Originally posted by BruceBanner So this is the second time I have heard this, thanks for continuing to educate me. Can you expand on this at all? What does it really mean when we say 'autofocus is rated to f5.6'?
Have you ever used a split prism viewfinder? That's where you find a line on your subject and get the bottom half to line up with the top half by turning your focus dial, and that means it's in focus, when they coincide.
So, the analogy is to the PDAF module above the mirror box. That dedicated sensor has pixels in strips. As I understand it, they have to be long enough for good discrimination between their ends, and the aperture has to be wide enough to get a nice length of the subject under them. For DSLRs f5.6 is something the makers can cope with.
Mirrorless cameras like Sony, Fuji, Nikon and Canon have this tremendous problem, then. They lack a mirror = no dedicated focus module.
Beholder3 calculated IIRC something like 12 percent of the sensor pixels are affected by having to use on sensor PDAF. Because of Bayer, that's the strips themselves and the immediate surrounding ones.
The RAW file is manipulated to cover these blind spots, you get no say in it, copying and pasting like Content Aware in Photoshop. Artifacts like banding can arise as a result, MJKoski has been disgusted IIRC raising shadows at ISO 100 in the Canon RP RAW files.
Panasonic as a result, refuse to use PDAF, saying it affects mirrorless image quality. Fuji use it very sparingly, only one medium format camera has it, and Digilloyd complained: "The Fujifilm GFX100 cannot capture some shades of blue sky without visible pattern noise, due to the inclusion of its PDAF pixels. "
Originally posted by BruceBanner
1) Are we talking PDAF here or CDAF when it comes to this aperture AF limit?
PDAF. CDAF does not use the method of making two ends of a line coincide on paired pixels.
Note that it is important for you when you focus on your subject whether it is a vertical line or a horizontal line. If you compose off centre, the ones on the outside can only focus on horizontals. The ones in the centre can do either, they're 'cross' points. Pentax have historically been pretty good at providing those, which is not appreciated by many Pentaxians. The Canon 6D full frame camera had only one cross point, for instance.
Originally posted by BruceBanner I always understood in PDAF the centre spot and spot directly above and below were 'rated to f2.8', I just took this to meaning they are better more accurate AF points to use, and that to use Spot AF mode (not centre point SEL spot) would actually even increase AF accuracy a little more. I never really thought about the other AF points in terms of f number accuracy.
No, in this case, we are talking about the *widest* aperture they're rated for, not the narrowest. So this means focusing an f1.4 lens is inherently hit and miss, because the focus point is really focussing at f2.8, not f1.4. On average, the focus plane will be the same, but because f2.8 has a greater depth of field there will be more variation acceptable to the sensor before calling it a match.
Originally posted by BruceBanner
3) If 5.6 is the maximum aperture for autofocus, then what is Pentax doing with the PLM 55-300 which has f6.3 as its minimum on the long end? What does that mean? Do we take it that Pentax are not the same as the others and have AF points rated higher than 5.6?
No, it's simply an inferior lens - I have one, by the way, and love it regardless. I also have a Sigma 150-500, guess what, also f6.3 at the long end, and never to be completely trusted.
There are mirror lenses that are f8. Ok, but they're manual focus. Just understand that focus confirmation can never be guaranteed to work on them, because that uses the PDAF module.
Sports photographers pay $10,000 for a 400mm f2.8 prime. One reason is they need all the AF points to work when tracking players. A wildlife photographer can aim for a $12000 600mm f4. They're likely to affect that with a TC, so 1.4 will give them f5.6.
Originally posted by BruceBanner 4) What does this mean when we use say a FA77 at f8? It's less accurate than using f2.8? Are we getting that depth of field trade off where AF accuracy is less but because focus plane is deeper the AF inaccuracies are harder to detect?
No, in taking a shot DSLRs go through a cycle. For the focus stage, they open up to the widest aperture anyway for increased accuracy. This is why I'm baffled to see anyone here suggesting to a member with a problem to troubleshoot by microadjusting at different apertures. Doesn't happen.
Originally posted by BruceBanner It's been more of a gut feeling than anything else, but my experience with the PLM thus far feels like it is sharper at its wide open apertures
Unless the ISO climbs from stopping down, giving more grain, it's user error on your part, Bruce, I'm afraid to say. There is no lens of any brand that is not sharper closing it down a bit. The reason is you're progressively excluding the 'junk' light from the very edges of the elements, the good stuff without aberrations is directly in the centre, along the axis.