 |
brainiac
Offline
[X]
|
 p.4 #7 · Is film really going away??? | |
> The ground glass for manually focusing an AF DSLR with an off-center subject usually leaves much to be desired, especially with wide angle lenses. Accuracy is not grea, and it gets worse with slower lenses...
I find accuracy of ground-glass off-center focussing very good on my 5D and 1Ds3, using Canon's Ee-S and Ec-S screens. With slower lenses it requires concentration, but at least it can be done. On a rangefinder there is no way of focussing your subject if you want to put it off-center, so I'm afraid it's really one-nil to the SLR. Bearing in mind that I almost never put my focussed subject in the middle of the picture, the rangefinder method almost guarantees misfocus, which increases the further from the center the subject is, and the wider and brighter are your lenses.
>...With any camera though, you just need to be aware of the inaccuracy that a focus-recompose technique can cause and compensate for adjust for it.
Being aware of it is not enough. The mis-focus varies depending on how wide your lens is and how far from the center your subject is. Those two continually changing variables make it almost impossible to use dead-reckoning to adjust accurately when using a lens at f2 or f1.4.
> ...a great deal of the famous photos taken with Leicas and other RF do show that focus isn't one of their strengths. But this also shows that sharpness isn't everything to make a great photo.
OK. Accurate focus isn't important to you. I get paid to take pictures. Lots of them are out of focus, but I do try my best to get focus as good as possible. A camera which had no way to accurately focus at wide apertures would be a severe limitation. Why pay so much for a camera that can't accurately focus off-center subjects?
> With a SLR and a fast lens distracting objects will be blurred out while you compose and during a rapid shot you may not account for them...
That's a great point - thanks for raising it. I mostly use alt lenses on Canon, and manual screens which give a very good approximation to real depth of field, so in that case the point doesn't apply, but I agree that with auto-aperture lenses, depending on the set aperture, SLRs can be _as_misleading_as_a_rangefinder_ by not showing the depth of field. They do have d.o.f. preview for that reason though, and liveview works well with d.o.f. preview too. A rangefinder doesn't even have d.o.f. preview.
>...a SLR still is not WYSIWYG so you still have to rely on knowledge and experience...
...unless you're using a Canon manual screen, alt lenses, or d.o.f. preview, in which case the viewfinder really is about AWYSIWYGAYCG (As WYSIWYG As You Can Get).
> Plus, a lot of wide open shooting with fast (1.2 and faster) lenses would be in low light - a situation where an SLR is at its worst with a dimmer viewfinder and the practically impossible task of using a matte surfaced focusing screen.
I shoot with bright lenses in very low light all the time, and I don't find it practically impossible at all. Here's an example with Oly 28 f2 at f2, and 5D at iso 3200, where a rangefinder could not have accurately focussed what my SLR did:
http://cyberphotographer.com/5D/nathanlo.jpg
> ...field curvature is not designed "into" the lens, rather they are not designed "out" (meaning corrected). And any problems you would have would be seen as focus being in front of the subject, not behind it. And in general, RF lenses seem to be much better corrected, in my experience.
I didn't say curvature was designed in. Field curvature is not the problem. The reason why a rangefinder can not focus off-center subjects is because the lens has a PLANE of focus. Lens makers prefer to make planar lenses so that brick walls look uniformly focussed. As a result, the center of the lens circle focusses at a _different_distance_ than the periphery. If you point your RF patch at the subject, and the subject is six foot away, and then recompose to place your subject nearer the side of the frame, then the lens is focussed at let's say 8 foot (in the direction of the subject), while the subject is still only 6 foot away. At f1.4 or f2 that constitutes a serious back-focus problem. On rangefinder the center patch is the only way of confirming focus, and in those very very common circumstances it is completely broken by design. OK - you could draw a line along the ground in the direction of the lens axis, and mark the distance on that line from which a perpendicular bisector passes through the subject, but I promise you that you will be less accurate and much slower than me and my SLR, which allows me to focus planar lenses on off-center subjects by eye, with no less accuracy than it allows me to focus the center of the frame.
Edited on Jul 18, 2008 at 06:43 PM
|
| Jul 18, 2008 at 06:38 PM |
| |
|
 |
), can we focus on whether it is a practical problem? 
