*****************************************************************************

    FAQ  -  $Revision: 1.3 $ $Date: 2003/01/21 14:55:58 $ 

*****************************************************************************
------------------------------------------------------------------------------
Q: When try to run AAMC I get: "could not excute: invalid directory(...)"
------------------------------------------------------------------------------

A: Nothing is wrong. Just change the settings under 
   "Project->Settings>Win32 Release->Debug" and 
   "Project->Settings>Win32 Debug->Debug" to reflect your directory structure.

   If you dont't you will get this error during Excute or Debug:

   "could not excute: invalid directory(win32 error 267)."
  

------------------------------------------------------------------------------
Q: My texture model looks all scrambled when I try to build my own model. Why?
------------------------------------------------------------------------------

A: Check that all input images are in 8-bit grayscale. 

   24 bit images will usually be automatically converted to 8 bit ... but not 
   always. It seems that the BMP support in VisSDK is somewhat buggy. So, if
   everything fails try convert the image to 8 bit grayscale (linear) and 
   open and save the image from Windows paint. This should produce BMPs that
   VisSDK can read.

   You should also remember that the unmodifed VisSDK, and thus the AAM-API, 
   can only read BMP files. For other file formats refer to the optional 
   howto's on this subject.


------------------------------------------------------------------------------
Q: Why do I run out of memory during model building?
------------------------------------------------------------------------------

A: First of all, the ACF option "Training method==0" uses a *lot* of RAM.
   Avoid using this option on anything but very small training set. 
   If you do anyway, your RAM usage will depend on:

        i)   The number of training examples
        ii)  The number of texture samples
        iii) The number of model parameters          (see the acf file)
        iv)  The amount of training set subsampling  (see the acf file)