ATJ wrote:Gary,
Here are a couple of threads:
Yes, there's a lot of crap in the DPR threads. Nothing really conclusive in the threads you're pointing to either.
Let's go through what I've just done. First of all, let me declare my prejudices: I believe that most times when people see focus issues - including backfocus - I tend to believe that it's a user error.
Similarly, when I see images of mine that I believe to be soft, my first thought is the same: user error. It's a problem with technique, and in that instance, that would be my technique.
I started by setting up the light tent and tripod. I've used a typical backfocus chart for this, if only to have a somewhat controlled - but contrived - target for the focusing.
Not wanting to stand too much on ceremony, things were set up to inexact tolerances - near enough being good enough. When, in doing a real shoot with a DSLR, does one accurately measure angles, distances, etc? Sure, using a LF camera and wanting to take account of Scheimpflug, you probably would. I'm not, and I didn't.
The main thing I was trying to determine was if there is an issue in focusing when comparing shooting using matrix metering, vs spot metering, because that is where I have heard of issues, and that is where I believe I may have even seen some issues.
So ... the main thing was to shoot a couple of pairs of similar images: 50mm f/1.4 @ f/1.4, matrix, centre focus position vs spot, centre focus position, and matrix, off-centre focus position vs spot, off-centre focus postion.
My expectation was that the critical images would be these lattermost ones, and they're the ones that you see here.
50mm f/1.4 @ f/1.4 matrix, off-centre focus position
50mm f/1.4 @ f/1.4 spot, off-centre focus position
In pixel peeping at these two images, there seems to be no real difference. A very slight difference in terms of the actual point of focus, but that can be easily put down to the manner of the tests I performed.
Certainly, the differences were not of the order of a failure to acquire focus, or focusing in the entirely wrong location, which is what I was anticipating, based upon what I'd been seeing thus far.
The differences I'm talking about amount to mere millimetres in terms of where the actual focus points fell. That does not, to me, display any evidence of any focusing issues.
So, I needed to revisit what I saw on Monday last. Here's two images from that session ...
The second of these images is markedly sharper. It's shot using spot metering, the former was sot using matrix metering, and these are adjacent images from the night. FYI, focus point on the first was on the lady to our left (Rinske), on the second it was on Freddy, the trumpet player to our right.
Looking more critically at the images (compared with when I'm actually shooting at a jam) we can also see that there's a significant variation in the exposure settings applied to the image. This is expected, based upon the conditions and changes I made.
Note too that Freddy's hat is blown in the first; the second image seems better exposed to me, suggesting that matrix still cannot cope too well with this sort situation.
What I failed to take into consideration last Monday was what the actual changes were: I accepted that changes would apply, but i didn't think about them, about what was actually changing. Silly me!
Both images were shot through the 85 f/1.4 @ f/2.8, -1.3 EV. To get the different exposures then, the shutter speeds must, obviously, be different.
1/40 in the first, 1/320 in the second!
And with that sort of shutter speed in the first image, I am exposing my images to the risk of camera movement and/or subject movement. In this instance, both.
At which point any discussion of problems with the focusing mechanism of the camera becomes moot: all bets are off.
Looking closely at the first of these images, I can see evidence of movement within the image, and any problem here, clearly lies with the user.
So, from my perspective, I am satisfied that there are no focusing issues of a generic nature with the D300 - as I mentioned earlier, we did this test using two different bodies (Leigh has an eval unit from Nikon Oz right now) and they both behaved consistently with the backfocus chart.
That takes care of the focus issues, AFAIC.
Let's return now to the metering: I've mentioned in other threads that matrix metering seems greatly improved on the D300 when compared with other
models I've played with. I still hold this view, but it's still evident to me, and for many of the images that I may be shooting, that I will still need to use spot metering
mode.
The extent of the blackness in these two images seems to me to have affected the metering. This is an expected outcome,
btw: there's a lot of dark areas and high contrast when you're shooting most types of stage scenes, and from what I'm seeing here, the metering system is trying to accommodate and compensate for the dark areas, rather than accepting that they need to be dark, and thus discount them from the metering calculations.
Hence, spot metering is, for me, still the way to go.
Finally, and back to focusing, this camera's focusing still impresses me. Ok, it's not even been two weeks, but it focuses bloody fast, much moreso than the D200, and the D70 isn't even in the same game.
And I note in one of the threads a comment about the camera "not focusing" at all ... from what I'm seeing, the camera already seems to be in focus at that distance, and the issue is more that the camera doesn;t need to refocus itself if it has determined that the subject is already in focus. A quick insertion of some interference into the sight path confirms that the focus mechanisms have not fallen asleep, so it's simply a matter of the camera determining no need to fix what isn't broken, so to speak.
That about sums up what I've done with this today: nefs are big - 10-12MB each - and live beside the jpgs that you see; just substitute NEF for JPG in the filename to grab them, noting that the extensions are all uppercase.
And returning to me previously declared prejudices ... they remain unscathed.
HTH; cheers.
Posted: Sat Dec 08, 2007 11:48 am