I can use this no problem with two shapes that I have just created. But for some reason, when I try to sculpt a shape that has been in the project for a while (maybe after a save?) both objects disappear when using this filter.
It's frustrating the crap out of me.
Anyone see this before?
Jim
Dssapearing happens because:
the operation bulds a new network wich hides the oeprators, but because it gfails to produce anything you see nothing.
Now why does it fail? Well most likely your object is not etirely true. Problem wih a surface moeller is that its not chekking constantly to see that what you do is valid. However objects in REAL world behave in certain very real ways. Wich can quite easily be circumvented in 3d, by choice or by mistake. Now boolean works ONLY on objects that correspond to reality, the sort of imaginary logic that is possible in the software is not possible for a boolean. This is no the booleans fault per see, because theres no way for a simple either or logic to answer to a question with 3 maybes as a answer.
But all thsi is secondary concerns, result of a boolean operation on a poly surface is in general unworkable. You often hear people saying its because the boolean is unreliable, fine they don't use it mission accomplished, but thats far from the truth. But rather the problem is quite on contrary on the fact that booleans are quite literary not doing what you THINK they are doing. Thats because quite literary a polygon object isnt what you think it is (tough the more dense the mesh gets the closer it is to what you think it is so a 2 million poly mesh is already quite close, but still not there).
Now while i will tell you to avoid boolean theres a several things you CAN do to meka it work.
PS: do a forum search theres a lot of VERY detailed descriptions of why booleans are rpblematic. And i ust dont wnt to repeat myself on thet brach i think ive said all taht i will ever need to.
hahah joojaa, we should just have a "pinned" topic in every 3D forum on highend3D:
"Booleans. Thinking of using? THINK AGAIN! - PLEASE READ!"
Yes, it appears so. problem is that our entire school syetm is geared to make people think in boolean like manner. real world often don't work that way tho.
Well, solid modeling for engineering uses Booleans, doesn't it? There's nothing inherently wrong with the concept of Booleans. The problem here is twofold:
Newbies are not aware of the problems that can arise. For example, tessellation into long thin triangles produces sub-optimal results if the object is deformed.
Maya's Booleans are pretty bad. Most entertainment-oriented 3D apps have traditionally fallen down in this area, Maya is no different. But try Max's new Pro Booleans. It's much more stable and feature-rich than anything else I've seen. And why? Because it converts polymeshes to solid models and then back again. And like many of the cool features in Max, it started its life as a third party plugin.
Yes, its a question of polygons versus nurbs. Theres a 3rd problem!
Nothing even in power booleans that fixes this problem you still just dont want to use booleans, tough it is good attempt to fix the problem yes. (maya booleans indeed never fail if you have the patience to fix the geometry problems you have, ie make the solid check, this can be automated but i prefeer not to since it builds a avenue where the user is fooled into a even bigger trap than before). The question arises whether or not the booleans actually complete or not.
engineers use solid paradigm modelers, their models are enforced to not include non worldly geometry. But more importantly they use nurbs models. Theres is indeed nothing bad in using booleans on nurbs. Reason:
(Nurbs mathematically model the end result perfectly, polygons on the other hand do not, so herein lies the problem. A polygonal model is a construct that appears to have a shape it does not have. And the boolean works on the actual shape. A subdivision surface might be a candidate for accurate booleans, but then you'd loose all the control you now have)
Damn i have a nice picture of this but my web repository is down ill post it later.
THis is why polygonal surface modelers suck at booleans and theres nothing you can do about it. You can fix the surface automatically to let the boolean always happen. But thats just the start of the problem. And even in the advent of perfect boolean ability its not used, excpt on angular models (where polys indeed are what they seem).
PS: making it a solid and then back has certain possibly adverse effects on your model. Wich is why id keep the two opeartions separate. But i guess whatever works. As long as it dont cause your more trouble.
PPS: even the best boolean engine I've EVER seen the PTC pro|e wildfire fails occasionally.
OK,
Well as someone just learning this software, I don't pretend to understand everything that was said. I do get...DON'T USE BOOLEANS...got it.
Now I have to relearn some things...ugh.
Thanks for the help!
Jim
I'm a lowly student, too, but I saw an excellent demonstration of why not to use Booleans in a tutorial - you can see for yourself, if you look. Take a cube and a sphere and do a Boolean subtract, and then look at the mesh where they intersect. You'll have 3 sided polys, 5, 6, 7, all mixed up every which way, stretched, squashed, pulled. Then you can also get strange or hidden geometry inside the model, too, depending on what you do. It has no brain in any real sense, and is just making geometry, good or bad, to get the job done, brainlessly. And what Joojaa says:
Now while i will tell you to avoid boolean theres a several things you CAN do to meka it work.
booleans, are sensitive to the normals and nonmanifold geometry, so your normals should be completely consistent and mesh completely manifold.
holed geometry needs to be split up.
Sometimes rarely the order of a mesh, especally imported ones throw the boolean a conundrum that it cant solve, if thst the case just renumber the vertices, (detach a face and attach all others to that bit, then see below, theres a usefull tool fro this in the devkit that avoids this trick)
Points must actually be merged and there , may be no lamina faces or open edge connections.
Numeric precision problems jitter/quantitze object a bit<
I think is that you have to become the brain in order for it to work properly, in which case why not just model it?
Apart from the facts Joojaa already mentioned there is one last bug (or maybe it is a feature ) when booling in Maya:
Sometimes you need to scale your objects up by a given amount (x10, x100 ...) before booling and scale down afterwards. This is because Maya does not allow a Boolean operation to create polygons smaller than a given value, so if your polys are that small the boolean operation will not work.
Please correct me, if that has been fixed in the 2008 version, as I have yet only encountered it in the 7.0 version.
I havent encountered that but it certainly makes sense that scaling up or down would work on quantization errors since it would re quantize. However whenever you'd encounter this problem you allmost certainly don't want those very small faces at all because they are more harmful than of use. Ive fount though that renumbering vertices accomplishes the same thing mostly so that should take care of it (and certainly harmonizing local space).
The Boolean scale up / scale down trick has been around since day one of 3D computer graphics. Typically this is not needed much anymore, so long as the operands are not too small relative to the entire scene. If the operands are the only objects in the scene then calculation should be more accurate.