I am working on a simulation for a guy who is making a render engine. It’s a bullet through a teapot. The thing is that the shards need to be manifold. I also want them to be extremely sharp. It will exported as an.obj so the edge split would be applied, and when that happens it actually splits the edge. Does anyone have an idea how to get super sharp edges while keeping it manifold (and keeping the surface smooth)?
Enable “Autosmooth” and play with the angle , it can replace edge split modifier without a problem, and can also work with edges manually marked as sharp without the need of the modifier
That is cool. I haven’t seen that one. I will have to check with the guy to see if the engine will respect this setting. Very nice, and thank you.
Can you think of any other kind of modifier that would add real geometry to make it super tight?
Photox, it’s a silly question, but… why in your screenshots you have the whole round, i.e. bullet and casing? 
Off topic alert!
C’mon stan, it’s so simple. It’s all ball bearings, hey!
That’s funny. I have no idea how a real bullet looks, or should look. It’s a bit of a stand in really. I googled ‘bullet’ and took the first centered referrence. So what, the piece at the end shouldn’t be there?
Yes, Stan is right - the copper casing is left in the gun or ejected to the side while the bullet projectile is the lead tip that penetrates the object.
I’m wondering if you might be able to use a texture map to define the edge. i know th fracture tool allows you to place a material on the cut edge, so it is not far fetched to assume that you could creat a bump map to give it an edge. though it may not be sharp enough. not sure, could be worth a shot
I’m not using cell fracture. It’s snap retopo. Cell fracture does not, by default, create manifold pieces (a requirement), now you can have the add-on make them manifold, but then you get to see the ugly side of the boolean modifier, which is of course the horrible topology. Which no amount of modifier can help. It’s a good idea though in general. Although the issue is more with the shading on the smooth surface as it approached the sharp edge.
As for the bullet, I’ll fix it.
You could also try using double edges. It’s not a beautiful solution… no, in fact it’s a messy solution, but hey, if it works - it works.
Sorry about the offtopic, the bullet just immediately caught my eye
Stanpancakes, are you talking about actually having two sets of edges (it must be manifold, so this is out) or are you talking about a compositing node?
Why would I be talking about compositing node when it’s a modelling problem?
I was talking about double edges, i.e. degenerate polygons at each edge (so it would be manifold). I haven’t actually tested it though.
Instead of double edge you could look at a very small bevel value with edges marked with a bevel weight, using the bevel modifier.
The results will depend in the topology of the original objects
Hmm, yeah, Richard’s solution is way better. Double edges produce awful shading artifacts. Bevel with width = 0.0001, segments = 2 and profile = 1 looks indiscernible from edgesplit unless you zoom very close.
