Have a look at THIS picture. Spot the difference? Yeah, the one on the left is DEFAULT mapping, the one on the right is PLANAR mapping. Why do we need a third option?
Well, take the left most object. Note how the ENTIRE texture has been mapped onto the face, whereas the right hand object is missing the edges on the right and left sides. Got that? Now, the DEFAULT texturing would be great were it not for the fact that since the object is made of triangles, rather than quads, there has to be a diagonal break down the middle. Effectively making the face two polygons, rather than one.
I know a number of people modified their textures and uploaded numerous copies of the same image to fit on various tapered objects, increasing the number of images on the assett servers, as well as the amount of L$ paid for uploads. But what if you either can't do it or the build prevents it?
Consider that you are wanting to build a large curved screen. You need to use a single texture and replace it with the video stream yeah? We all know how to do this, it's easy, but when you're spreading this over numerous prims it's not as simple anymore. A series of square prims is also easy, but when you're curving the screen, a la iMax, you can't do it any more. Ever noticed how nobody has ever built a curved screen that is more than 10m diameter? This is why.
So, finally, what I propose is a third mapping technique that regards the face as a quad effectively, or an alternative type of object (a tick box maybe to switch between tris and quads?) that can be mapped properly. I know OpenGL can do it simply, it has been doing it since before it was even called OpenGL when it was first introduced by SGI. Now can Linden Labs introduce something for us that will allow us to build what we really want?
Discuss it here!
