Which one is better?

[an1m3-lov3r]

TurboSmooth or Smoothing Groups?

[Freespace-3DT]

Turbosmooth when close to camera, Smoothing Groups when far from camera, in my opinion.

- Moved to CG Chat section

[NinthJake]

Turbosmooth all the way for me. I have never bothered trying to learn to use smoothing groups, don't really know what the meaning behind it is.

Plus, turbosmooth can be used while modeling (I have no idea if smoothing groups can do that as well)

And the ultimate reason I don't use smoothing groups, Blender don't have 'em :P

[Perversonality]

Turbosmooth with smoothing groups for ultimate control :)

[poopipe]

they don't do the same thing - ergo you can't make a comparison.


turbosmooth adds geometry and smooths the mesh according to some cube based algorithm I've forgotten the name of.

smoothing groups affect lighting by blending (or not blending) edge/vertex normals between adjacent faces - they do not add geometry.
Setting smoothing groups is the same as setting hard/soft edges in Maya (whatever the equivalent is in blender etc. ) the only difference is that you select by face in max and it's stored in convenient groups.

[Tycane3D]

Definately smoothing groups.

It requires a bit more work to use smoothing groups properly, but unlike with turbosmooth it gives u much more control and it doesnt add unnecessary polygons (like on flat surfaces that could do with 1 quad instead of 4,8,16 etc).
down side is that u will have to make ur own rounded shapes etc and then assign the group, but the modelling tools in most software packages make it pretty easy if i know how to use them.

just look at a chamfered box in 3dsmax for example, it only subdivides the edges, and than only in the same direction of the edge, where as with turbosmooth it will just subD every polygons, giving u a higher polycount than necessary., where as if u want to make the same cube using turbosmooth u will have to add an 2 edges near every border edge, and than turbosmooth it...higher polycount.

It just takes a bit longer to do thats all.

[Nukie]

I'm with freespace and poopipe here.

You can not compare the two simply because they do a different thing.

For most of my models, which are for stills or animations, I use turbosmooth, because I find it necessary to have a complete smooth model for the things I do.

Now if you work on models for real time environments I wouldn't go with turbosmooth because (as said before) it adds geometry, and in most real time situations you wont want that. So you use smoothing groups for that.

In some situations, on high poly models, you want to combine the two for more control.

[poopipe]

just look at a chamfered box in 3dsmax for example, it only subdivides the edges, and than only in the same direction of the edge, where as with turbosmooth it will just subD every polygons, giving u a higher polycount than necessary., where as if u want to make the same cube using turbosmooth u will have to add an 2 edges near every border edge, and than turbosmooth it...higher polycount.

It just takes a bit longer to do thats all.


your argument only works in that situation (ie with a parametrically generated primitive) or for realtime work - outside of that, doing things that way works out to be inflexible, time consuming and expensive.


The user needs control over the level of subdivision on an object depending on what they're doing with it.

I dont want to build or make changes to the shape of a high density mesh when i could make changes to a low density mesh that is then subdivided.

I don't want to rig or animate a highly detailed mesh when i could be working off the low poly (0-level) version of a subdivided mesh.

I don't want to be framing a close up only to find the mesh im working with isn't dense enough and I can't subdivide it because the mesh hasn't been built to be smoothable.

I don't want to unwrap a high density mesh when i could be working on a low density one that gets smoothed later.

and so on...


we use subdivision surfaces for a lot of very good reasons

we use smoothing groups for completely different reasons

[Buzzy]

an1m3-lov3r - I gotta side with poopipe here. I'm not really sure how or why you are comparing smoothing groups and turbosmooth. They are completely different tools. Its like asking which is better - a pen or a wrench? Well, do you need to change an alternator or do you need to pay some bills?

Just because they both have the word "smooth" in the name, doesn't mean they do the same thing.

And I don't understand your picture either. The second image is not illustrating smoothing groups at all.

[an1m3-lov3r]

ok... fyi but its not comparing. lets just put this way.. which one do you prefer?

[Perversonality]

For high poly I would use a combination of smoothing groups with Turbosmooth. There's an option to take them into consideration in the TS modifier which gives you more control over sub-div without always having to create lots of edge loops, though it doesn't always get around that.

For low poly I would never use TS. I might use tesselate to add density and volume to a block based mesh but TS/MS are less predictable, so I wouldn't use them at all

[Tycane3D]

your argument only works in that situation (ie with a parametrically generated primitive) or for realtime work - outside of that, doing things that way works out to be inflexible, time consuming and expensive.


The user needs control over the level of subdivision on an object depending on what they're doing with it.

I dont want to build or make changes to the shape of a high density mesh when i could make changes to a low density mesh that is then subdivided.

I don't want to rig or animate a highly detailed mesh when i could be working off the low poly (0-level) version of a subdivided mesh.

I don't want to be framing a close up only to find the mesh im working with isn't dense enough and I can't subdivide it because the mesh hasn't been built to be smoothable.

I don't want to unwrap a high density mesh when i could be working on a low density one that gets smoothed later.

and so on...


we use subdivision surfaces for a lot of very good reasons

we use smoothing groups for completely different reasons

yea true, but then i am looking at it from a realtime perspective coz realtime models are usually modelled a bit more carefull and specific, and if done right a turbosmooth (with seperate smoothing groups) will take care of any tesselation problems u might have,and keep the polycount lower with the same quality as a mesh modelled to rely on turbosmooth, and in that respect i find smoothing groups better. (and that was the question, right?....wich is better)

now if u want to ask "wich ones smoothes better...yea..turbosmooth obviously", its not the same class of tool, u use them togethor.