snapsy
Online
• • • • • •
Upload & Sell: On
|
 p.15 #16 · Z8 extremely unreliable/inconsistent AF in mildly low light | |
SCoombs wrote:
I have finally managed to record my own videos. Mine are not as well "organized" as Snapsy's, largely because my models are, as you will see, a bit less cooperative. On the plus side, this means they better model a more realistic shooting scenario.
I tested this at 3 apertures: f2/8 and f4 on my 70-200S, and 1.8 on my 85 1.8S
Due to the fading of exterior light, EV varies in these tests from a bit over 4 to around 3, if I've done the calculations correctly. Anyone curious to double check that for themselves can use the values as recorded in the video to check.
Note that sometimes in these videos I zoom in to show the focus, but often I am just doing it to check the focus, so don't take my zooming on any given shot as a claim of anything unless you actually see the focus being wrong. Usually when it is I will zoom out to show the focus point and then back in again.
My observations are as follows:
- Most notably to me, the SR box was turning yellow or white to indicate loss of focus, allowing me to avoid pressing the shutter. This did not happen in any of my testing before I updated to firmware 2.01 and also did a full rest on the camera.Previously it just stayed green the entire time.
- The f1.8 lens, at least in this limited test, absolutely performed the best. I did not record a single miss using that lens in spite of the fact that it was the last one I did and the lighting was the lowest.
- The f4 aperture performed the worst, which is unsurprising.
- The 2.8 aperture had a few misses and it had the problem I am trying to demonstrate in that it focused on the back eye when the SR box was on the front one. It seemed like it stayed green when actually out of focus more often than with f4.
All of this testing was without flash. I did one test with flash at 2.8 and nothing extraordinary happened other than one more miss where the camera said it had the front eye and actually focused on the rear eye.
- This was also tested with a relatively small custom area box. Previously, I had used the full area because it had proven more reliable in the past. I only did one round with the full area AF tonight, at 2.8, and it seemed to perform worse than the smaller box. This makes sense - it is just contrary to behavior I had previously observed.
- Also not super surprising, but it kept focus on my lighter eyed children better than my darker eyed ones. Worth noting here is that the worst performance was not on my darkest, brown eyed child but on one with lighter blue eyes just slightly darker than my lightest eyed child, on whom it performed the best.
This leaves me with some very preliminary initial conclusions:
1. Firmware 2.01 or the system reset or both may have improved performance. It's still not perfect, but it initially appears noticeably better.
2. At 1.8 it seemed to perform almost flawlessly. It's noteworthy that many people reporting no problems also report shooting at 1.2 or 1.8.
3. IF - and that is a big if - the camera will consistently display the yellow or white box instead of green when something is not in focus, I'd consider the issue resolved. It seemed to mostly do this at f/4. At f/2.8 it was, perhaps we could say, overconfident at times.
4. I want to experiment further with the smaller wide area box vs. the full area. I started with this camera never using the full area box and only started using it grudgingly after finding it was giving me more in focus results than the smaller areas. I did this past November shoot literally thousands of photos to test this and the result of full-area being better was extremely consistent, but it may be that for whatever reason this was not normal and it is now working the way most people find it to - with smaller boxes being better.
5. I am cautiously optimistic here, but at the same time there is very clearly an issue as I would say that in at least 10%+ of my test shots the SR was targeting the front eye but focus was on the rear eye. I have a strong example of this shared below.
Here is the test at f4.
At around 0:10 I zoom in to show a focus going to the wrong eye (the REAR one instead of the front one). In fairness to the camera, the SR box had turned white just as I pressed the shutter here so the camera was trying to tell me it was not focused. However, I think that there was also a shot I show at this time when it was still green but the focus was on the wrong eye, and either way it's still odd that even though it was not on the correct eye it WAS focused on the other eye.
Around 0:30 the same thing; focus on the rear instead of the displayed front eye, but SR box had turned white.
Here's the f1.8 video. Nothing to comment on here, really. It briefly turned white a few times but basically it did what it was supposed to. I'm not sure if it comes through in the video or not, but the SR box also "felt" a lot smoother and stickier.
Here is one f2.8 video. https://studio.youtube.com/video/gXosGNW7fLk/edit
Most notable I think is the shot taken about 2/3 of the way through that clearly shows the phenomenon of the SR being on the front eye but the actual focus on the rear. Here is that photo:
https://live.staticflickr.com/65535/53695380485_b4564387cc_b.jpg
Here is an extended video at 2.8. I'm not sure if there was anything particular about this I was trying to point out. I actually did a whole bunch of recordings that I didn't upload but most of them got interrupted by a kid doing something so I restarted. This one is just one that ran for a bit without any distractions for the most part. The most notable thing here is that it mostly behaves BUT it does lose focus a lot - it just happens to TELL me about it so I can avoid taking missed focus shots.
Here's one more at 2.8 where we see the "reports focus on the front, actually focused on the rear" thing going on. In this case there is a very brief turn to white or yellow, but otherwise the box is green the entire time. I've watched this back in slow motion but I can't tell if the missed focus shot I share below was at the moment the box showed a loss of focus or not:
https://live.staticflickr.com/65535/53695286959_3807843c52_b.jpg...Show more →
Great job on the videos. I posted this on your DPR thread but posting here as well:
On several of the videos I see the MF override distance scale appear on the screen. In the f/4 video it appears just before the photo is taken. This scale appears whenever the focus ring is being manipulated while AF-ON is being pressed. Are you inadvertently hitting your lens focus ring when handling the lens? Here's your f/4 video, slowed down by 10x to show this:
I edited to show the two photos taken. As you indicated, the focus indicator turns white just before the exposure. In the first the MF ring is bumped - also the white left rangefinder arrow is active, showing the lens is now OOF relative to the AF point due to the MF manipulation. In the second photo the subject is moving quickly just before the exposure and the indicator is white the entire time.
I hope it's ok to repost your video like this - it's unlisted so it'll only show up here. If this is an issue let me know and I can take it down.
|
. 