As far as I know, the scanning of a film in the telecine machine yielding a video signal of the full pixel count desired is a one stage geometric process. Whatever content is on the film is desired is matched up with the scanner mechanism depending on whether or not the film is anamorphic and whether or not the DVD to be pressed is to be anamorphic.
Let's say we are putting through a 1.85:1 anamorphic film. If we adjust the telecine so that the entire film frame is covered by the telecine scanner (more correctly scanning a little past the top and bottom edges), we immediately get a correct 16:9 or 1.78:1 optimized anamorphic video signal. Although all anamorphic DVD's are optimized for 1.78:1, anamorphic films are more often optimized for 1.85:1 so the correct telecine setting is not exactly the film frame.
Let's say we are putting through a 1.85:1 hard matted (prerecorded black bars) spherical (non-anamorphic) film. If we adjust the telecine to scan the entire film frame we get a 1.85: letterboxed video signal. If we adjust the telecine to scan just the picture portion (actually a wee bit more on top and bottom) we get a correct 16:9 optimized anamorphic video signal.
No more scaling loss occurs in any of the above scenarios compared with the others.
Whether the desired part of the film image is optically stretched or zoomed to fit over a fixed scanning window or whether the scanning is optically zoomed or stretched (no change in pixel count) to cover the desired part of a fixed size projected film image depends on the make and model of telecine. After the scanning is done, any extraneous material captured (such as the bottom edge of the frame above or top edge of the frame below) is blacked out in terms of scan lines or pixels.
Video hints:
http://members.aol.com/ajaynejr/video.htm