Question for Modelers and Sculpters - Booleans

So I've been working on upping my skills with modeling of late.  I've read many things here and elsewhere on the topic of using Boolean operations.  The prevailing opinion is always don't use 'em or the world might stop orbiting the sun and thousands of crickets will take up residence in your bedroom.

I get it... I think.  I haven't run in to a circumstance yet where only a Boolean would solve whatever it is I'm trying to do.  However, I am curious of what are the actual dangers of using a Boolean to quickly achieve the structure you're looking for?  I'm assuming the problem would be the operation creating some invalid geometry, or something along those lines.  If I were forced to use Boolean for some unknown reason, what would I need to look out for after applying it?

I know there are other sites out there where this question would probably reach a larger number of people who could answer, but I figured I'd start here.

  

Comments

  • grinch2901grinch2901 Posts: 1,247
    edited March 2016

    Booleans tend to leave n-gons, polygons with more than 3 or 4 vertices. Many programs can't handle them; most renders love quads (polygons with four vertices) and/or triangles and anything else is a crap shoot.  If you always use the same rendering engine and you don't plan to share a mesh with others AND that renderer cand handle them (test it) then you can get away with it. Otherwise the mesh isn't necessarily portable between render engines and may not work properly.  Also tends to leave weird artifacts like dark shaded areas where they aren't in shadow.  Sometimes you can get rid of that by adjusting the smoothing angle in the surfaces tab but sometimes not becaus the geomety is odd there. 

    One thing to do is go ahead and try a boolean and look at the resulting geometry, it might  be possible to add some edges here and there and clean it up into all quads. If it's too off the wall though then maybe reconsider using another technique.

    Post edited by grinch2901 on
  • KurzonDax said:

    The prevailing opinion is always don't use 'em or the world might stop orbiting the sun and thousands of crickets will take up residence in your bedroom.

     

    +1

    Seriously though, the resulting hideous topology may or may not be a problem, but the infinitely sharp edges and corners resulting from boolian operations are generally unacceptable to me, and for organic shapes never. I think the time taken in solving these problems would be better spent in finding an alternative method. I simply never consider boolean operations as a solution in modelling, but I guess it may depend on what you are working on—for example it may not be a noticeable problem in small details.

  • Oso3DOso3D Posts: 15,095

    Note: I am very much a beginner and may be utterly wrong...

    My impression is that if you don't care about where polygons end up, it's not a terrible idea to do Booleans and then run it through Instant Meshes. This will nicely, cleanly retopo your surface into neat, even quads or triangles (or both).

    That said, I'm also seeing some reasons you might not want to retopo, because sometimes you want flow of polygons to work a particular way.

     

  • Jan19Jan19 Posts: 1,109

    Hi, ZBrush will do ok Boleans.  You can then re-mesh or retopologize.

  • mjc1016mjc1016 Posts: 15,001
    edited March 2016

    Usually, Booleans work in the program they are created it...in the native format for it.  It's the translating them to another format that really messes things up.  And it isn't so much Studio hates them, as it's the obj format that does.  And since that is the most common way to bring modeled items into Studio it really is a mess.

    I haven't tried to see if millighost's Blender dsf exporter handles them any better...

    Post edited by mjc1016 on
  • KurzonDaxKurzonDax Posts: 228

    Thanks, all.  I don't have a specific need to use one on anything I'm working on at the moment.  So far (in my admittedly brief experience) I've found better ways to accomplish things, usually by breaking up the geometry in to more logical pieces or something along those lines.  I was just curious what drove the the 'booleans are bad' philosophy.

    Now if someone can give me a silver bullet for creating perfect UV maps, I'd owe that person my deep, dark soul.

  • mjc1016mjc1016 Posts: 15,001
    KurzonDax said:

     

    Now if someone can give me a silver bullet for creating perfect UV maps, I'd owe that person my deep, dark soul.

    Practice, practice...and more practice.

    Blender has decent mapping tools, live updates to the maps and a several choces on how to map (various projections, etc).  There are options to each one...some things like island spacing are common, but every one has something unique to it.

    The rest is practice until your fingers fall off and your eye implode (well, maybe not that much...).

  • Jan19Jan19 Posts: 1,109
    KurzonDax said:

     

    Now if someone can give me a silver bullet for creating perfect UV maps, I'd owe that person my deep, dark soul.

    UV Layout. :-)  It'll do almost perfect UV maps. 

     

  • Perilous7Perilous7 Posts: 40

    Using booleans are fine, you just have to have a mind of the project you are doing beforehand, they nearly always cause you to do some sort of clean up of triangles to quads etc moving around of the boolean objects to get cleaner cuts etc , if using it is quicker than you doing the job manually then its all good. a technique ive found is that if you duplicate your object,boole it then use that object as a "cutting guide" for you to do manually (how easy this will be depending on your 3d prog), this also teaches you about following the surface as you do the cuts too :) maybe get to the point where you dont need to use booles so much :) Jobs a good un

  • BejaymacBejaymac Posts: 1,965

    It's not so much that they create n-gons as both DS and Poser can handle them, the problem is that Boolean usually leaves concave polygons (see attached pic) and many 3D engines can't deal with those.

    You have a quad poly mesh and you use Boolean on it to cut a hole in that mesh, if you look at my pic that polygon could be one of the corners of your hole with points 1,3 & 4 being the corner, now a lot of engines will fill that area in with black or white artifacts leaving you with little choice but to triangulate the entire mesh to get rid of the artifacts.

    In DS you also have the problem that concave poly can also cause the dreaded black polygon where the normals have been broken due to the internal triangulation order of the vertex, if DS uses 1, 2 & 4 and then 2, 3 & 4 then everything works fine, but if it triangulates it as 1, 2 & 3 you get broken normals and a black polygon.

    Concave.jpg
    500 x 500 - 62K
  • ValandarValandar Posts: 1,418

    If I ever even CONSIDER booleans, I run it through ZBrush's Dynamesh after, then ZRemesher for retopo. Then kick myself for thinking "the easy way out" is going to result in any form of useable geometry.

    Another problem with booleans, though not so much with modern systems, is you have absolutely no control over the polygons on the edges. You're almost always better off simply extruding an addition out or a subtraction in, then bevelling the resulting border.

  • Kendall SearsKendall Sears Posts: 2,995

    It really depends on what you're modeling for.  Using booleans for shell modeling is usually a bad idea as the process can leave horrible artifacts.  Using booleans for functional modeling (simulations) or solid modeling (3D printing, milling) can be just fine since it isn't just the outside "skin" that is affected.

    As with most things, sweeping "rules" tend to be hyperbole to keep the noobs from making critical mistakes early.

    Kendall

  • hphoenixhphoenix Posts: 1,335

    The biggest problems with using booleans in a mesh modeller (voxel modellers don't have this issue, as they re-topologize the surfaces from the voxels) are as follows:

     

    1)  Creation of n-gons.  These aren't handled well by most GPUs, as those typically use triangles (most drivers simply convert polys and quads to tris.)   This results in the next issue:

    2)  Poor triangulization of n-gons (triangulation algorithms are pretty simplistic in many modellers, and often just pick one vertex and then cut the n-gon to every other vertex except the already connected edges.)  This can make for some ugly shading, and often very ill-arranged topology.  Which also often leads to the next issue:

    3)  Very thin or degenerate triangles.  These cause issues for almost all objects.

     

    Put it all together and Booleans (except in the most simple cases) usually result in some seriously messed up geometry.  Cleaning it up is a lot more work (usually) than just modelling it without the Booleans.

    Now, IF you have good Decimation or Re-topology software (or built-in to the modeller) you can take those booleans, run some subdivisions, decimate, and re-topo and often you'll get much better geometry that requires a LOT less clean up.  Sometimes, depending on the resulting geometry, you'll have to run the sequence in different order.  Takes some experience to know beforehand....so save your work (to different file names) throughout, so if something doesn't look like it worked right, you can go back and try it the other way.....

    Cleaning up Boolean operations on meshes is slow and tricky even for experienced polygonal modellers......

     

  • StratDragonStratDragon Posts: 3,278
    edited March 2016
    hphoenix said:

     

    I recall desperately trying to use them starting out but when i did there was a lot of post modeling needed to clean up the mess they made.  Most modelers now have the tools necessary for avoiding them. Booleans are from another age where there was no other way. The prefered alternitve  to booleans by most modelers is to simply model correctly the first time.

    While they have their place in some situations they can more likely cause problems further down the pipeline if you aproach modeling with them as necessity. 

    Post edited by StratDragon on
  • Oso3DOso3D Posts: 15,095

    I used to use 3d Booleans in Carrara/RDS back when the idea of vertex modeling scared me. And there are quick, basic structures that work fine -- I made some groin vaults using either extrusions or 3D booleans. It's also not so bad, UV mapping-wise, if you are using procedural shaders or other stuff where UV maps are irrelevant.

     

  • KurzonDaxKurzonDax Posts: 228

    All great info here.  I really appreciate everyone's input.  

    @Bejaymac Great explanation of the concave poly

     

Sign In or Register to comment.