Please help my Constitution Class

Posts

• Lee80193 Posts: 458Member

  May 2009 #2

  

  One thing to remember about using booleans in max is that it can really screw up your geometry, but it can be fixed if you take the time. Im guessing that when you chamfer the edges that you get a lot of weird looking artifacts... you probably have to many loose verts in there... The other thing I noticed was that the edges around the boolean are have been welded. When you clean up your mesh after a boolean you really want to try and create quad sided polys, not tris... Another thing to look at is that your vertical edges line up, if they don't you can get some artifacts... see attached image for some ideas on how to clean up your mesh... The big thing will be removing the extra verts caused by the boolean. Good Luck. hope this helps
  

  my_crap_01.jpg 79.9K
• Jedilaw0 Posts: 0Member

  June 2009 #3

  MadKoiFish had a good tutorial on this topic a while back, but the link appears to be dead.
• Berkut1 Posts: 0Member

  June 2009 #4

  I can't speak for vektor but it looks like he may have used subdivision modelling there to achieve that look
• Vektor332 Posts: 0Member

  July 2009 #5

  

  Sorry I didn't notice this one earlier. Hope it's not too late to offer some useful advice.
  
  I actually did not use subdivision modeling. All of the shapes at the back end of the secondary hull were created with Boolean operations. The trick is to create an edge loop or edge segment (sometimes called "control edges") to act as a buffer between the mismatched geometry of the two objects. This is a wireframe screen grab from my modified version of the Enterprise to show what I'm talking about:
  
  
  
  When you Boolean subtract one object from another, a lot of "damage" results from the imperfect way in which the underlying geometry of the two objects intersects, usually in the form of extra vertices and edges that create long, skinny triangles. This damage propagates outward until it hits an edge that wasn't affected by the Boolean operation. The further away those unaffected edges are, the longer and skinnier the "damaged" triangles tend to be, resulting in smoothing errors. The best way to minimize this is with "control edges" as shown by the red and green lines in the above screen grab.
  
  What we're mainly concerned with is the edge shown in blue, which is what we were trying to get from the Boolean operation in the first place. The goal here is to keep the vertices along that edge as smooth and regularly spaced as possible. Unfortunately, the immediate result of the Boolean operation is a mess of irregular vertices and lots of long, skinny triangles (whether or not the edges of those triangles are visible is irrelevant, they’re still bad news). By cutting the red and green control edges close to the blue edge, we can keep those triangles relatively short and prevent them from causing smoothing errors. This can be done either before or after the Boolean operation, but it’s probably more reliable and requires less clean-up if done before.
  
  Next, we need to deal with the vertices along the blue edge. It’s a combination of vertices from both of the original objects. Let’s call them Objects A and B, A being the lathed secondary hull and B being the extruded shape that was Boolean subtracted from it. At least half of those vertices are extraneous and should be eliminated, but which ones?
  
  Well, next to the red highlighted control edge, the vertices from Object A (on the blue line) are too far apart to give us a nice smooth curve without visible segmentation, so we go in and manually weld each of those vertices to the nearest vertex from Object B. This leaves us with only the Object B vertices, which are nice and regularly spaced and close enough together to appear smooth. This is where the control edge really comes into play as it isolates the effects of the vertex welds and keeps them from spreading out over the larger surface.
  
  Similarly, next to the green highlighted control edge, the vertices from Object B (on the blue line) start to get too far apart and suffer the same segmentation problem, so we weld those to the vertices from Object A. In this case, the control edge is not really needed as the vertices and edges of Object A never move, but terminating the control edge part way around can create its own smoothing problems, especially if the surface curves in more than one direction (which this one doesn’t), so we continue it all the way just to be safe.
  
  So, after the Boolean subtraction has been performed, how do you know which vertices came from Object A and which ones came from Object B? Sometimes you can tell by the way they are spaced or how they connect with the rest of the geometry, but the simplest way is to retain a copy of Object B and just look for the vertices that line up with its edges. All the others then belong to Object A.
  
  Sometimes, if one or more of a polygon’s vertices are not co-planar when it gets cut by a Boolean operation, you can wind up with extra vertices on the invisible edges that don’t appear to come from either object. As long as the polygon in question is not unusually large, these can almost always be eliminated by welding them to their nearest neighbor.
  
  That’s basically it. The same principles based on control edges can be applied to any Boolean operation where the curvature or angles or mesh densities of the two objects don’t match, as they virtually never do. The control edges let you isolate the mismatch and then you simply apply some careful vertex editing to get the best result.
  
  Hope that makes sense and isn’t too verbally convoluted. It would probably be easier to understand if I had more screen grabs showing each step of the process, but the modifier stacks on this model have long since been collapsed so I would have to do it with a whole new set of objects. Maybe I will some day. The whole topic of smoothing errors, why they occur and how to avoid them is something I have rarely seen explained elsewhere.
  
  Good luck!