Twitter discussion: What Andrew is currently doing.

[Tony Nemo]

Well, updates have always been free, and upgrades to full versions have been minimal. Having to rely only on new seat sales stifles development, and it is not a model I would want to see Andrew adopt.

The workman is worthy of his hire. Andrew certainly is!

[michalis]

But first 3DC has to add a "zbrush' room among the others

A room for surface sculpture AFTER re-topo. Its already there but lacks of good tools (zbrush tools!). Now please tell me that you don't like zbrush sculpt tools... the only real rower of zbrush (as the performance of course). Voxel sculpt tools of 3DC are great too and can't live without them.

These are two different approaches, both of them are great. I'm just using both apps and this makes me happy.

[SonK]

Andrew,

Please support Modo with AppLink

[philnolan3d]

A room for surface sculpture AFTER re-topo. Its already there but lacks of good tools (zbrush tools!). Now please tell me that you don't like zbrush sculpt tools... the only real rower of zbrush (as the performance of course). Voxel sculpt tools of 3DC are great too and can't live without them.

These are two different approaches, both of them are great. I'm just using both apps and this makes me happy.

At the moment the Sculpting room can't even support large numbers of polys. Since Andrew is working on Multi-threading for the paint room now I wonder if that will address the poly count problem.

[hOss]

Add my vote to SonK's for a modo AppLink implementation! As it stands now, I'm really glad modo handles the .LWO format as good as it does to work well with 3DC!

-Patrick

[ozukaru]

I think further sculpting after retopo is not a concern in 3d-coat (since voxels, I mean in zbrush you must retopo even to be able to go on sculping properly (to redistribute polys), not here in 3d-coat is not necesary you only retopo when you are done with sculping) am I wrong?

[philnolan3d]

I can imagine some people may like to create a base with voxels and then finish the rest of their sculpting in the painting room or maybe the Sculpting room if that gets updated.

[haikalle]

CUDA support for linux version !! That must be a dream. Linux version was fast already, but with cuda. Have to try this...

[GeeJay]

I can't believe it! Finally the Linux version supports CUDA.. A dream come true.. This is true cross-platform support, I am so happy! This kind of dedication is so amazing, I am totally sold on 3D-Coat now.

[ozukaru]

CUDA support for linux version !! That must be a dream. Linux version was fast already, but with cuda. Have to try this...

but what do you do after complete a model in 3d-coat linux? ... I mean, there is a 3dsmax version for linux too?

well I supose you can use blender.

[kay_Eva]

Will there be an AppLink for Blender 2.5? That would be nice ^^

[GeeJay]

but what do you do after complete a model in 3d-coat linux?

Well, import it to Blender, Maya, Softimage or Houdini. Render it with VRay, Thea, Octane, 3delight or any other renderman renderer. Do post production in Nuke or Shake. Paint it with Mari. And last but not least send a bill to the client. At least that what I've been doing for the past couple of years, so there must be some kind of life without Windows, right?

And Blender add-ons are Python based, so they are cross-platform if you respect some basic rules.

It seems to me you either have a very limited knowledge of 3D infrastructure or just want to troll.

[haikalle]

@kay_Eva, you might want to check this out.

http://www.3d-coat.com/forum/index.php?showtopic=5853&view=findpost&p=43978

[philnolan3d]

Very interesting tweets tonight.

This night was hard to sleep (children). And looks like I came to very easy thought how to replace marching cubes with ...

algorithm that produces smooth surface that consists of quads only and results with much less faces than mcubes (no thin or small tris).

I have not found that this method was mentioned somewhere. Interesteng... I will implement it when will get some time.

[haikalle]

If Andrew can pull this off. It would be amazing... I just hope that there is no big bumps on the way coding this.

Life is interesting. Sometimes you need crying children to be inspaired.

[GeeJay]

Marching cubes is a common algorithm to convert point clouds to surfaces. It is used for example to convert particle based fluid simulations to meshes. I guess the old quadrangulation algorithm was marching cubes.

[haikalle]

Isn't the marching cube, the algorithm that is used to create voxels...

[GeeJay]

Ah, that would make sense. And, it would make a possible enhancement to it even more tasty!

[haikalle]

I'm not quite sure but I would guess that... Now the voxel density is uneven and if Andrew has found something to make it more even, then it will be a huge

thing. But I think we have to wait and see...

[philnolan3d]

Yes, Marching Cubes refers to voxels.

http://en.wikipedia.org/wiki/Marching_cubes

[Psmith]

Marching Cubes is actually the process Andrew uses to create an organized polygonal mesh FROM a voxel object.

I believe what he is indicating is that he has arrived at a better solution for automatically creating this mesh from a voxel object - one that contains only nicely spaced quads, (no n-gons or triangles).

But, at this point, it is all in his "noodle".

Greg Smith

[Javis]

Marching Cubes is actually the process Andrew uses to create an organized polygonal mesh FROM a voxel object.

I believe what he is indicating is that he has arrived at a better solution for automatically creating this mesh from a voxel object - one that contains only nicely spaced quads, (no n-gons or triangles).

But, at this point, it is all in his "noodle".

Greg Smith

Yes sir! You are on the dot, but then... You always are. Man, I really wish we could have gone for a longer beer outing in L.A. during Siggraph. That surely would have been great.

Anywho, yep, marching jello cubes. Andrew always comes up with the greatest ideas when he's around his family. True inspiration.

[artman]

Also I made experimental option for voxels export - export as dense quads. Possibly it will be helpful to bring raw meshes to ZB.

Andrew,would it be too difficcult to implement direct merging to Ptex using that mesh?

Or without a carcass mesh 3Dcoat would have hard time giving a good painting performance?

It would be cool,users would not need to draw guides and create a good carcass mesh

and most importantly:there would be no displacement needed...

so that means no possibilities of artifacts on areas that are difficult to project/displace.

It would give great workflow: Voxel-Painting.2 step-way.

And then user can use Texture baking tool to get all maps directly from painted hires.

[Andrew Shpagin]

Andrew,would it be too difficcult to implement direct merging to Ptex using that mesh?

Or without a carcass mesh 3Dcoat would have hard time giving a good painting performance?

It would be cool,users would not need to draw guides and create a good carcass mesh

and most importantly:there would be no displacement needed...

so that means no possibilities of artifacts on areas that are difficult to project/displace.

It would give great workflow: Voxel-Painting.2 step-way.

And then user can use Texture baking tool to get all maps directly from painted hires.

Yes, I am thinking same thing too. It is just step toward this.

The only possible drawback is that Ptex carcass mesh coul'd not be tooo big, no more than 1 M approx.

[artman]

Yes, I am thinking same thing too. It is just step toward this.

The only possible drawback is that Ptex carcass mesh coul'd not be tooo big, no more than 1 M approx.

hmm.,,,thats a big drawback..1 million would require displacement too,its too low for usual Voxel sculpts...

but maybe projection will have less chance of errors because Dense quad mesh will be closer to hires than autoretopo generated mesh.

So that would be still great if displacemt works well on that mesh.

If not,then it would be pretty much the same than autoretopo,faster..

without ngons/tris to fix but wihout edgeflow...so half better/half less-good.

[philnolan3d]

Remember that a voxel mesh would most likely not give a good poly flow for animation, which is a big reason why we retopo in the first place.

[jedwards]

Yes, I am thinking same thing too. It is just step toward this.

The only possible drawback is that Ptex carcass mesh coul'd not be tooo big, no more than 1 M approx.

I would definitely like to try this. I haven't used ptex at all since it was implemented because I don't like having to subdivide a low rez mesh to work on it. Painting the sculpt (or in this case a very good optimized replica of it) and baking to an optimized mesh would bring it much closer to a zbrush-like experience using polypaint, only without needing to have proper uvs on the high rez mesh in order to bake.

[jedwards]

Does the zbrush navigation scheme mentioned in the latest tweet include using shift to snap to ortho camera views too like in zbrush? This would be a very welcome addition - honestly it's something I feel should be available in all 3dc applications.

[MADjestic]

I haven't used ptex at all since it was implemented because I don't like having to subdivide a low rez mesh to work on it.

As far as I understand, subdivision is necessary only if your mesh has non-quad polys, so you should be able to avoid subdivision step if you want to. Even in that case, you should be able to bake ptex texture into a regular UV space texture of your low-res mesh, granted you have proper UVs on it.

[jedwards]

As far as I understand, subdivision is necessary only if your mesh has non-quad polys, so you should be able to avoid subdivision step if you want to. Even in that case, you should be able to bake ptex texture into a regular UV space texture of your low-res mesh, granted you have proper UVs on it.

Cool, I didn't know that. Thanks!