Z8, 24-70/2/8s @ 62mm , F4, AF-c on tripod 1/4s, iso 25600. I took 7-8 shots all in focus.
Face detection and eye focus came on left eye, no hunting but took a couple of second to confirm focus.
Z8, 24-70/2/8s @ 62mm , F4, AF-c on tripod 1/4s, iso 25600. I took 7-8 shots all in focus.
Face detection and eye focus came on left eye, no hunting but took a couple of second to confirm focus.
If it took a couple of seconds to confirm focus on statue with high contrast light around the eye, what performance would you expect on a human child with flat lighting who is constantly shifting position?
p.17 #3 · Z8 extremely unreliable/inconsistent AF in mildly low light
snapsy wrote:
If it took a couple of seconds to confirm focus on statue with high contrast light around the eye, what performance would you expect on a human child with flat lighting who is constantly shifting position?
what high contrast. I took this at 7[pm all light off and the only light source was my projector that I was watching movie on about 6 meter behind and 5 meters to the side.
We are talking about how many % in focus right? Mine is 100%
p.17 #4 · Z8 extremely unreliable/inconsistent AF in mildly low light
suteetat wrote:
because I already showed pictures if you followed my link on thread of the week winner under my name shooting Z8 at f8 at night time underwater with F mount macro lens/FTZ adapter on erratic moving subject while scuba diving with just a few flash light as the only light source in total darkness and AF did not struggle nearly as much as this and keeper rate is much higher. May be my Z8 is the abnormal one with super AF ability at night?
If your Z8 is the "abnormal one with super AF ability at night", then I recommend you submit your super camera to to be the duplicate test candidate. There'd be no better camera to test under SCoombs' EV4 conditions than one that is claimed to have "super AF ability at night."
Z8, 24-70/2/8s @ 62mm , F4, AF-c on tripod 1/4s, iso 25600. I took 7-8 shots all in focus.
Face detection and eye focus came on left eye, no hunting but took a couple of second to confirm focus.
2 seconds to confirm focus on a stationary object?!? While that might meet your criteria for adequate performance, that kind of performance would never be acceptable on a moving subject like a child.
p.17 #7 · Z8 extremely unreliable/inconsistent AF in mildly low light
suteetat wrote:
what high contrast. I took this at 7[pm all light off and the only light source was my projector that I was watching movie on about 6 meter behind and 5 meters to the side.
We are talking about how many % in focus right? Mine is 100%
Look at the light around the left eye. The contrast is so high it's practically a specular highlight.
p.17 #8 · Z8 extremely unreliable/inconsistent AF in mildly low light
snapsy wrote:
If it took a couple of seconds to confirm focus on statue with high contrast light around the eye, what performance would you expect on a human child with flat lighting who is constantly shifting position?
This is fair, but also do we expect our cameras to provide lightning fast subject detect AF on a moving subject in -2 EV conditions? Because that’s a pretty tall ask.
p.17 #9 · Z8 extremely unreliable/inconsistent AF in mildly low light
Jman13 wrote:
This is fair, but also do we expect our cameras to provide lightning fast subject detect AF on a moving subject in -2 EV conditions? Because that’s a pretty tall ask.
I think -2EV is too much to ask. The 4EV lighting the OP is presenting seems more reasonable. The camera is rated down to -7EV, albeit with a f/1.2 lens and AF-S.
p.17 #10 · Z8 extremely unreliable/inconsistent AF in mildly low light
We need to be cautious not to take the bait when irrelevant material is submitted in an attempt to run the discussion off into the weeds. 4EV, -2 EV and -7EV are totally different situations. @SCoombs gave us great data to analyze in the EV4 test case. Flooding the thread with irrelevant data like pictures of stationary artwork serves no purpose other than to derail any meaningful progress.
If people claim to have super cameras that never fail to properly focus, then I think they should make them available for comparative testing. Otherwise their claims are just noise being injected into the discussion.
p.17 #11 · Z8 extremely unreliable/inconsistent AF in mildly low light
Yawn, it was good enough to track all kinds of critters in almost completely darkness underwater with f8 and small focus light. They were not stationary and moved a heck of a lot more than a child sitting for a picture.
p.17 #15 · Z8 extremely unreliable/inconsistent AF in mildly low light
snapsy wrote:
I think -2EV is too much to ask. The 4EV lighting the OP is presenting seems more reasonable. The camera is rated down to -7EV, albeit with a f/1.2 lens and AF-S.
Agreed. FWIW, I have found the Z8 to be less accurate in shallow DOF situations with some lenses when shooting people than I found my Canon R5. However, most of those misses tended to happen with the 105/1.4E. Native Z lenses, and even my adapted Sigma 85/1.4 and Sony 50/1.2 GM have had very good accuracy. I shot an event indoors in mixed lighting using almost exclusively the 50/1.2GM and 85/1.8S and had perfect focus on over 90% of shots.
Even shooting in dimmer light portraits with the 105/1.4 and the Z 70-180:2.8, I had good overall accuracy with only a handful of misses. Though I did have the 105/1.4 miss randomly in bizarre situations that shouldn’t happen.
p.17 #16 · Z8 extremely unreliable/inconsistent AF in mildly low light
I see an invalid apples and oranges comparison when it comes to the statue photo and the bait photos vs. the child's eye photos/videos.
The bait photo task asked the camera to focus on a single, flash-illuminated subject against a totally black background. The subject was perpendicular to the lens axis and occupied a narrow depth of field. That is the best case scenario for a camera, as it is a simplest test case where there is no alternative subject for the camera to choose as a focus point. It would be hard for AF to fail under those test conditions.
The task of subject identification in a uniformly-lit home interior using eye-detect AF is a totally different task. There is a lot of competing 'signal' in the home interior photo, which the camera's AF algorithm has to reject when searching for an eye. Scanning a scene takes comparatively more computation time than scanning a black background.
In one case the AF system is being asked to perform a simple task. In the other case the AF system is performing a far more complex task.
Although two of the bait pictures have eyes, one does not. Eye-detect AF can not possibly have been used in that 'no-eye' case. The bait poster didn't tell us that eye-detect AF was in use on the bait shots. If eye detect was not in use then we have a deliberate oranges to pears comparison, which only wastes out time. We know that eye-detect AF couldn't have identified an eye in the second bait photo, and we don't have any reason to consider that eye-detect AF was used in the first or third bait photos during the same underwater shoot.
@SCoombs' tests were a very good demonstration of eye-detect AF in a uniformly lit scene, where the AF algorithm has to sort through a lot of data to perform eye detection. I don't think the bait photographs share anything in common with SCoombs' test condition. They're nothing more than a red herring.
Z8, 24-70/2/8s @ 62mm , F4, AF-c on tripod 1/4s, iso 25600. I took 7-8 shots all in focus.
Face detection and eye focus came on left eye, no hunting but took a couple of second to confirm focus.
Let me confirm with you: you say this photo was taken at f4, 1/4 of a second, and ISO 25600?
That's *insanely* dark. I can't fathom how the photo has come our with the combination of brightness and noise levels shown here. This is 6 stops darker than the photos I have shown in my previous videos when, without the flash, any photo was almost unintelligible for the level of noise.
p.17 #18 · Z8 extremely unreliable/inconsistent AF in mildly low light
coralnut wrote:
2 seconds to confirm focus on a stationary object?!? While that might meet your criteria for adequate performance, that kind of performance would never be acceptable on a moving subject like a child.
Yes, I've messed around with the subject detect on statues, paintings, etc. and it's always been much more capable than on real living people. The lighting he describes doesn't really confuses me for what the photo looks like here, but even with that aside SR working more consistently on a statue wouldn't surprise me.
p.17 #19 · Z8 extremely unreliable/inconsistent AF in mildly low light
Jman13 wrote:
This is fair, but also do we expect our cameras to provide lightning fast subject detect AF on a moving subject in -2 EV conditions? Because that’s a pretty tall ask.
I personally don't at all, but I'm trying to get it at EV4, not Ev-2. Also, as I've said above, for Ev-2 that photo blows away my expectations for just overall image quality at that light level and ISO.
p.17 #20 · Z8 extremely unreliable/inconsistent AF in mildly low light
Not only is his super Z8 especially adept at focusing in the dark, it's especially adept at doing so while not rendering noise in the image at ISO 25000. Amazing.
?