On Jan 18, 2012, at 2:19 PM, Robert Purves wrote:

> 
> Robert Covington wrote:
> 
>> It's definitely crashing at the FDecompressImage call (Quicktime) and whatever Apple things it calls it would seem (by the log), but I can't tell if it's my matrix causing that or a bug in the QT call, in Intel platform.
>> 
>> The QuadToQuadMatrix returns no error when creating the matrix. : ( 
> 
> Possibly relevant:
> <lists.apple.com/archives/Carbon-dev/2006/Oct/msg00730.html> 

Thanks, might be a clue. GetQDRectMidX is not described...but if copying out the subimage bounds rect defined by the trapezoidal points to a new source GW might fix it, I'll try that. Just doing a click test where the points were aligned with the window edge (0) did not help.

Since I'm describing an image within the main image (all coordinates positive) I don't see how the baseAddress could be going negative. : (
 
I'm tired these days though. Will probably just remove the feature.

rc