Dynamic choice of resolution used for image feature extraction by the synther, based on load provided

Perhaps the Photosynth team would rather keep things the way they are for gauging the results of improvements made to the synther, however...

I'd like to see the synther take a quick inventory of the CPU, RAM, and hard drive space available to it, the number of photos given it when the 'Synth' button is pushed, and the resolutions of all of those images and use that data to dynamically alter the resolution at which it extracts image features.

If it's a 40 image synth, then why not use full resolutions for feature extraction, matching, and scene reconstruction?

Currently many smaller synths, in many cases from new users (although I don't mean to suggest that only beginners make smaller synths), suffer from not having much of a pointcloud or fracturing into pieces due to novice ideas of what enough overlap is. Why not give them a boost, compared to larger synths, and let them generate all the pointcloud that they are capable of?