[TopAce]

But I must have one subobject for the model itself to avoid some visibility errors in-game.

[Flipside]

Hmmmmmmmm... actually, since things switched to HT&L I've been doing a few 'bad' things for Alpha/Beta versions, and FS2 has been very forgiving

[KARMA]

I use object union too sometimes, what's wrong with it?

[magatsu1]

it sucks donkey doo-daa. I've yet to try TS v5 booleans yet though.

I prefer point editing. I'm guessing modifing one object to the shape you want is better than joinig several objects to create the same effect (see Bob's post)

[TopAce]

I am not so sure if I had to ignore object unioning. You may be right, you used FSO more than I did.

[mikhael]

Regardless of FSO or not, its still best to avoid intersecting polys. Why invite possible trouble when you can ensure you won't have it? Unioning the objects is a good thing, TopAce.

The main concern with the union, of course, is that the geometry that results from the union is sometimes not accurate. Truespace lacks the depth and breadth of Lightwaves point and poly editting/construction tools, so you may have a lot of trouble cleaning up those errors in Truespace.

[magatsu1]

Nah. Easy Peasy.
I think Truespace is just "misunderstood":nervous:

[KARMA]

I don't suggest to use intersections, but sometimes, they could be useful to reduce the pcount a lot and the amount of time too

[TopAce]

Time is only secondary - Everything for GAMEPLAY!

[Flipside]

Lightwave will actually be pretty forgiving with Subtractions and even Unions on occasion, as long as you remember to triple the polys, and to merge points afterwards It's not 100% reliable, the more complex the object, the higher the risk, but it CAN be done

[mikhael]

I find exactly the opposite, Flipside: Tripling the polys adds more edges  to the equation and makes things more error prone.

What I do is select the faces on the target layer, cut them and paste them to another layer. If they're all coplanar, I'll merge them. I then do the boolean on this bare face. After that, I do my clean up (usually just a merge points and possibly subdividing the non-quads intelligently), then I move them back to the original object and merge points again.

It seems like extra work, but it really does make booleans far more reliable (not that Lightwave has bad booleans to begin with).

[Flipside]

I might well give that a try Mik, my current favourite is to subtract the main model from the Subobject (in a different layer), move in really close and move the subobject about 1nm from the hull. With HTL, you can also get away with deleting the inwards facing polygons, not sure if this works normally though. But this still relies very heavily on the complexity of the surfaces you are joining.

One handy trick I've learned is keeping the UV Map open in one window, that way you can select individual points on the model from the Atlas map, which is a lot easier sometimes, especially when you are working on models with 'Internal' components.

[Krackers87]

another thing that helps sometimes is to save the .cob as a .dxf

then open the .dxf and save it as a .cob again. It tends to clear of Fu)(ups