OpenSubDiv is on its way back

news

(Veezen) #61

It means that we can’t expect anything similar to OpenSubDiv in 3Ds Max?


(so3Datel) #62

only in our dreams :cry:


(Veezen) #63

< cry in Blender >

Well, but is there any proof for that? I remember that they were talking about it during some YT stream and I think they mention that they want to heavly improve that feature, especially in modeling part so I thought that this is what they wanted to do :<


(artell) #64

Not sure what you’re trying to show with this video, i’ve only quickly watched it since 20 minutes is a bit long… But Open subdiv in Blender will replace the current Subsurf algorithm, so it looks pretty much the same as what you can see in this video. Faster subsurf for still meshes and animated meshes, which is a milestone feature.


(<== Lost? Click Me) #65

https://lists.blender.org/pipermail/bf-blender-cvs/2018-October/115460.html
This commit makes OpenSubdiv to properly work in edit mode. - Sergey


(Hadriscus) #66

Yep, this and improved creasing - as far as I’m aware everything we can see in the video is already possible. Blender has had edge creasing for decades, even though it might not be as high quality as OSD’s. (?)

;D


(so3Datel) #67

OpenSubdiv for 3ds max has:

  • more accurate mesh smoothing (tris, ngons)
  • more accurate mesh smoothing with creasing edges and alone vertices
  • 2 algorithms for smoothing mesh
  • 4 algorithms for smoothing uvs https://youtu.be/dZD37Pb6fgo?t=213
  • adaptive tessellation in viewport and render
  • calculating with CPU or GPU at choice

http://graphics.pixar.com/opensubdiv/docs/subdivision_surfaces.html

sorry for my english


(Hadriscus) #69

Only tesselation is missing. All the rest is brought along with OSD.


(eobet) #70

Will creasing have any kind of granular visual representation in the viewport?

I know 3DS Max has a “heat map” sort of view, but I’d like to have the edge color represent the crease weight.


(Romanji) #71

I think the most important beneficial ability of 3DsMax way of modeling is the fact that you can stack multipliers (non-destructively) on top of each other while keeping the ability of editing the mesh intact.
That means you can use a mirror + subdiv and then edit the resulting mesh further without applying the modifiers.
In Blender (and most of the other DCCs) you can only stack the multiplier on top of your mesh.
This is actually a feature the BF should look into. An non-destructive edit mesh modifier could be very helpful.
Or this could be archived with the “Everything nodes” upgrade.


(Michael Knubben) #72

The Edit Mesh modifier is a crutch, and actively works against the nondestructuve nature of the modifier stack. If you edit anything underneath the edit mesh, it’ll fuck things up.


(Romanji) #73

True, but the only way around that is using a complete procedural approach like Houdini does. Which makes the modelling process much slower.
The modeling stack with the edit modifier is like an elaborate undo system, that’s how most people use it i guess.


(Indy_logic) #74

I agree to some extent but It’s also very useful for fixing things when sht get’s fcked up. We get shots in the lighting department all the time with penetrating geo from animation. This is usually just a simple edit poly mod and some vertex tweaking to clear up the issues.

I think eventually, an edit mesh modifier would ultimately be a positive thing for the most part. There are of course issues that you would need to be aware of while using it but the benefits outweigh the negatives.

This is really the ONLY feature that I think is really great about Max. I’ve used Max for years and years and I truly do NOT like it. But yeah, props to Max for the modifier stack.

Also, just wanted to point out that no one actually models like what was shown in the video. It would be terribly in-efficient.


(Michael Knubben) #75

I agree, but at that point it’s almost the same as copying the mesh and collapsing the stack. I would probably use the modifier myself if it was available, (in the exact situation you describe) but the people I hear clamoring for it often don’t know how to use modifiers at all.
If development time was put into any of this, I’d prefer it be on (other) new modifiers, or an improvement of existing ones, like the bevel.


(Hadriscus) #76

I think it’s great idea and have defended it in the past - however it has to go hand in hand with powerful element reordering tools, maybe even a proper spreadsheet view, like houdini has.


(Indy_logic) #77

Actually I’d rather it just be simple. I think it would actually be pretty easy. In my head it would just be a storage container for additional mesh data that would override the data below. So, if you stick an edit mesh on top and move some points, they just get stored as offsets to the data below. Same goes if you do something distructive like adding or deleting verts. It just overrides whatever is below.

I mean, I’m not saying it’s something you could do in a day or anything. But I have a feeling it doesn’t need to be as complicated as many have made it out to be.


(Hadriscus) #78

If you want procedural, non-destructive modeling (I do too) it better be editable from the start, otherwise it doesn’t live to its full potential - say you need something changed at the very base of your modeling process, like an eight-sided cylinder instead of a six-sided cylinder - change that and all the resulting operations are disrupted. Unless… you’ve got selection tools that operate on rules instead of fixed elements (select every nth face instead of specifying an index). In that case the changes are properly propagated throughout the node graph. I would like a full-fledged nodal toolset like this, it would be so cool. We have Houdini right now, but hey… :slight_smile:


(Lumpengnom) #79

The editMesh/editPoly modifier in Max is extremely useful if you know how to use it. It can be esspecialy useful for animating parts of a mesh in combination with volume select modifiers and similar things. You have to know its limitations of course but just because it has limitations doesn´t mean it is useless.
If you use it in combination with volume select modifiers (or other modifiers indepentent of vertex count like bend or twist) you can often even change the underlying vertex count of the base mesh without it causing trouble.


(Indy_logic) #80

Max doesn’t do this but it’s still useful just the same. Like I said before, it’s just about the only good thing about that software. :wink:

I don’t really care for full procedural modeling actually. We have Houdini at work too but it mostly used for effects. I know a lot of people like the procedural modeling workflow but I find it clunky and far less fluid. Another app that has a procedural modeling workflow now is Modo. But they mix it with a non-procedural/direct mesh modeling approach.

But, to each his own.


(Hadriscus) #81

I don’t see it as anything more than an ‘interfacing’ matter - imagine a Blender that would construct a node tree as you model, creating a node for every mesh operator called - as is done right now - from the viewport. There wouldn’t even need to be a change in workflow. Ah, sure, it’s just me fantasizing, but… it’s not just a geek’s fantasy. Recently I’ve been wondering how to rig some really extensible/changing creatures, and thought, we need some flexible ways of creating/extruding geometry, and it’s kinda clunky to get there with the current modifier design. Anyway, I’m straying a bit far from topic now…!