Quote - For the OpenGL/3D part of things I'd suggest OGRE. Cross-platform, open source, a large community, and on Windows you get to choose between DirectX and OpenGL, without sacrificing portability to MacOSX and Linux. And the HLSL support is neat. There's also normal mapping, parallax mapping, and a lot of other goodies incorporatied right into OGRE. Think about the possibility of realtime animation rendering at modern game quality... Combined with good animation tools such a package would definitely gain some foothold in the 3D CG world.

I love dockable windows and toolbars in GUIs.
In general, I'm not that enthusiastic about multimodal apps. Switching modes can be a pain. Dockable palettes are usually a more user friendly way.
Then again, some tasks are clearly different. The Setup room for example, that is quite different from "normal" Poser use, so a special Setup node is not unfriendly to the user.
Actually, something like the Setup room would be a suitable plugin candidate.
Materials, however, are part of the standard Poser workflow and should not require a separate mode. Dynamic cloth and hair - and rigid body dynamics? Why not incoporate ODE into this app? - questionable. I use it as part of my standard workflow, but not everyone is that much into dynamic cloth and hair.
Functionality comparable to the face room - I'd say a separate mode. We're talking morph generation here, not a typical part of the standard Poser workflow.

OGRE is a great suggestion, I'll definately look into that.  Dynamic cloth will be built in, but I hope to make it much more user friendly and useful than the poser version.  ODE looks fantastic, I can think of a ton of uses for it, not sure I can get them all up and running in the alpha version but eventually I can see a lot of extremely useful features that ODE could provide.

The face room - well, I've never had much use for Posers face room, it's a wasted tab for me since it only works with such limited geometries.  I haven't really done enough research yet into the particulars, but it would be great to be able to generate a decent representation of a 3d model from a couple of 2d pictures and have that built in to the programs functionality.  I'll probably have to do a lot of reading and experimenting before I can even begin to tackle a project like that though.

My thinking on this is a bit different though - I've tried the face shop demo and in all honesty I've never been able to get anything out of it that remotely resembles a human face, at least not one that hasn't been infected by some horrible disease.

The demo videos I've seen make it look so easy, but when I outline the jawline in particular I've noticed that very, very minor variations in the outlining process lead to huge variations in the resultant figure.  If I'm going to do a "face room" type of app I want something that is easy to use and will give you usable results with minimum effort, something that won't go outside certain parameters unless you shut off a limiter switch so your end results don't completely suck.  So I wouldn't look for anything like that in the alpha release - Lol

Mostly I'll be concentrating on reading in files, formating them properly, morphing, posing and rendering.  I'd also like to include an easy to use, straight forward rigging system and hopefully a simple, easy to use animator as well.

Once that's done I'll start looking at adding other bells and whistles, at some point it might have it's own modeler, uv mapper, etc - but these are all features that are not on my A list, at least not at the moment. 

-Never fear, RenderDog is near!  Oh wait, is that a chew toy?  Yup. ok, nevermind.. go back to fearing...