Hi Todd,
Actually, with a P40+back shooting a full frame 35mm transparency, the arithmetic makes it an effective 35MP raw image (7320/1.5)v*7320h = 35.7MP. The important thing though is that I've now got the resolution perfected to 5161 PPI based on photographing the USAF, looking at the center and all four corners. Alignment is really a delicate business but it works. This is really very close to Imacon and good drum scanner territory. You are right that the capture process is much faster than scanning, but there's stuff in front of pushing the button that consumes some time. An alignment check should be performed at frequent intervals, especially after changing any magnification or positioning of the media, and it is necessary, at least with this camera, to allow a good 20 second interval between the mirror-lock-up and the capture. This wouldn't apply to mirrorless cameras, but it applies in spades to a 645DF or any other mirrored DSLR. This sensor is so sensitive at these resolution levels that ALL vibration needs to be excised before capture. Effective raw image size can be increased substantially by photographing the 24mm dimension of the media magnified across the 44mm dimension of the sensor in two captures, one displaced from the other so both ends of the film are photographed and blending them. I haven't tried that yet to see exactly how well it works and what raw dimensions I end-up with, but should be interesting.
Capture One does have an invert function, but only in the version of the application targeted for the conservation/institutional market and it costs about 2500 dollars, so I'll stick with my established workflow for now. I would really like to see Adobe add the invert function into LR as you mention, because if it were possible to confine the whole workflow to Lr it would save a huge amount of storage (a rendered P40+ file is about 220 MB in 16 bit) and cut the workflow time substantially - let's hope.