Accurate enough for what? They could well be derived from completely different methods, so the one thing you can't do is accurately compare between manufacturers. You also can't compare different focal lengths! So whilst manufacturer presented MTF graphs might be "accurate enough", I was just adding a cautionary note to those who were comparing Nikon figures to Canon figures. Just like MPG estimates, they have very limited value and shouldn't be a serious basis for a purchase.
Lensrentals now has the equipment to measure lens resolution without being mounted (to a camera as we know it) so they can, at least, give genuine comparisons for lenses across manufacturers.
MTF is definitely not the final word, but its a starting point, and a lens is not going to be better than the manufacturers posted MTF. The values that the camera manufacturers are good enough precisely because of this, there is no use being accurate to 5 or 10%% because it is only part of the story.
Do you have a link to Rogers optical bench and colliminator? as well as the PHD's he has hired to run it? The last time I saw any information was that he used Imatest like all the other testers.
Measuring the MTF of a lens with a optical bench, slit illuminator, not only takes 500K of equipment, but also needs some very skilled technicians to intrepret the results. I had all of those in my lab at a large aerospace company, and it was a big investment assembled over several years of R&D budget. We had several lab techs and engineers / scientists using it. Its not something easily done.
Somehow, I wonder if Roger has acquired this ability, or even why he would want to, since Imatest is plenty good for what he does. But please give me a link and impress me.