Why no Catmull Clark Subdivision in Edit mode

Is there a way to use Catmull Clark Subdivision in edit mode? I think for most people it is the preferred method, and the method that is used in the modifier section. It’s frustrating that Blender chose to use two different methods of subdivision.


1 Like

mmmhh… I do not get Your question, or better your frusrations… The issue is only to show the modifiers also in edit mode… there is the switch within the modifier pannel…



Happy blending…

P.S.: Might be a good idea to move this one into “Basics & Interface (UI)” :slight_smile:

Might be better to move into the “Sorry, I didn’t look” forum.

That’s not the problem, that lets me see a subdivided object in edit but it doesn’t let me work on it. Let’s say I’m making a face object. I rough it out of a box and then subdivide once. Refine it some more and subdivide again. It requires me to exit edit mode, apply the modifier and return to edit. Without applying and can’t really further edit the mesh.
Another time that this comes into play is when I have a mesh object with many different parts and I only want to subdivide a piece of it, not the entire object.

1 Like

The box to the far right shown in post #2 pointing to show subdivision in edit mode, if you click the last box on the right the mesh conforms to the subdivision and you can edit it in its subdivided form.

sub apply to object not a part of object !

happy bl

Destructive subdivision works differently from its modifier counterpart in that it doesn’t convert to quads, nor subdivide ngons, unless you count subdividing edges as subdivision. Because of this, CC smoothing only looks good a fraction of the time, unless the user always models in quads - which cannot be expected, therefore vertex smoothing is standard.

The destructive subdivision would have to be rewritten to resemble the modifier. I prefer that we have two different types of subdivisions - they allow different usages.

Or you can subdivide without smooth, and apply CC smoothing manually using something like this.

I understand that CC’s were intended for display and rendering. And would require a complete rewrite to make a destructive tool. But I think CC is not the issue and I have to agree with the OP.

I think there is something wrong here. I think that there is a bug or a failing feature. If you look at the example. It should not subdivide that way. When trying it out, I noticed that for some reason the subdivision is hanging on those edges . It works properly on 90 degree angled edges. You can take a cube and turn it into a ball. But you toss any other angle its way and it fails to subdivide those angles properly. Smooth subdivision should not work like this. And there are other similar subdivision tools in other apps - such as Maya and LightWave that work properly without CCs. CC is mainly a very specific math on how to subdivide. As well as a host of other features, not to mention being render - specific. But putting a subdivision next to a CC subdivision should not have that big of a fundamental difference. The basic curves are so similar that it would take a much closer observation to tell. This is grossly wrong in a fundamental way. I would report it as a bug.

IMHO you’re comparing apples and oranges here. Subdividing a mesh is something completely different than turning it into a subdivision surface.

The former just increases the meshes’ density, thereby smoothing them out when used with smooth shading while keeping the general shape. The latter, however, transforms the meshes’ “real” geometry into control points for a “virtual” interpolated surface. That’s not even remotely the same thing.

I absolutely agree. That’s the issue, if someone mixes up “vitual” and “real”.

With a non applied modifier, (especially those which are deforming the base-mesh), the original toplogy of the mesh is not changed. So You have to work on the original. The only advantage is, that you can controll/see the effect, which would be real in case of applying the modifier and therfore “truning the effect into real topology”. Nothing else and nothing less. And IMHO this has nothing to do with a bug or something wrong. It’s ment as THE feature as it is presented. Everything else is another feature.

Well I think the subdivide smooth feature is either broken, or it never worked at all. So it might explain the confusion.

It has nothing to do with CC subdivision. That is another discussion. This is about proper smoothing of a mesh in a destructive manor - which absolutely has use in Modeling. It is a feature that should work. Plain and simple.

It works properly in Maya and in LightWave Modeler. Can’t test XSI and Houdini right now. But two out of two is good enough for me.






Blender as you can see makes complete hash of the mesh. This is not a working feature by modeling standards. For the same reason you don’t always mirror a mesh with a modifier, and you fully expect the mirror function to work destructively, you fully expect all destructive tools to work as their counterparts do. Be that mirror, smooth, spin screw and so on. Even if they don’t work exactly the same with the exact same controls, they should be at least useful in most cases. Subdivide smooth is very very limited in use - at best. I say it is broken or never worked properly.

What I have illustrated here, is that you don’t need to use CC Subdiviion. That is another animal entirely. It has more technical implications than immediate visual ones.

You should be able to model destructively as much and as often as you like with as many tools - working properly - as you can have available.

Personally I think this is a great find and it should be either a bug report or a feature request.

1 Like

Well I step out of this discussion, because I do not get the sense in general. You show two different options in blender, where (in the screen-shot of the blender) the left shape of the mesh is exactly equal to the shape of the other applications. (If you would have shown the edit-mode screen, it would habe been more obvious). As I see in the other apps You are using always the CC-option. And then you compare this with a total different aproach on the smoothing option in blender, which natually has a different effect, cause it’s depending on the poly-count of the mesh where it’s applied to.

But as said, I’ll step out.

Happy blending :slight_smile:

Subdivide smooth doesn’t apply the same kind of “tension” on the mesh as a subdivision surface would. However, you can simulate the added tension by using the “smooth” command on a mesh that has been “subdivided smooth”:


Left: Subdivide smooth + several iterations of smooth / right: Subsurf modifier.

I agree to Mike J. Gee, though, that this discussion still is about apples and oranges.

True to an extent. But the smooth command is another animal entirely. That, while can be used to fix this issue, has another purpose.

Ok let me add more explanation.

In both LightWave and Maya there is a “subdivide smooth” tool. In Maya, you can subdivide smooth using CC (completely destructively) or Linear which grays out CCs. (completely destructively) To save image space I showed two examples in Maya. One destructive with CC and one destructive without CC. You can do this in edit mode on the mesh destructively. And as you can see there is very little difference between the CC version and the Linear version. I was only interested in the final shape. Then in Maya on the same image I showed the CC preview mode. 3 on the keyboard, which is the open GL display of CC Subdivision. This is similar to Blender’s CC Subdivision Surface Modifier. Just as in Blender you can adjust both the preview and the render subdivision level. But we don’t need to get into that.

Result of the Maya test:

Both CC destructive smooth and Linear (non CC) destructive smooth are fundamentally the same. Main difference is it took more iterations for linear. (check my tutorial for subdivision math). Linear and exponential in Maya refer to the main differences in subdivision math between CC and non CC subdivision. This is subdivision math - not shape.

Both of these were compared to Maya CC display. Virtually the same.

Conclusion - both Linear and CC subdivide smooth as well as the comparative feature to a modifier work exactly the same in Maya - as far as the end result shape. I am sure there are differences. But it is negligible. And as I said before CC is a whole other technology as well to do with rendering and displacement and so on.

In LightWave, we have had Metaform smooth which started as a plugin back in 1995. This is Modeler’s version of Subdivide Smooth - in Blender. We had this long before Subdivision Surfaces. And LightWave has had its Supatch mode to replace the need to use Metaform - just to finalize a mesh, before it even had CC Subdivision. But it still exists, as a tool you can use. And it has always worked as a subdivision tool - properly. And there is very little fundamental difference between Subpatch Mode and CC Subdivision Mode in LightWave either. I mean on a very basic basic level. The shapes are basically the same. It is the subdivision math that is different. Linear = 1,2,3,4 etc and Exponential = 4, 16, 64, 256, 1024 etc.

Conclusion in LightWave.

Both LightWave’s Subdivide Smooth equivalent and its CC Subdivision Surface Modifier equivalent work exactly the same as far as the shape is concerned.

Along comes Blender and tries to implement the same feature in Subdivide Smooth. This has nothing but absolutely nothing to do with CC or modifiers or anything else. It is a modeling tool.

And it is completely limited. It has never worked properly on 3D meshes outside of maybe a cube.

I just checked it in 2.49. So I doubt it was never intended to be used outside of simple one sided-shapes.

So you make note of that and move on. Or your request it work properly on 3D shapes or sort out how to code it. Those are the options.

1 Like

But does Maya and LW use the barycentric subdivision as Blender’s edit mode does? Meaning, will a triangle stay as a triangle after subdivision, or quadrified?

The current behavior of subdivide smooth is very similar to the “To Sphere” operator with the exception that that it only performs displacement on new vertices and not the original geometry. This creates a “pillow effect” with pinches or creases on the original geometry. This can be replicated by subdividing a cube once without smooth, then once again with smoothing. Couple this with the barycentric subdivision and the result is… well, hashed as you put it.

No. You are right it subdivides differently. And I am not trying to necessarily assert this is “broken”. I am merely trying to point out its limited use and the fact that in a variety of uses it is well, basically useless. This is one of the first things I noticed when trying Blender for the first time in 2008 with 2.49. As a fairly well-versed modeler, I expected something called “subdivide smooth” to perform as the name suggests. And as I was seeing in Modeler. As a result. I stopped using it. Assuming correctly that it was coded with a very limited use case in mind. Or that it was broken, or shortsighted, or all of the above. Bottom line, how useful was it to me? Not very. Is it practical to only have a “proper” subdivision as a modifier and not have the same functionality as a destructive tool? No not at all. And it is not an appropriate response - in my opinion - to continue to assert the wrong workflow - a modifier - for something you need as a destructive tool which is what the OP was asking for, basically. You would be equally wrong to suggest that - in the case mirror in destructive mode was not working - you use a modifier for that instead. That, in my opinion is apples to oranges. :wink:

For the record I couldn’t agree more that the current behavior is a potential candidate for review. Apart from quad spheres, I’ve had little use of Subdivide Smooth because of its limitations, and because I rarely find the need to destructively subdivide AND smooth something in one go.

Just to throw some ideas on how it could be improved:

  • Factor in original geometry when smoothing
  • Offer option to use CC or Doo-Sabin for more accurate volume displacement
  • Quadrify triangles

Indeed. Yeah those would sum it up. I am not a coder. But I wonder if anyone has created an addon for this? If not it could probably be done that way.

Thank you Richard. That is exactly what I was trying to get across.
I’ve tried to show the difference. The red vertices are the original vertices and the yellow vertices are the additional ones that are added by subdivision. You can see how the Subdivide Smooth works by smoothing the added vertices and holding the original vertices in place. This method almost never produces pleasing results.
Catmull Clark, Subdivision surface, subdivides and then smooths the original vertices and holding the new vertices in place.


Hi.

Please, dont mind cubes too much. There are real cases you to get a proper smooth subdivision of a part of a mesh. Like an arm or a leg of a figure you sculpt.
You used as little detail as possible to keep it easy. After subdivision, you start working out the detail on these parts. Including loops nescesary for animation deformations.
(Of course, This kind of refinement is not the same as pressing ‘subdivide’ button on the multires modifier.)

Atm, I fear this requires me to split my mesh, do subdivisions on parts as I wish, and put everything together.