Please help me understand why I can't bevel some edges with polybevel
6498 10 1- decksounds
- Member
- 109 posts
- Joined: March 2017
- Offline
I must be missing something here but I'm not sure why it works sometimes and not others. If I get non-manifold errors I will use a clean sop with manifold-only checked. Often, including in this example, I get the error “Skipping Boundary Edge 2344…” I've tried every option/combination of clean/polydoctor/fuse/facet as well as all of the polybevel settings and still no luck.
If I could just figure out how to make the polybevel work in all cases, it would greatly improve my Houdini modelling game but I get into these situations where it just won't work.
What is going on here in this particular case? The edge I want to apply polybevel to is highlighted in the second image.
Thanks for looking.
UPDATE: I think I solved the problem, please let me know if this is correct. The primitives of my base mesh were non-planar enough to throw off the bevel node, creating bad (degenerate?) faces. I think this was messing up some of my boolean operations as well.
UPDATE 2: Using the crease node to control the bevel amounts adds great flexibility, I was having a lot of trouble with overlaps until I figured this out.
If I could just figure out how to make the polybevel work in all cases, it would greatly improve my Houdini modelling game but I get into these situations where it just won't work.
What is going on here in this particular case? The edge I want to apply polybevel to is highlighted in the second image.
Thanks for looking.
UPDATE: I think I solved the problem, please let me know if this is correct. The primitives of my base mesh were non-planar enough to throw off the bevel node, creating bad (degenerate?) faces. I think this was messing up some of my boolean operations as well.
UPDATE 2: Using the crease node to control the bevel amounts adds great flexibility, I was having a lot of trouble with overlaps until I figured this out.
Edited by decksounds - Sept. 6, 2019 18:39:37
- decksounds
- Member
- 109 posts
- Joined: March 2017
- Offline
Maybe this is related but I notice this one division from the prior bevel in the boolean that created this edge, is DARKER than the other divisions.
What does that indicate? And why are there two normals there? I have “consolidate points” checked in the clean sop.
What does that indicate? And why are there two normals there? I have “consolidate points” checked in the clean sop.
Edited by decksounds - Sept. 6, 2019 02:48:36
- Andr
- Member
- 900 posts
- Joined: Feb. 2016
- Offline
- decksounds
- Member
- 109 posts
- Joined: March 2017
- Offline
I'm sorry, I can't share the hip but it happens so often that I guess I am looking for any input on what causes this error in general. If I do a few booleans, there is a good chance that a bevel won't work somewhere and it's really killing my progress.
There are no caveats in the documentation and very few threads on the issue. I have literally spent 20 hours trying to bevel ONE problematic edge. I've started the model over a few times now trying to avoid the issue, but because I don't know what is causing it, I eventually run into the same issue or some other edge wont bevel. Is it the ngons, is it the verts with more than x number of edges, normals, etc. and then how do I prevent those issues while keeping things procedural?
I really would like to do all of my work inside Houdini but now I feel like I will have to go back to doing my hard surface in other software which sucks because Houdini has so many advantages over a destructive workflow.
I've attached another attempt at the model and it is giving me issues at the same spot. You can see that every edge OTHER than the one I want to bevel, can be beveled… I assume it is because it is the result of a boolean subtraction… but I can't figure out why. (first image has the edge in question highlighted, second image shows how it won't bevel like the rest).
There are no caveats in the documentation and very few threads on the issue. I have literally spent 20 hours trying to bevel ONE problematic edge. I've started the model over a few times now trying to avoid the issue, but because I don't know what is causing it, I eventually run into the same issue or some other edge wont bevel. Is it the ngons, is it the verts with more than x number of edges, normals, etc. and then how do I prevent those issues while keeping things procedural?
I really would like to do all of my work inside Houdini but now I feel like I will have to go back to doing my hard surface in other software which sucks because Houdini has so many advantages over a destructive workflow.
I've attached another attempt at the model and it is giving me issues at the same spot. You can see that every edge OTHER than the one I want to bevel, can be beveled… I assume it is because it is the result of a boolean subtraction… but I can't figure out why. (first image has the edge in question highlighted, second image shows how it won't bevel like the rest).
Edited by decksounds - Sept. 6, 2019 13:20:29
- decksounds
- Member
- 109 posts
- Joined: March 2017
- Offline
There must be something wrong with my approach.
Here you can see the base mesh that I box modeled as well as the cutters for the subtractive boolean operation. The cutters are all boxes other than the extruded curve that I used for the problematic cockpit area.
Here you can see the base mesh that I box modeled as well as the cutters for the subtractive boolean operation. The cutters are all boxes other than the extruded curve that I used for the problematic cockpit area.
Edited by decksounds - Sept. 6, 2019 16:01:19
- anon_user_89151269
- Member
- 1755 posts
- Joined: March 2014
- Offline
decksoundsYes, there's: you've swallowed this boolean cookie (pun intended) whole.
There must be something wrong with my approach.
I can't help you without a scene file - figuring stuff out based on images and your description has a low ROI - but I think you'd do yourself a favor by approaching the modeling of this shuttle, or w/e it is, with a more traditional sub-d modeling technique.
Booleans are great, but they are a tool to be used when they're the best solution to a problem, not intended as the bread and butter of your modeling tasks.
Apart from this, be aware that the polybevel has problems. RFEs have been filed.
Edited by anon_user_89151269 - Sept. 6, 2019 17:22:20
- lfranceschini
- Member
- 72 posts
- Joined: April 2016
- Offline
The “Unique Points Along Seams” setting on your boolean subtract may be the issue, it creates disconnected geometry where one surface comes from the original geometry and another surface comes from the cutting box. You can see this effect by appending an explodedview node after your boolean. See if unchecking that option on boolean gives you the result you want.
- decksounds
- Member
- 109 posts
- Joined: March 2017
- Offline
pickleddecksoundsYes, there's: you've swallowed this boolean cookie (pun intended) whole.
There must be something wrong with my approach.
I can't help you without a scene file - figuring stuff out based on images and your description has a low ROI - but I think you'd do yourself a favor by approaching the modeling of this shuttle, or w/e it is, with a more traditional sub-d modeling technique.
Booleans are great, but they are a tool to be used when they're the best solution to a problem, not intended as the bread and butter of your modeling tasks.
Apart from this, be aware that the polybevel has problems. RFEs have been filed.
Thanks a lot for taking the time to help out, I hear you on the right time and place thing.
I was able to get near perfect results using the crease node and making sure my boolean shapes consist of all coplanar primitives. Some of my quads were out of whack.
Here is the latest result.
Edited by decksounds - Sept. 6, 2019 19:10:46
- decksounds
- Member
- 109 posts
- Joined: March 2017
- Offline
This is what I mean about the primitves having verts out of co-planar alignment. Polybevel is very sensitive to this, from what I am learning.
I feel like the documentation should read, "The node can handle very complex inputs, so long as the geometry is comprised of primitives containing mostly co-planar vertices.“ … or something to that effect.
This is probably why the polybevel+booleans work really well with ”stock“ geometry like boxes and cylinders. I'll have to be more careful when manipulating primitives from now on.
Is there a ”make primitive planar" function in Houdini? That would be super useful.
In this example you can see the dramatic difference in the bevel resulting from moving one of the points out of planar alignment.
I feel like the documentation should read, "The node can handle very complex inputs, so long as the geometry is comprised of primitives containing mostly co-planar vertices.“ … or something to that effect.
This is probably why the polybevel+booleans work really well with ”stock“ geometry like boxes and cylinders. I'll have to be more careful when manipulating primitives from now on.
Is there a ”make primitive planar" function in Houdini? That would be super useful.
In this example you can see the dramatic difference in the bevel resulting from moving one of the points out of planar alignment.
Edited by decksounds - Sept. 6, 2019 19:13:06
- anon_user_89151269
- Member
- 1755 posts
- Joined: March 2014
- Offline
I'm not sure whether the bevel's poor results is related to non-coplanarity, that's up to the devs to figure out. What's certain is that in xsi I get great bevels for the same exact geometry.
The best we can do as users, is to file well documented (pics, scene files and comprehensive explanations) RFEs to Support and they take it from there.
So do this since you are having problems with the bevel. Your RFE will stack up on top of mine and probably others, increasing the chances of getting a fix.
The best we can do as users, is to file well documented (pics, scene files and comprehensive explanations) RFEs to Support and they take it from there.
So do this since you are having problems with the bevel. Your RFE will stack up on top of mine and probably others, increasing the chances of getting a fix.
- decksounds
- Member
- 109 posts
- Joined: March 2017
- Offline
pickled
I'm not sure whether the bevel's poor results is related to non-coplanarity, that's up to the devs to figure out. What's certain is that in xsi I get great bevels for the same exact geometry.
The best we can do as users, is to file well documented (pics, scene files and comprehensive explanations) RFEs to Support and they take it from there.
So do this since you are having problems with the bevel. Your RFE will stack up on top of mine and probably others, increasing the chances of getting a fix.
While watching the H18 preview, I see there is a new setting called “asymmetry tolerance”, I think this may relate to the issue with the “coplanar” issue I was mentioning in this thread. Very excited for the new release!
-
- Quick Links