'Floating' Geometry on a low-poly?

Andreas

Is floating geometry acceptable on low-poly models these days? I've heard it does happen in games these days, the only problem might be with how the engine does the lighting; I heard that can produce weird artifacts?

By floating geometry I mean if you stick join two meshes together, but instead of cutting one into the other, you just push them till they overlap. Like pushing a trigger mesh into the main mesh of a gun, without cutting the trigger into the body of the gun.

Also, if the model is for portfolio purposes, is it generally acceptable to use this method?

SnowGhost

Floating geometry certainly make UV mapping a more complicated, but as to your question I would like to know as well. I do it but I'm no pro and amateur at best in uvmapping and texturing.

gamedev

Yup, it's widely used in games now days. The only exceptions I'm aware of are w/ engines that use shadow volumes (Doom 3 is one example I believe). Unreal Engine 3 however, supports floating geo just fine. If the object is to be vertex lit as well, consider placing verts appropriately where floating geo and surfaces intersect.

Andreas

Acceptable for portfolio pieces though?

Ghostscape

intersecting meshes are fine. This isn't the same as floating geometry, which is usually used to refer to small pieces sitting on top of a high-poly surface.

It causes weird artifacts primarily with vertex lighting...which few games do nowadays. And it isn't always an issue with vertex lighting, either.

Farfarer

For really low stuff - DS spec, for example - it might be a trade-off between precious extra polygons or precious lost texture space (and, obviously lighting), depending on the model.

Either way, it's a generally accepted technique but it's something you might want to bear in mind.

gamedev

Acceptable for portfolio pieces though?

Yes.

PixelGoat

SnowGhost wrote: »

Floating geometry certainly make UV mapping a more complicated, but as to your question I would like to know as well. I do it but I'm no pro and amateur at best in uvmapping and texturing.

How does it make UVmapping harder? I personally find it to be much easier to unwrap intersecting pieces, than meshes that are completely watertight. Its just a click of a button and then you´re done.

Vrav

Maybe he means it makes it harder to conserve space on your UV map. Personally, I make floating or intersecting bits of low-poly geometry to fill up empty space, but... I'm weird like that.

But no, I mean, I don't see floating geo being any different than a UV seam somewhere, so long as you avoid Z-fight-inducing intersections. If you're using tangent space normals, you could even give the floating protruding bit the same UV space as the surface underneath, and just pop it off for instant LoD—sounds sensible to me.

Ghostscape

PixelGoat wrote: »

How does it make UVmapping harder? I personally find it to be much easier to unwrap intersecting pieces, than meshes that are completely watertight. Its just a click of a button and then you´re done.

When you have an intersection, you wind up with parts of your UV that aren't seen, which means you're wasting space unless you stack bits, which can be awkward to work with.