Any news about Sculpt Tool development?

I would like to know if Sculpt Tool is being developed and advanced?
I tried ZBrush 3 other day and man, I loved it. If Sculpt tool could be sped up a bit (with back faces culling maybe?) and few new tools added like masking some parts with vertex color painting and hiding them, I would love it even more :wink:

Blender have only one developer which worked on Sculpt mode and Multires: Nicholas Bishop. I also thinking often about future of Sculpt in Blender. In my opinion it need more developers…

More developers per area would be nice but it’s not like we pay them to develop wanted features, they’re just commited to making Blender better.

Hi, I’m that one developer that was mentioned. :wink: Rest assured, improving sculpt mode is a very high priority for me, but I’m a bit busy now with school things. New features, and as important, better implementations of existing features, will come as soon as possible.

Thanks for the reply Nickolas. I am very happy that you are still working on it!
I have perfomance question. Why is it 700,000+ polygons model renders very slow in view port in Blender when I am sculpting and at the same time it renders very fast in ZBrush3 ? Is it Blender related issue or Sculpt tool related issue?

Great to hear from you Nicholas!

Erm…that is all I have to contribute for this week.

Koba

I’d say that would be a Blender issue, since Blender has trouble rendering large amounts of polygons regardless of what you’re doing in the view port. Which is a bit of a joke to be honest, this should be a priority area for improvement.

I once got Blender to handle 4 million polygons in object mode, to note one thing, edit mode is a lot slower then object mode, I do hope the GLSL guy can speed up viewport drawing for those who don’t need all those extra bells and whistles.

What about reconstructing the layer of subdvision?.
It’s a really important feature that lacks.
But for me what should be improved is the rendering and using modifier on multires meshes.
No programs have multires meshes,if they can work in a good way they can make the difference,for now rendering huge meshes is really problematic,and posing with armature doesn’t work well.

I agree with renderdemon, expecially about using modifiers on multires meshes.

I do hope the GLSL guy can speed up viewport drawing for those who don’t need all those extra bells and whistles.

I hope they implement a real time cavity shader like Zbrush 3 when they do the GLSL goodies. It really makes sculpting fine details easy to see.
I am very happy with the current version of blender too :smiley: .

Zbrush uses pixals ( 2.5D ). Not pure polygons. If you zoom in a object in zbrush it will get very slow. It all depends how many pixals are on the screen. Thats why it goes slower when sculping in a higher workarea resolution.

So its a totally different system of handling the information.

Renderdemon,

I agree that reconstruction subdivision levels from a highly subdivided mesh is desireable. Here are some thoughts of mine on a possible algorithm.

this only works for mesh without holes/edges (ie doesn’t work for planes but does/should work for a cube, uvsphere etc)

find ‘stars’ (points that have either more or less than four edges meeting at the vertex)

find the connections between them that is your primary topology (you can now count how many cuts are on the line between each of the nodes - the minimum cuts determines what your maximum subdivisons are - and hence determines how many loop cuts are at each of the different lines between nodes before subdivision).

if doing pure subdivisions doesn’t result in the same topology, then there were some loop cuts

calculate the loop cuts

add the loop cuts on your primary topology

subdivide the reconstructed mesh to the same level

move corresponding points to match

propagate changes to the lowest reconstructed subdivision level

Of course it is probably more difficult than that and might be cases where it doesn’t work other than those I’ve identified :slight_smile:

We can probably find someone who is a math major who has had some topology classes that would give us relevant input and maybe algorithms.

LetterrRIp

They’re called pixols, not pixals :smartass:

Also, the main tool is a full 3d mesh as far as I can see. It only becomes a 2.5D representation once you drop the tool onto the canvas.

Well,to be honest I really don’t know anything about reversing subdivision surface,but I have found some paper on the web that maybe can help,I’m not sure,as I never studied how multires surfaces work:

http://wscg.zcu.cz/wscg2006/Papers_2006/Full/B89-full.pdf

http://www.cgl.uwaterloo.ca/~rhbartel/Papers/doo.pdf

I think that brecht have the math skill to help.

Another thing really important is that currently the sculpt tool isn’t much easier to use with other programs,like Zbrush(here the recontruction of subdivision could probably help,as the mesh every time can be recreated without thinking about vertex and face indices)…
I know that Blender wants to stay alone,but maybe isn’t in mho the right choice.
If the workflow could be improved,more people we’ll try to use Blender and other sculpt programs together.

Since we’re throwing out suggestions, I’d much rather see work in the other direction - in better integration with the rest of Blender. Sculpt is ok right now for making a nice picture on an empty background, or for static scene props, but not all that practical for production work.

Right now because of the way multires works, it’s almost like working in a separate app. You have to model, unwrap, and do everything first, then once you start sculpting you’re locked down for good. You don’t have much freedom at all to go back and use other things in Blender, for example making topological edits to the base mesh. We also tried to use sculpting on an animated character at our studio, but had to use Zbrush instead with the displacement modifier, since it’s pretty much impossible to use a multires mesh with armatures. I’d like to see:

  • Ability to make topological changes to the level 0 multires mesh
  • Proper interpolation/deformation with armatures
    (and not so related to sculpt but would be cool)
  • Better integration/consistency with the rest of Blender’s paint tools, in the sense of UI layout, features such as pressure sensitivity, the F key size adjust, lazy brush, etc.

broken…
I’m not sure “even” Zbrush has the ability to make topological changes to the bottom level… Not that it would be astounding that it can’t do something.

Anyway, my logic (my logic! ;D ) says that in order to do that the multires “modifications” would be removed from that face which will be extruded or deleted or subdivided. After that when you exit edit mode, the same level of subdivision would be applied to that extruded/subdivided part. Basically like the sub-surface modifier. I don’t know how hard it would be to code, because I have no coding knowledge at all.

Zbrush has been able to do this for a long time (since 2.0 IIRC). Silo can do this too, although when I tested it, its results were rather… mixed.

Cheers,

Personally I didn’t like Silo 2 and Mudbox :confused: ZBrush is good, but it’s interface rather complicated and clumsy.
When tangent space normal maps baking will be implemented, Blender will be ultimate tool for game developers. And again, speed is a big factor for me atm :confused: