DotTune: New AF tune technique, no photos required
CanadaMark wrote: snapsy wrote: CanadaMark wrote:
You are misunderstanding me I think. He\'s getting a value of -19 on one end, and +4 at the other - using a midpoint of -7 or -8 is nowhere near precise.
I understand you and you\'re incorrect. The phase detection feedback provided by Nikon bodies via the rangefinder is a continuum of phase measurement values. The fringe of AFMA values that yield a solid confirmation represent the the largest measured phase differential for which the camera still believes produces relatively sharp photos. The purpose of having a range rather than an abrupt single value is to make manual focusing to the rangefinder easier; this is also why the range of confirmed values grows about about 2.5x when the camera or lens is in MF mode, again to make it easier to bring the lens quickly into near focus. Since the range of confirmed values (both +/-) represents a continuum of phase measurement values, the midpoint of that range represents the optimal AFMA value. The absolute size of the range has no bearing on the accuracy of the midpoint - it\'s simply a function of the focal length and maximum aperture of the lens. For Nikon\'s AFMA in the D5/D500, using the full confirmed range is not necessary since the camera is not required to provide relative feedback like it does in the rangefinder - instead the camera simply selects the AFMA value with the minimal phase detection differential, which yields the same result as the midpoint of the confirmed range used by DotTune.
CanadaMark wrote:
The only precise way to do it is currently impossible, which would involve a separate value for every combination of FL and distance. Nikon\'s automated implementation does not solve the problem either except for a single data set. If you have to max out AFFT to fix your lens at any combination, you should be exchanging it because you will never zero it out across the entire range. Further, if you have to settle on some middle-ground value that makes your lens OK everywhere but never great, that\'s not a proper solution either. Viewfinder \'tuning\' has been around since the D3 in 2007 and it has the same problems today as it did back then. Until Nikon opens up more data points, you\'re always going to be better off exchanging the lens for one that requires no AFFT. Sigma has the right idea with the dock, but even that is an extremely painful process.
You\'re conflating the frequency of distance-specific AFMA variation with the underlying usefulness of the mechanism itself. I\'ve tuned 30 lenses across 10 Nikon bodies and of those only 3 produced material distance-specific variation, and two of those were third-party bodies (Sigma ART 35mm and 50mm, both of which were correctable using Sigma\'s dock).
We will have to agree to disagree, and I still don\'t think you understand what I am saying based on your response.
And to get more specific, you described why taking the midpoint of a large confirmed AFMA can\'t produce reliable results. I provided specific technical information about what your statement is incorrect. Rather than address this you reply that my response indicates that I don\'t understand what you\'re saying, then finish it that I\'m being misleading. As always I\'ve been very open to objective technical discussion about the technique but one thing I can\'t counteract are opinions based on feelings rather than an objective technical basis.
Jun 21, 2016 at 02:55 PM
Previous versions of snapsy's message #13613147 « DotTune: New AF tune technique, no photos required »
