What would it take to get a Boolean mesh cutting tool working in Unity?

It seems like the internet is littered with half-finished, abandoned attempts. I know that it took Autodesk and Blender quite some time to come up with robust solutions, but if it works in Blender, is there a reason it can't work in Unity? I need to be able to cut shapes out of large, complex, non-convex meshes in a way that doesn't destroy topology -- is this impossible?

Is there anyone who, for any amount of money, would be able to do this? Which person? How much money? Please send me their way. Thank you!

 

Have you checked SabreCSG (forum thread)?

 

At this point I think I've looked at every CSG/Boolean asset that exists for Unity, although I definitely could have missed some. SabreCSG looks good for level building, but in the forum thread it looks like it has issues with concave meshes. All of my meshes are concave and have between 30,000 and 200,000 verts, and we can lower poly count some but not enough to make a difference. This is why existing solutions don't work for me -- it isn't worth it to repeatedly buy assets to test if they work, only to have them break on our complicated meshes. I pretty much need a solution as powerful as those in 3D modeling programs.

 

High quality CSG is really, really hard. Often times just because some complex boolean works in one bit of software doesn't mean it works in another. I've not used Blender's CSG, but Autodesk has two completely different solutions, Boolean and ProBoolean, the later of which used to be a plugin for 3ds Max by a third party that Autodesk bought. There have been several times when I've had to try both solutions to get a working mesh, even with relatively simple mesh combinations. Complex meshes just make the problem exponentially more difficult.

Every person / company I know of who has looked at CSG goes "it can't be that hard, can it?" ends up spending months or years and never getting it "perfect" because it really is essentially impossible.

 

Thanks for the confirmation -- that's sort of what I suspected. I'll keep one eye out for future solutions, but for now it's good to know it's not really on the table. I appreciate the help!