More rambling.... In Hollywood, back in the old days, they would use the painted backgrounds but would film inside a studio. Do graphics engines ever use this concept? I've often wondered how they display distant backgrounds as well as they do. I imagine it might be like this... The" studio" area might be like a 100 or 200 meter radius sphere with the viewer at the center. Inside that area, it is 3d. Beyond that area, the engine creates a background photograph by rendering the background in 3d and taking a photo. Then it maps the photo onto the "studio" sphere. Periodically (frame rate) and/or for logical reasons it generates this 3d background. A separate part of the engine handles the "studio" area. Another part handles the "background" area. The background could be layered. Imagine there are two layers. Layer one is medium range background and layer two is long range background. I am being abstract. Inside the "studio" area, the 3d scene is handled normally at a high framerate. In another process, the medium background is handled at a lower frame rate. In another process the long distance background is rendered at even a lower frame rate. Perhaps as slow as once every 3 to 5 seconds. Periodic updates to the background is the frame rate. That's the target rate for updates to the background in terms of time. However, the background engine could also know which direction the photo faces relative to the viewer and what direction it was facing when it was originally taken. If the viewer is moving at a rate of speed that causes the photo to face away from the viewer beyond a certain amount of degrees before then next frame is scheduled, it could render the background in 3d and create a new picture and re-adjust when the next scheduled frame is to be rendered. A very sophisticated background engine might be able to comprise the background scene of many smaller pictures all mapped to the inside of the "studio" sphere and update them at framerates that make sense for each one. For instance if there was a fast moving boat in the background, the engine might update that area of the background faster. Well, it might mean you could take a really high res picture of the background and update often enough that it seems believable. Or for a game like eq2, when an object "pops" out, you generate a 2d photo of it and map it onto the sphere. Then you can still treat it as a solid object for shadow purposes.
