Generate new mesh based on selection on existing mesh

Generate new mesh based on selection on existing mesh

Select a mesh, or a portion of one. Set how far the new mesh is to be generated over the old mesh. Set how "fine" the new mesh is going to be. Hit generate. The new mesh will now be floating on the of the old mesh. Ideally, the new mesh will take on the overall shape of the selected mesh (think of it as generating a breast plate armor piece over a selected original chest area of a body mesh) - but is not an exact copy of the original mesh. Another example is if one selects several sphere meshes and use this tool on it, it would generate some kind of pod-like "shell" mesh over the spheres.

This is a helpful feature for making armor or clothing or pod/shell. A bit specialized, but a time saver to get a rough mesh going.

Hi.
Mmmh, I think it´s nearly already there. For that case I use "split", "flip normals" and "shell". Works fine for me.

With kindest regards
Frank

frank beckmann said:

Hi.
Mmmh, I think it´s nearly already there. For that case I use "split", "flip normals" and "shell". Works fine for me.

With kindest regards
Frank

Click to expand...

Thanks! Haven't thought of that one before. It requires some clean up work after, the shell - which may or may not be a big pain since one may not need the backside (copy of original) polys.

The other concern is that I don't want to have the original mesh in the newly created mesh. Licensing reqirements you know. I was thinking more along the line of the application generating a new mesh in a similiar shape (but off-set or shelled as defined by the user). The mesh can be coarser or finer depending on the setting.

I was hoping to avoid creating polygons over the mesh by hand.

Programmatically generating a new mesh from an older, copyright mesh is probably about as defensible as, say, selling a photograph taken by someone else that you've applied "Unsharp Mask" to.

That said, there are perfectly legitimate reasons to want this kind of function -- Silo has it in spades (you can draw new geometry on an existing mesh and it just builds it). It probably makes more sense in a dedicated modeling program than an all-rounder such as Cheetah.

podperson said:

Programmatically generating a new mesh from an older, copyright mesh is probably about as defensible as, say, selling a photograph taken by someone else that you've applied "Unsharp Mask" to.

That said, there are perfectly legitimate reasons to want this kind of function -- Silo has it in spades (you can draw new geometry on an existing mesh and it just builds it). It probably makes more sense in a dedicated modeling program than an all-rounder such as Cheetah.

Click to expand...

I don't mean the application just subdivides the selected existing mesh. I mean the application would make some decisions about creating a new mesh that conforms to the existing mesh, but would optimize it based on user settings. So, while it is derived from the original mesh, it isn't a copy of it. Some "thinking" went into it to create the new mesh. Granted most of the thinking was done by the application.

I understand your point about a more dedicated (um, expensive?) modeling application to handle the less mainstream features. I just don't want to use (and pay for) 5 modelers to get the job done. Perhaps we can do this in a script/plug-in.

Sixer.
Honestly I must say I have no clue what you are aiming for. As there any example out there, where I can have a closer look?

Frank

frank beckmann said:

Sixer.
Honestly I must say I have no clue what you are aiming for. As there any example out there, where I can have a closer look?

Frank

Click to expand...

Hi Frank,

I am not sure there is an existing example that does exactly what I want to do the way I want for it to get done. The closest working example that I was made aware of is Silo. In Silo, one can "paint" a cross hatch of lines over a portion of a mesh and it will generate a new mesh based on the cross hatch while retaining (more or less) the curvature of the original mesh. The result is a different mesh, with different polycounts and structure.

Zbrush 3 (based on the demo video) "seems" to have the workflow that I suggested, but I can't tell if the new mesh is a scaled up version of the original mesh.

The method you had suggested works fine (minus the cleaning up that may be needed). The only issue with that is if you are trying to - as an example - create a bodysuit for a figure from a vendor (say, Daz or eFrontier), you can't legally re-distribute the mesh (or part of it). So you can't share or sell your creation. The goal is the new mesh not being merely a scaled up version of the original mesh.

I think the reason that it's not been done, is that the program would in essence, have to know how to poly model by itself. I think this is just to complex a thing to program. Silo's much copied topology brush is the best idea I've see along this line, excluding a dedicated app like Topogun.

I had a look at Zbrush3 demo too, but I can't remember seeing anything like your idea. Where in the demo were you thinking of?

squareitround said:

I think the reason that it's not been done, is that the program would in essence, have to know how to poly model by itself. I think this is just to complex a thing to program. Silo's much copied topology brush is the best idea I've see along this line, excluding a dedicated app like Topogun.

I had a look at Zbrush3 demo too, but I can't remember seeing anything like your idea. Where in the demo were you thinking of?

Click to expand...

In the Z3 demo (and the Next Step page), look under mesh extraction. You'll see that an upper torso armor plate and a torn t-shirt is made from a selected area of the mesh. It looks like it isn't an exact copy of the selected mesh - take a close look at the t-shirt.

I don't think the application need to know how to poly a model. It'll have all the information in the selected mesh. It may just need to "un-subdivide", smooth, or optimize the mesh. That's where the "fine/rough" setting comes in. For example, "fine" would produce an exact copy of the mesh, whereas "rough" would approximate the general shape using less polygons.

I think I would be happy if I can get a rough shape/mesh from this feature. I can then go in and tweak. I just didn't want the new mesh to contain an exact copy of the old mesh. It would become an issue if I decide to sell/give-away the model if it contains parts of someone else's mesh.

Sixer

I just realized a quick and dirty way to implement this (still have to add some code in C3D, but wouldn't need for C3D to figure out the topography of the mesh) is to:

1: Do a "shell" like operation, but only retain the "top" part of the shell. Or just copy the selected polys and scale them in the x, y, z directions.

2: Do a polygon reduction.

3: Do a subdivide.

Keeping in mind a user setting for how much scaling, polygon reduction, and subdivision is applied.

Sixer

I'd take a look at Silo. It's a dedicated modeler with some very useful functionality along these lines.

I have looked at Silo, and I'll learn it once I have some free time. Right now, I know C3D better.

C3D can do most of what I want in this feature now. It can shell and subdivide. It just doesn't do poly reduction yet. I don't think this feature as being all that specialized.

Like I've said before (I think on another thread), I don't like to jump between too many applications during my work-flow. I much rather (in this example) work with three applications to do my modeling - one to do the basic shaping, one to fine tune, and one to detail - than to jump between two applications (for example) everytime I want to extrude.

Six