I’ve been working a lot with mapping a single image to multiple faces of a prim, and multiple prims of an object. The overall benefit to this is that the object’s images appears to load faster. This is because there is only one image. This makes for less bandwidth used to see the final result.
Working with these kinds of textured prims is a nightmare if you do not stick with separating your images into a fixed set of rows and columns. Sometimes, these rows and columns work against the very nature of what I am trying to do.
One example is a texture map containing 4 circles. Each circle can be mapped to the end of cylinders. Once you have those mapped, there is still a lot of empty space that wasn’t used in the original image. This would be the space surrounding the four circles. You can start to add additional images in those locations. However, it gets harder to map them accurately.
To get around this problem and make things really simple, I created a small application using Microsoft Visual Studio. I load up an image, and then select the area that I want to map. It’s really easy to use. The mapping information is updated in real-time and is easy to just copy/paste into the second life client.
It’s not done yet, but it’s close. I would like to eventually get rotation and horizontal/vertical flipping as well, but that may have to wait for the next version. It’s almost a wonder why the second life client doesn’t have this built in already.