well I use maxscript (3dsmax scripting language) wich is the same as actionscript for flash except that the syntax is different.

I model my scene (low and highpoly) render the high poly stuff and export the low poly ones. After that I grab my camera positions and put them into my flash engine. Here is a screenshot of a technical demo scene I am working on:

and this is how the camera looks like
so all I need from the camera is the X,Y,Z position, the X,Y,R rotation and the the FOV. This can be automized with maxscript but it only works well if my coordinate system is the same as max uses.
As for the units I use generic units in 3dsmax so a top view cam in my engine show exactly 128 pixels*128 pixels if a cube exported with 128³ is pushed to flash. Btw this is part of my engine:
http://board.flashkit.com/board/showthread.php?t=759766

I am pretty sure that back then with Resident Evil they did the same some MEL (maya scripting language) oder maxscript to get additional data to their engine and render as they were used with the rendering software.