Joojaa,
Thanks for the info and tips. I'm running WinXP SP3 and my HD is formatted NTFS.
I thought about the idea of breaking up the render into smaller sections and blending them in over several frames after I posted. If the fg map causes the render to die I can always do this. BTW, the fg map I'm currently testing with sections of animation has grown to 186 MB with FG rays = 1000 for about 300 frames (1280 x 720).
A couple of related questions on how the fg map actually works.
I notice that the fg map grows slightly with each frame, but not nearly what the original fg map was for the first frame. What is mr actually doing when it appends to the fg map when fg rebuild is off? If the initial fg map is for 1000 fg rays then it is not casting another 1000 fg rays for the second frame and adding it to the initial 1000 fg ray fg map. I believe more fg rays would cause the render to be brighter (as I've seen when I compare the same scene rendered with fg rays = 100 to fg rays = 500).
This also brings up an issue for rendering the initial fg map: does it matter what is in the scene for this initial fg map render? I've taken the precaution of making all objects invisible except for the lights and walls of the environment for the initial render. I assumed that trace depth (=2 in my case) might give different results if there is a lot of geometry in the scene. However, since the fg rays are cast from the camera (which is stationary for my scene), and the fg map is "modified" or added to for each frame this might be an unnecessary precaution. So, if the camera is not moving, would it really matter what is in the scene for the first frame?
Thanks again for the help you've given me and many others over the years.
Matt