I create this thread to talk about this topic that all people want to talk but we are messing the gsoc custom normal thread.
Ok, in the last message we are talking about two different tools for two different workflows. Some people was talking about the expected “mesh fusion” solution and pitiwazou about the bevel in boolean new edges that have Houdini. For the last one I think that only a vertex group “output” in boolean modifier is needed to created the same “node” effect, and will be really usefull. And I believe that will be really easy to make this to somebody that know the code.
For the mesh fusion, I think that a modifier like “bevel boolean” addon by Rodinkov Ilya will be a good solution.
In other way, I will want to talk about the “bevel node” for cycles that IMHO is something that solve similar problems and few people want to see.
using bool tool with auto operators between two meshes (ex. A and B) and bmesh boolean, after an operation (ex A - B using Ctr + Shift + Num-) and entering edit mode you can see that the faces which was operand mesh (B) are selected; so you can easily select the boundary verteces and create a vertex group for bevel
A - B using Ctr + Shift + Num-
so this kind of operation may be usefull to create vertex group for bevel
The ‘bevel node’ can not do, however, the same work that an actual bevelling on boolean seams can do. In the case of Mesh fusion like bevellings there is an actual creation of geometry, when in the bevel node it is only an impression of having bevels. An impression that is far from being perfect, given that the so sharp tips of the many sided corners make evident that is a trick only (as is evident in the sample image above).
It is good to have a ‘bevel node’ too, it can be used in cases where there is no any interaction with many sided tips (as is in the cases with cylindrical shapes, for example). But it is a totally different thing from an actual, geometry modifying bevelling on the boolean seams.
Also, the demonstration in the Houdini video is on a low poly object, where the boolean seam goes unobstracted all its way… this is something that can be achieved even now in Blender (as happens in the case of ‘Hard-ops’ plugin). The needed thing is to have a boolean seams bevelling method which works unobstracted even in objects having a dense and complex polygon structure. Ilya’s plugin is a good try towards such a direction… despite its having a need to be more easy as to its work route (now it depends heavily on many manually organized modifiers).
Sorry, maybe it’s me, but I don’t understand your message very good. It seems that you mix concepts or something, but I’ll try to answer.
About round edges shader. Why can not be done a node shader for bevel effect? The “impression” is far enough perfect for video game artist when you are baking the normal maps. Did you see wolfenstein saga? they use that shader all the time, and it’s one of the best game in this generation. With that shader you don’t need to redo the project cage and a lot of other problems that needs hours to be done.
About houdini boolean. The difference is that the booleans system allow us to be interactive, without destructive workflow.
About mesh fusion modifier. Also like the other examples, the good of a modifier like this is the non destructive pipeline. Actually the addon is hard to use and create new mesh that can not undo without lost other work.
DcVertice you did misunderstand me… I did not say that the bevel shader node is unnecessary, quite the contrary: I do consider it a must have for Blender. The thing that I did mean is that it is a quite different discussion from that about a tool of actual mesh bevelling of boolean seams (something as Mesh fusion or Hard mesh do). Bevel shading node is something which requires a thread of its own maybe. : - )
As to the bevel shading, there is a whole discussion about it in the thread below. And there are some very successful tries for having such a thing through osl programming too in the thread… it is definitely worth looking:
(btw, most unfortunately, Photobucket have gone commercial and so my related images in the thread can not be seen… but still there are the images of others)
I want to talk about the three tools, not that the tools are excluding between. And I have talk about bevel shader here because I think that the thread that you talk is dead and is more technical.
I wish to make a debate about that tool that BF ignore, when I think that a coder could do it official without problems. Maybe I’m wrong.
One thing that it’s good to talk about because sometimes we forgot it… Mesh fusion make SHIT of mesh, that you can not use out of mesh fusion. After a mesh fusion you will have strange faces, you couldn’t use subd,… mesh fusion make a “final” mesh where it’s hard work with it.
As Pitiwazou said Mesh fusion produces final results. And they are of excellent quality. They are final, it is true -you can not edit them anymore- but they are fine for hard edge modelling tasks. It would be good, of course, to find a way of producing such meshes that will remain editable, no doubt about that, but… even having such a final mesh producing tool as Mesh fusion would be a superb enrichment of Blender’s tool arsenal.
I am putting below three links, the first is Groboto’s (its authors are the same people who made Mesh fusion), the second is Mesh fusion’s and the third Hard mesh’s (which is an add-on for Maya which does the same work as Mesh fusion)… all of them produce astonishing results:
I personally think that you could have 2 bevel procedures:
With standard bevel you essentially insert 2 edge loops and push them appart based on input (crude simplification). This leads to artifacts with big values
Some of the other tools as Hardmesh instead seem to DELETE anything inbetween the beveled area and create a complete new mesh between. Blender equivalent would be Bevel -> Delete beveled part -> Bridge outlined edges (+ superellipse ):
Cedric made this demo. Note how Beveled area has newly created topology. Manual replication in blender using Bridge gives relatively good results (just flat).
The good idea will be all inside one modifier to do only one bevel operation, and I think that second option is the best. But the important thing… Where must we put the money to somebody make this in a good modifier stack?
