There does not appear to be a real convincing consensus of IQ difference with the 1.4x 2 vs. 3. Some detailed measure-baiting has revealed one is sharper in the center, the other sharper at the edges,. I can't recall which.
I am confident that Canon DID make IQ improvements with both of the MkIII T-Cons, but with the 1.4X, I could not tell the difference in use. The 2X seems more improved.
IMHO the difference in IQ is such that it is a subjective conclusion as to whether it is enough to warrant the extra cost. My opinion is that it is not. My opinion is that the reason to go with a MkIII T-Con is to take full advantage of the latest in Canon;'s AF tech.
As for AF, perception aside, there is virtually no way that the MkIII T-Con's can improve AF performance unless it is paired with the correct equipment that will allow the superior AF to function. Sadly that list of specifically equipped lenses and camera bodies is very limited and does not include the older Gen 1 IS super telephoto lenses. We can not expect a lens designed and launched in the 1990s to be able to work with the latest Canon AF system advances of just the past few years. Therefore, if you are missing any one part of the three ingredients needed to achieve improved AF function, (T-Con MkIII + latest camera bodies + latest lens) then you can not get the new fangled AF. It's easy to tell, count the active AF points @ f/8. (f/5.6 lens with 1.4X, or f/4 lens with 2X)
If you get all of the AF points, you are in AF nirvana. If not, then you have old school T-con AF with the brakes applied.
Back up a few years, from the time Canon has had autofocus lenses and teleconverters, Canon has imparted an intentional decrease in AF speed from the moment a t-Con is mounted. Canon AF has preferred to forgo speed in the name of AF accuracy where T-Cons are involved. Similar to it's AF @ f/5.6 only bodies, Canon has preferred to remain conservative as to how much light it's bodies require to try to AF at all, and with T-Cons at what speeds it will try to achieve AF. toss on a 1.4x and drop your max aperture from f/4 to f/5.6 and the whole system has always automatically shifted down a gear. Precision over speed. Put on a 2X, it shifts down two more gears, and only allows the center AF point. (on older bodies that even allow AF @ F/8. Most bodies, no AF at all)
It was not until the 1DXII, and the very latest lens/T-Con combos that Canon has had AF quality taken to a high enough level to discontinue the "AF Handbrake"
Yes, some slowing can still occur, but this is due to actual lighting or contrast issues, not Canon dropping the AF speed at the mere presence of the t-Con.
When the MkIII T-Cons first came out, the reports on this forum and elsewhere of improved AF spurred me to do a lot of testing on my own, and frankly the results all flew in the face of any perceived (but never actually "measured") improvements in AF that people were casually reporting. Likewise I've read no measured reviews from trusted sources that have the ability and time to actually measure such things that state or even imply that one can garner the AF speed improvement without the entire food chain required to do so in place.
EG:
- Even with the correctly equipped lens, the new 500mm IS II and a T-Con MkIII there was no real AF improvement until I also had a 5D4 to pair it with.
- With the EF 500mm f/4L IS version 1 side by side with three bodies swapping MkII and MkIII T-Cons, no real perceivable difference in AF.
Get all three together, in my case 5D4, EF 500mm f/4L IS II, and 1.4x or 2x MkIII
Everything changes!
You get all AF points working @ f/8 and there is no immediately noticeable AF slow down. Canon's old "T-Con hand brake" is no longer there. In good lighting, AF appears to be as fast and confident as the naked lens.