I have never noticed anything untoward about centre-point focus in any lens I have ever used, apart from this one. [...] I suspect I am not perhaps as intensive about my gear as you
That could be. I didn't use autofocus much during the film era but never noticed anything under a loupe which was unambigously autofocus wander as opposed to PDAF or photographer limitations. Using designed for film lenses on digital bodies didn't change this, at least with the lenses I had. It was with designed for digital long glass on mirrorless that the evidence for autofocus wander became incontroverible since focus assist zooming shows AF point placement more clearly. Prior to moving to m43 I ended up documenting---more or less accidentally---a number of long glass APS-C focus fails analogous to what you've reported in this thread. Some are plausibly attributed to PDAF and microfocus adjust limitations but others occurred with CDAF or hybrid focusing (PDAF with CDAF confirmation). No IBIS but all of them were with OIS lenses and OIS active. Some occurred on tripod, but always in configurations with the ballhead loose, meaning the lens was semi-handheld and OIS would presumably be correcting for photographer induced wobbles. I never noticed autofocus wander below about 300mm equivalent but 1) I also didn't move to autofocus macro until after switching to m43 and 2) I only rarely used shorter in ways in which autofocus wander would be observable.
In general, it seems to me the the systems I've used treat the autofocus boxes show in the EVF or on the LCD as more like guidelines. This leaves the camera body free to focus at points along the outside edge of the box as well as inside the box. It's not something that's likely to be noticed unless DoF is thin and the image critically examined.
As you can see from the three tripod shots above, it does not seem to do the same thing when on a steady base. I did try some handheld last night with both Ibis on and off, and saw no difference either.
Cool. I'm sorry, though, which three are the tripod ones? All sets of three on this page and the previous page seem to be described as handheld. Also, just to make sure I'm understanding properly, no difference with IBIS on or off means the autofocus wander also occurs with IBIS off? If so, that suggests some sort of interaction with handholding. But I don't, as yet, have any hypothesis about why the G6 and G80 would differ in this regard that seems credible.
Have you tried a larger sized auto focus box?
I have and have found one size larger than minimum can give better results with some autofocus systems in some circumstances. That hasn't been my experience of DFD on the G7, though. Sometimes people have problems with AF misses when using small AF points for macro but, in my experience, it's been associated with photographer limitations in placing the point. Pretty sure that's not a trouble for Roddy.
whether you fit it with the clip-on or a step-down ring it seems to do so little for magnification, and so much for narrowing the focal plane, that I do not use it all with this lens
Both are expected, I think. In addition to focus range compression,
- The effective focal length (EFL) of two lenses is EFL = f₁ f₂ / (f₁ + f₂ - d) where f₁ and f₂ are the focal lengths (30 and 125mm for the Panasonic 30mm with a the DCR-250) and d the distance between the two lenses' neighboring principal planes. The choice of clip on or step down makes only minor changes in d which, not having measured the combination, I would guesstimate might change the EFL from something like 17.8 to 17.5mm. As a result, both attachment methods produce essentially the same lens and much the same overall behaviour.
- The effective aperture (EA) of a coupled lens pair does not seem to be well understood at low magnifications (say, less than 4-5x). Given all the element details for both lenses it wouldn't be hard to calculate but that information's hard to get which, so far as I can tell, means no one's tried to develop reasonably accurate estimators for EA and thus depth of field. Among the lenses I've measured, it seems the EA of a coupled pair is 1-2 stops faster than what one likely gets from a macro in the 0.5-1x range.
There are potentially other contributing factors, also not well understood, but even more technical and probably too far off topic.
The greater the zoom, the better the Raynox DCR-250 responds. At least with my P14-140.
Well, that kind of depends on what means by zoom and better. I think what you're saying is that, for a given front lens in a coupled pair (focal length f₂), choice of a longer focal length (f₁) for the rear lens produces greater magnification (M). If so, that's expected, as the maths linked in my previous post reduce to M∞ ≈ f₁ / f₂ when the rear lens is at infinity focus and the principal plane separation d is considered insignificant (
Ingles-Le Nobel). In the case of a DCR-250 and a 30mm lens, this produces M∞ = 0.24, which is an unremarkable accomplishment with most 30mm lenses. With a 140mm lens, M∞ = 1.12. For the coupled pairs I've measured actual magnifications are ~3% percent lower than these predictions. So, if details are important, one might want to reduce these a bit to provide some accounting for d.
Among the autofocus macros and telezooms I've modeled or measured, it seems both f₁ and d decrease as the rear lens is moved towards close focus. As a result, M(some focus distance) ≈ M∞ ≈ f₁ / f₂ has been a reasonable default in my experience of M = 1-3x. Coupled pairs which actually produce this quasi-telecentric behaviour seem uncommon, though. The lower the magnification for a given rear lens, usually the more endocentric the behavior. Higher magnifications appear have a greater chance of being hypercentric.
It also does not take a Raynox - achromats do nothing for it at all. I have to say too, though, that I have taken many photos with other lenses with achromats very successfully on auto-focus.
