> Anyway what you are talking about there are definitely some great ideas,
> but still are very much in the realm of requiring quite a lot of work, so that
>part that I mentioned about it taking a lot of effort and not knowing when
> it will be possible to happen still applies.

Oh, most definitely it will take a lot of work! I didn't mean to sound like it wouldn't. My comment about "make it more generalized" was intended to be about weighing down the interface and workflow with specialized tools. All this would require a boatload of new under the hood work to make happen. Most, if not all, the commands would have different actions to get the same interface results as the NURBS tools. But I was seeing more of the interface being much the same (just maybe some more options for the cage tools if needed) so that the much of the same workflow can happen and you don't end up with that compartmentalized feeling a lot of other programs have; it's rather annoying to find that you can do this or that command on one type of object but not another. The ones that raise my ire are usually a generalized concept like an lattice deformation.

> Typically existing functions in MoI are very oriented towards precision and
> using the mouse to be able to do object snaps and stuff like that, but
> usually a sub-d modeling tool's functions like extrude are wired up to be
> less precise and more just give you some quick sculpting like feedback as
> you wave the mouse up or down instead of being a "snap to objects" kind
> of thing. So to make the same kind of sub-d modeling feel I'm not sure that
> you would get that by using the same existing extrude command.

Well, you can already just click and drag the extrude tool. ^_^ And having the precision is also a good thing when doing hard surface models (that you want to have as sub-d meshes for import into something like ZBrush). Personally, I think the tools are a good 90% there for adapting, interface and workflow-wise, to a typical poly/sub-d modeler. The only times I feel like I want a more gooey feel is on fillet/chamfer (something like being able to click and slide the mouse while the fillet updates in real time).

Plus holding Alt turns off snapping temporarily; although that turns off ALL snapping, including things like the auto-construction lines with tend to be the only way to get things to extrude in the 3D view without flying off to be on the ground grid. I think a good way to avoid that would be some sort of fast way to temporarily pin the construction plane. Like little squares running in the XYZ directions off the selection. Click the square and the CPlane would snap to that until you start another tool or select another square to change things.

As for the lack of Booleans, I only say that because Booleans are kind of the bane of poly modeling; if you don't have overly dense meshes (that are rather impossible to deal with), you end up with a lot of odd N-Gon shapes that just get more and more painful to deal with and clean up (and which subdivide like utter garbage). Now, if you had it so that the Boolean would intelligently snap the mesh cage vertexes to avoid adding any extra vertexes, I'd say you'd have something worthwhile. Something where you could control a slider to see which shape has the more bias to it's original vertex position. Of course, if you don't prepare the two meshes well, then such a system would likely be useless (one has more edge density than the other, etc). In that case, the only option is to toss out edge loops or add in new ones, both of which will change a sub-d surface. I guess the only best way is a compromise. (Booleans are sooooo much nicer in NURBS.)