So much great new research data is available lately! Reikan Focal Online Stats.
Between Roger's LensRental blog findings on AF, and now a really great gem of a find! Reikan Focal MFA statistics!
My thoughts...the 50mm 1.8 is a lens lottery lens like I thought it was. Rather soft wide open, copies vary a lot...AF is pretty much crap at wide-open IMO (ok the graphs don't literally translate to that but there are a lot of people fiddling with MFA for this lens).
Likewise the Canon 85mm 1.8 does absolutely stellar (as does the 1.2)...and for all the hoopla of the Nikon 85mm 1.8G being more 'modern'...just take a look at the results, still a lot more lens lottery being played more on the Nikon side than Canon IMO...besides our Nifty Fifty which is the black sheep of the Canon family in lens lottery, I think Canon lenses do fairly well. Most of the lens I own score high...including the 40mm f2.8, another highly consistent, accurate, and not needing MFA lens...and according to charts is *bloody* sharp wide open.
It also seems to indicate to me that there is not a *huge* difference in AF accuracy for the lenses I use between the 5DMKII and 5DMKIII if we are just talking center point focus and recompose...AF accuracy seems to be stellar as long as that given lens is stellar. We do know that the 5DMKIII + certain new lenses get more AF magic sauce, but actually for the 40mm f2.8 which supposedly has it, the statistics don't show a big difference there IMO (although I guess the 5DMKIII does seem to have the better wide-open consistency on the 40mm f2.8).
There is a link on the site that tells you how to interpret their findings. IMO with Canon's push towards Robotics, Canon's QA/QC consistency with lens is only going to improve...yes the 50mm 1.8 is an absolute stinker in this regard, but most everything else does very well and especially the new stuff (although the 85mm 1.8 is 20 years old and does just as well in consistency as the very new 40mm f2.8 pancake).
Preliminary results seem to show that Nikon not only is clearly winning in dynamic range, but in the area of dynamic focus and sharpness variation as well.
Does absolutely confirm that the 5DMKIII improves accuracy and reduces variability across the board...very impressive. At the same time the 5DMKII is not as 'terrible' as people like to claim especially when compared to other brands and bodies.
I know my 6D is where it should be with regards flange to sensor distance, whereas my 5D2 was slightly out, meaning my manual focus Voigtlander lens needed a new focus mark adding.
All my lenses are a much tighter fit on the 6D also & none have needed MFA or recalibration (including the Canon 50mm f/1.8 & 40mm f/2.8 pancake).
Yes I would have to guess the statistics target the market audience for that software. Meaning most people willing to buy that software are probably *serious* gearheads who care about critical focus to the nth degree and are the type to MFA all their lenses. I MFA none of mine and feel they are all rather good to excellent copies. Regardless if there is a good deal of bodies/tests for certain combos (for example kit lenses have lots of statistics) and it does seem to hold tidbits of data you can pull. If Reikan Focal delivered a free version of their software, they might get a much larger influx of data to pull from.
For example some online tests have tested the 40mm f2.8 to be absolutely superb wide-open, and some have tested theirs as needing to stop down. These statistics on Focal Reikan seem to show that both are valid results...*some* people get 40mm f2.8's that are bloody sharp wide-open, and some get copies that need stopping down according to the relative sharpness per aperture ribbon. But overall the consistency of that lens is an order of magnitude higher than the 50mm 1.8 which is just tremendous in sample variation (comparing ribbon widths).
I also believe the only thing that matters is how much confidence *you* have in *your* gear and the level of results you get. I personally believe the results I get with my 85mm 1.8 and 40mm f2.8 or 17-40, etc with my 5DMKII (recently manufacturered), is very high. But then again those results clearly show that I should expect that for that combination of body/lenses as on *average* it is delivering close to ideal focus 98% of the time with some variation 5-7% range However the 5DMKIII also clearly cuts down on AF variation in most any lens by almost half except the 40mm f2.8 (which coincides with my eyes as the 40mm f2.8 is rather extremely consistent in focusing). However I think for the price difference, the 5DMKII is clearly 'ok' and obviously its been a workhorse camera for many years...it also just shows that if perfect focus is something you desire, the 5DMKIII absolutely delivers. It also seems to deliver far more precision and improvement on the extremely fast L primes...but for the regular EF primes the improvement is more modest (except the improvements are rather huge for the very popular 50mm 1.4). I think this coincides with a lot of peoples anecdotes in their experience with the 5DMKIII.
On the Nikon side looking at the results it seems some people can be wonderfully blessed and can get body/combo's that are superb in sharpness, but on average can also be much lower. What is interesting with the variation ribbons over all for some lenses I've seen is that Canon's 'average' aperture sharpness is towards the top of the variation ribbon, and Nikon's 'average' is towards the bottom of the ribbon. For example a D800 + 85mm 1.8G, few are blessed with copies that are bloody sharp wide open (greater than 90%), but on average wide open you get 70% of the IQ you get stopped down. Canon's 85mm 1.8G shows that its clearly softer wide open than at f4 obviously, but that overall people get more predictable performance from their lens as well as AF.
Thanks for the link. Im no pro and looking to get the 5D II I think its still a quality camera.
Well a few things seem obvious, and that is the 5DMKII for all the heat it gets for being a terrible AF camera, for the most part has accuracy comparable to the 7D (not taking into account tracking however, just center point focus recompose). The results coincide a bit with LensRentals. The 5DMKIII and 1D series are at a higher level (although it seems a lot of 1D are optimized for telephoto zooms and the benefits for shorter primes seem less).
It does seem to confirm a lot of forum chatter...in that Nikon's previous D300/D700 had very very solid AF with very little variation and high accuracy, and with the D800's variation may feel worse than the 5DMKII especially to the pixel peeper. However how much of the extra variation also comes from the extra resolving power may also be at play. I'd have to guess that the extra MPs easily makes up for the slightly higher variation to deliver better results most the time, however for the casual family events and candid shooter, I don't think it makes sense to buy a 36MP camera like the D800 and have a lower 'keeper' rate where obviously you do not utilize anywhere close to the full extent of its 36MP files, vs the 5DMKIII which seems to show you will nail focus most the time with most lenses.
I know doing PP work myself, I'd rather have the smaller files where 97% of my shots are tack sharp, than have massive 36MP RAW files where some are clearly OOF compared to others.