Good idea.

At work with my newer card, a setting of 1 or 2 on any simple model runs a smooth as glass, but at home a 15 would drag a bit.

When you talk about a possible automatic setting, I see it as a default "high" setting by the user, like 25 - when the viewpoint or objects are being moved.
Then as you sit at idle with a still model, an off-screen rendering of some type would creep in. It could be done in a way where it feels seamless.
Imagine an ultra-slick movement that subtly cleans itself up to where the edges are ultra smooth. But if would have to detect when you instantly need to change the view, so there is no lag.

Google SketchUp does a simplistic version of this by substituting boxes when you are moving the view on complex geometry. Kinda gross.
But I know that MoI has them all beat already. :-)

BTW, a thought I had that could be employed sooner - A simple timing check when loading a file. If MoI has taken more than a user set amount of time, like 30 seconds, then MoI would up the mesh angle or higher then reload the image.
- Could be a simple way, if not through a warning dialog, to let the user know that a monster is trying to open and the mesh angle is set too low.

...Just a few random thoughts about it...

Hi Mike,

> Then as you sit at idle with a still model, an off-screen
> rendering of some type would creep in.

I'm not sure about doing something that's totally view dependent like that - I'm kind of thinking more along the lines of the current non-view-dependent generated mesh, but having both a coarse one, and a finer one. Still view independent though so that once stuff is finally calculated it wouldn't have to crunch on that mesh generation stuff anymore.

The part that could be good would be removing having just a single global setting for it, which like you described in this thread can actually be harmful if it is set too low.

The coarse mesh would be the first one generated, so the initial file load would be leaner on memory consumption and also would just show up on the screen a bit quicker. Then the idea is that as you continue working some higher res (but still view independent) meshes could be created, if you had plenty of memory available, and as those higher res ones were completed they would then get used so the display would improve after that.

There are still some tricky things to figure out what that though, one is making sure not to run out of memory while making higher res meshes and another one is to handle stuff like what if you shut down the current model and open a new one while it is busy calculating higher res meshes in the background, it has to deal with stuff like that which is possible, just kind of delicate to handle.

- Michael

About Sketchup you have a cool plug by Fredo 6 who permit to keep many components! ;)
GhostComp http://forums.sketchucation.com/viewtopic.php?t=21469