using currently. once in a blue moon i have invalid surfaces created by a tool or the big one that gives me trouble is the fillet tool. anyway if you have trouble with any of this stuff and want me to do a chance or see what i can find then just save the 3dm and email it to me.