alundeb wrote:
Exactly. The time it takes for the diaphragm to close should be sufficient to make a small correction in focus without a noticeable delay, even in Tv mode where the aperture is not necessarily known at the time of focus lock. The correction could also be made focus distance dependent, as the focus distance is known by the lens, it is by the definiton of autofocus.
Canon, make the MA is a function of subject distance please. Not only your problem child will be fixed but all your things will perform better.
Ed Sawyer wrote:
Canon could fix it electronically with a lookup-table inside the lens that refocused at time of exposure based on reference data. But it would require more time and effort than they probably feel would be warranted.
alundeb wrote:
Exactly. The time it takes for the diaphragm to close should be sufficient to make a small correction in focus without a noticeable delay, even in Tv mode where the aperture is not necessarily known at the time of focus lock. The correction could also be made focus distance dependent, as the focus distance is known by the lens, it is by the definiton of autofocus.
mttran wrote:
Canon, make the MA is a function of subject distance please. Not only your problem child will be fixed but all your things will perform better.
That alone won't work. Not just subject distance, but also aperture. I think as controversial as this lens may be, nobody will dispute that there is _no_ shift at f1.2... regardless of subject distance.
Or, alternatively, we could ask Canon to please fix this damn lens. But that's just me.
stanj wrote:
That alone won't work. Not just subject distance, but also aperture. I think as controversial as this lens may be, nobody will dispute that there is _no_ shift at f1.2... regardless of subject distance.
Or, alternatively, we could ask Canon to please fix this damn lens. But that's just me.
At the risk of repeating myself, there is no focus shift at f/1.2 because focusing (whether manual or AF) takes place at f/1.2. Focus shift is only observable when:
1. The exposing aperture is not the focusing aperture.
2. The DOF is sufficiently narrow that missed focus is evident.
Regarding the compensation for focus shift via some kind of modified AF algorithm, it may be possible, but let's be very honest here: the practical implementation of AF as it stands presently is such that there isn't actually as much accuracy as you might think. Talking about a more sophisticated algorithm to compensate for focus shift presumes that the present AF system is capable of that level of precision in the first place--and we know that this isn't always the case. The accuracy is also dependent on ambient light levels and the contrast presented to the focus region.
Thus, the better solution, in my view, is to either live with some small degree of focus shift and adapt your camera technique in those cases where it could present a problem; or, design lenses without focus shift and accept the way such designs render blur circles. And as I pointed out earlier, the focus issues for the 50/1.2L seem to be about overall AF accuracy, not focus shift per se. So asking for a more complex AF system is a bit like putting the cart before the horse, when what is really needed is a more consistent and precise AF system.
wickerprints wrote:
I must take pains to point out that focus shift cannot be rectified by changing the way the lens autofocuses--indeed, for a manual-focus lens with focus shift such as the Zeiss ZE 85/1.4, the user can find critical focus with Live View and still miss when the exposure is taken, because focus shift is caused by the change in plane of sharpest focus due to the focusing aperture being different than the exposing aperture.
In liveview, you have the option to stop down while focusing. This eliminates focus shift. I always do this when focusing is critical on any lens.
stanj wrote:
That alone won't work. Not just subject distance, but also aperture. I think as controversial as this lens may be, nobody will dispute that there is _no_ shift at f1.2... regardless of subject distance.
Or, alternatively, we could ask Canon to please fix this damn lens. But that's just me.
wickerprints wrote:
At the risk of repeating myself, there is no focus shift at f/1.2 because focusing (whether manual or AF) takes place at f/1.2.
Yes, you are repeating yourself. _I_ understand that, yet you are replying to me. _I_ was replying to mttran, as my quote clearly shows, who said that if the adjustment were focus distance governed, the problem would go away. I was just pointing out to him that this would not be sufficient. So, please read more carefully - I agree with you, and from my billion posts about this lens in other threads, as people seem to realize, I do understand the physics and limitations.
I don't think the focus algorithm in general is the stumbling block, in AF in general. So, theoretically, a lookup correction table could be incorporated in the existing AF algorithm without necessitating redoing the entire method itself first. IMNSHO of course.
Basically a 2-D array of micro-adjustments based on distance and aperture would do it. It could probably even be coded into a complex formula and not require individual data points to be looked up.
I recently got the 50mm 1.2 (with datcode UY) to replace the Sigma 50mm1.4. No focusing issues so far. I have shot with different apertures and distances, so far I'm happy with the results.
wickerprints wrote:
At the risk of repeating myself, there is no focus shift at f/1.2 because focusing (whether manual or AF) takes place at f/1.2. Focus shift is only observable when:
1. The exposing aperture is not the focusing aperture.
2. The DOF is sufficiently narrow that missed focus is evident.
Regarding the compensation for focus shift via some kind of modified AF algorithm, it may be possible, but let's be very honest here: the practical implementation of AF as it stands presently is such that there isn't actually as much accuracy as you might think. Talking about a more sophisticated algorithm to compensate for focus shift presumes that the present AF system is capable of that level of precision in the first place--and we know that this isn't always the case. The accuracy is also dependent on ambient light levels and the contrast presented to the focus region.
Thus, the better solution, in my view, is to either live with some small degree of focus shift and adapt your camera technique in those cases where it could present a problem; or, design lenses without focus shift and accept the way such designs render blur circles. And as I pointed out earlier, the focus issues for the 50/1.2L seem to be about overall AF accuracy, not focus shift per se. So asking for a more complex AF system is a bit like putting the cart before the horse, when what is really needed is a more consistent and precise AF system....Show more →
Simple solution from Canon - a selectable custom function that would, with certain fast lenses, all the camera to activate 'DOF preview' while focusing when operating in Av or M and using aperture settings where focus shift is likely to be noticed (say, the f/2 to f/4 range or whatever). That way, the camera focuses at the aperture that will be used for the shot.
Not on my 7D or 5D2, no. Doesn't mean that the AF can't work with the aperture "stopped down", at least within the range where AF will work on the camera (i.e., f/5.6 or larger on non-1 series bodies).
- but the 50/1.4 is reputed to focus shift back and forth in sequential test shots. Sometimes. Maybe....