GE has to render, that is, display the polygonal shapes of a city and lay the photographic images on them. The processor has to plot the math to make the polygonal shapes, and lay the image on it, and then adjust for lighting, shadows and atmosphere.
If the models of the buildings were not optimized, GE would be slow and no fun. Images need to be compressed, and details axed. Often the photo texture has enough convincing detail that it makes up for loss of polygonal shapes.
These pictures show details i built in the model, before admitting I would have to loose them.
 |
| I call the little fire escape roof exits "huts" in my file names. We keep the huts, but loose… |
 |
| the venting and ducts, the little sun window bays, the chimeny… |
 |
| and the power station in the southest corner. This is a dear object to those that care, but we can't keep it. |
 |
| We're also going to loose the little double flat pyramids on top of the northwest hut. |
You don't really know how much you've got to axe, for instance in this model I built up the roof top a lot, which requires more polygons and more images, especially becuase
I took pictures of the tagging on the roof and all the huts. I compressed the files a ton, I have the user and over-all map performance experience in mind – ideally it would be as fast as spinning a globe. If every one puts hi res files of their favorite tags on every surface, GE would be too slow and not much fun.
 |
| The silver cell relays are made from one image – a .png with transparency. This is a great way to add detail and keep the polygons low. |
