I can repeatedly reproduce overexposure now

[ELAN]

Advertisement (gone after registration)

8 hours ago, Photoworks said:

for me it was only happing when shooting quickly multiple frames in single mode, with auto ISO and Auto shutter.

Same here.  My theory is that the camera is metering while the shutter is still in motion re-opening the shutter after the previous shot was taken. That’s why I sometimes see overexposure by 2-3 stops and other times by 5 stops. 

[MrPaulK]

15 hours ago, M11 for me said:

And as I understand it that workflow is not really relevant either. Probably you do not take a picture and then look at it on the LCD and while doing that you hold the camera towards the next sceene and at that same time you halfpress the release button and you take the next shot. 

Actually I’ve done this quite a bit usually waiting for some kind of timing thing and reviewing if I hit it on the last try. I’m not sure why you assert that isn’t valid. 

[SrMi]

10 hours ago, ELAN said:

Same here.  My theory is that the camera is metering while the shutter is still in motion re-opening the shutter after the previous shot was taken. That’s why I sometimes see overexposure by 2-3 stops and other times by 5 stops. 

Those who see overexposure could try shooting with the electronic shutter (if possible) to see whether the overexposure goes away.

[M11 for me]

vor 3 Stunden schrieb MrPaulK:

Actually I’ve done this quite a bit usually waiting for some kind of timing thing and reviewing if I hit it on the last try. I’m not sure why you assert that isn’t valid. 

No, I did not say that: What was described was the following: You look at a shot on the LCD and while you do that you hold your camera horizontally up in the air aiming at your next shot while still looking at the old shot on the LCD. While doing that you half press the release button . . . .

Now tell me if you ever do that.

But still; why should there be an overexposure in that instance. I have no explanation.

[dscdsc]

I was able to reproduce this on the first try using the same settings.

FW 1.6 but I've had it happen in older ones too

[Kwesi]

Teacher to students: All of you who have learned the secret sequence in your camera that gives you randomly over exposed images should move to the head of the class!

Now take that knowledge and go express yourself with it! 😇

[Tailwagger]

Advertisement (gone after registration)

22 hours ago, SrMi said:

“The patient says, "Doctor, it hurts when I do this."
The doctor says, "Then don't do that!”

Seriously, the new metering system may have some race condition as it now involves computations instead of reading an analog sensor.

It's not clear that the metering system is 'new' given we've had the same capability to metering directly from the sensor when in LV for over a decade now. The one known difference between the M11 and its predecessors is that now metering via the sensor is the default (and sole methodology) whereas previously the vast majority of users were likely relying on the analog sensor.  More usage, more likely the conditions are hit, seen and complained about.

I also was unable to reproduce the same effect with my M11 when cycling though a set of 5 photos, then retrying again by cycling through a set of around a dozen.  Might require more files to display or perhaps a set of intervening operations like zooming in and out prior to chew up more memory.  What might be more interesting is that I tried the same experiment on the SL2 as I have to believe that a lot of code is shared between these cameras.  I was able, twice out of a dozen or so tries, to cause a mini lockup. The rear screen blanked when the shutter was pressed, but the AF failed to actuate the lens motors and fully depressing the shutter did nothing. Once the shutter was released however things returned to normal.

22 hours ago, jqian6 said:

It occurs if I frantically browse through the pictures and all of sudden I depress the release fully to take a picture, that picture is overexposed.

Out of curiosity, are you shooting in 60MPx or a reduced mode (where we can assume there's even more work being done to the image prior to storage)?

As I said elsewhere I doubt the SD cards are the issue per se other than potentially altering timing.  From this additional failure scenario, however, it wouldn't surprise me in the least if many of the problem(s) being seen are related to internal, not external, memory.  This bug, along with some of the others, suggests a memory management issue.  Could possibly be a subtle mem leak that over time can result in there not being enough available in times of stress..' like having a bunch of images rendered and cached internally and cycling through them rapidly and concurrently pressing the shutter.  If the playback mechanism and various shooting, processing and display processing functions all alloc memory from the same mem pool, there could also be a problem related to general contention for resources. If someone holding a lot of the available memory on one hand runs into locks related to prioritizing certain operations over others, like writing the card, doing metering... not good. ie. something along the lines of the lower priority op holds the memory the higher priority one requires with the higher priority one holding some other lock the lower can't get. Certainly the sort of scenario that could result in the extreme in the locks ups or possibly operations, say reading the metering values, failing.

Edited January 27, 2023 by Tailwagger

[Kiwimac]

23 hours ago, hdmesa said:

I think it's typical of not just M but SL cameras as well – overexposure if shutter is depressed before the camera has time to meter the scene. Use of your SD card versus the internal memory may cause a slight time delay, just long enough to miss exposure. On my SL2-S, I can cause overexposure by zooming in to focus, the exposure brightens to the zoomed-in crop, then I immediately press the shutter button, which causes the exposure to be taken exactly as the camera zooms back out and before it can re-meter the full scene again. I have to instead half press the shutter to zoom out, then give the camera about half a second for the exposure to re-meter and "settle" on the right exposure.

Doesn’t the M11 meter in real time off the sensor? There shouldn’t be any delay at all. 

[jqian6]

26 minutes ago, Tailwagger said:

 

Out of curiosity, are you shooting in 60MPx or a reduced mode (where we can assume there's even more work being done to the image prior to storage)?

As I said elsewhere I doubt the SD cards are the issue per se other than potentially altering timing.  From this additional failure scenario, however, it wouldn't surprise me in the least if many of the problem(s) being seen are related to internal, not external, memory.  This bug, along with some of the others, suggests a memory management issue.  Could possibly be a subtle mem leak that over time can result in there not being enough available in times of stress..' like having a bunch of images rendered and cached internally and cycling through them rapidly and concurrently pressing the shutter.  If the playback mechanism and various shooting, processing and display processing functions all alloc memory from the same mem pool, there could also be a problem related to general contention for resources. If someone holding a lot of the available memory on one hand runs into locks related to prioritizing certain operations over others, like writing the card, doing metering... not good. ie. something along the lines of the lower priority op holds the memory the higher priority one requires with the higher priority one holding some other lock the lower can't get. Certainly the sort of scenario that could result in the extreme in the locks ups or possibly operations, say reading the metering values, failing.

I am shooting 60MP.

I think your memory leak theory can also be plausible.

I did an experiment again. I currently have about 300 pictures all in SD card, I bring up the review and then hold one of the arrow key constantly to cycle them through, while at the same time fully depress the release button about 20 times. First 5-8 pictures are kind of within normal exposures and after that the overexposure happens again.

I also did it with both mechanical and electronic shutter only, same kind of phenomenon. So that rules out the shutter problem, but leaves it to metering only.

The camera may have enough internal cache to cycle through first few hundreds pictures but later on when the cache is full the system has to push some old ones out of the queue and put the new ones in. This kind of operation may cause memory leak and slow down system, making the camera stop metering.

Of course it's just my theory, and nobody knows what really happens.

I am done doing Leica's engineers' job and going to use the camera like a normal person from now on.

Edited January 27, 2023 by jqian6

[jqian6]

5 minutes ago, Kiwimac said:

Doesn’t the M11 meter in real time off the sensor? There shouldn’t be any delay at all. 

there could be a delay in real time

[hdmesa]

14 minutes ago, Kiwimac said:

Doesn’t the M11 meter in real time off the sensor? There shouldn’t be any delay at all. 

It does, but due to the processing speed of the camera there is always a small delay because it needs time to not only calculate the exposure but also to change the settings then adjust the preview in the EVF.

Keep in mind that when zooming in on an image to focus, you're giving it another image to meter resulting in a delay. And when you zoom out, it has to re-meter the new zoomed out scene. It can't keep metering the entire scene while also metering the zoomed in image. Also, when reviewing an image on the LCD, the meter is either turned off or not being analyzed for exposure.

Edited January 27, 2023 by hdmesa

[MrPaulK]

2 hours ago, M11 for me said:

Now tell me if you ever do that.

Yes precisely that, as I described. As I said I’m usually trying to time something and reviewing the attempt when the opportunity arises to try again. Normally on a tripod, or arms resting on knees, or shooting without looking (loose framing). I don’t think it’s that unusual. Keep an eye on the subject while reviewing. 

[BradS]

23 hours ago, jqian6 said:

I am serious, guys, try it, it's kind of fun to torture it.

scenario is: turn on the playback showing a picture, then hold on to one arrow of the pad, the pictures fly through the LCD, at the same time, depress the shutter release button fully taking a picture, and it's over exposed by maximizing the ISO and the shutter speed is about 1/20s.

Now I kind of love to torture it.

 

I retired three years ago but when I was still doing software/firmware, I would have loved to have a tester with this attitude. A reproducable bug report is a gift of pure GOLD! Keep it up!

Edited January 27, 2023 by BradS

[SrMi]

3 hours ago, Tailwagger said:

It's not clear that the metering system is 'new' given we've had the same capability to metering directly from the sensor when in LV for over a decade now.

AFAIK, the difference is that now metering in RF mode is done in a power-saving way. Likely the sensor is read at specific intervals and, maybe, when the trigger is pressed. In LV, the sensor is read all the time. Do we see overexposure with active LV as well?

[Tailwagger]

2 hours ago, SrMi said:

 Do we see overexposure with active LV as well?

I believe so, but its hard to say definitively.  I've experienced it once or twice and I'm 100% LV.  I'll confess at the time I assumed it was pilot error.  But you raise an interesting point given another timer triggered event in the mix. 

[hdmesa]

1 hour ago, Tailwagger said:

I believe so, but its hard to say definitively.  I've experienced it once or twice and I'm 100% LV.  I'll confess at the time I assumed it was pilot error.  But you raise an interesting point given another timer triggered event in the mix. 

The SL2-S is of course live view all the time, and I can provoke incorrect exposure if trying to exit another task too quickly and forcing an exposure before the meter has time to "settle" and read the scene. I can literally see the image change as the camera finds the right exposure after exiting being zoomed in to manually focus. It feels like about a full second I have to wait for the exposure to be correctly metered.

Edited January 28, 2023 by hdmesa

[Kiwimac]

8 hours ago, jqian6 said:

there could be a delay in real time

I’ve owned and used Panasonic, Olympus and Fujifilm cameras that were all EVF cameras that I imagine metered the same way in principle. Never had an issue like the one described. 
 

I suppose it depends on the read speed of the sensor. 

[Tailwagger]

7 hours ago, hdmesa said:

The SL2-S is of course live view all the time, and I can provoke incorrect exposure if trying to exit another task too quickly and forcing an exposure before the meter has time to "settle" and read the scene.

Yep... that aligns with my test of the SL2 I mentioned above. More fuel on the fire that there's something, either shared design philosophy or code... given the same processor family... in the firmware's DNA that's not 100% dependable w.r.t. to multi-tasking. 

While tradition, heritage and a commitment to past practices can be a strength, it can be an impediment as well.  Not likely to happen in this millennium, but I cant help but feel that more good than harm would come from Leica open sourcing at least a portion of their code. Redact the DFD, PC, etc details; omit sufficient info to prevent third parties from rolling their own firmware if you must but I'd wager there are some folks who might be interested, clever and experienced enough to id the source of some of these issues. 

[Tailwagger]

6 hours ago, Kiwimac said:

I suppose it depends on the read speed of the sensor.

I think it more depends on the design and quality level of the software that's dealing with it. 

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[SrMi]

9 hours ago, Kiwimac said:

I’ve owned and used Panasonic, Olympus and Fujifilm cameras that were all EVF cameras that I imagine metered the same way in principle. Never had an issue like the one described. 
 

I suppose it depends on the read speed of the sensor. 

The sensor read speed is the same as with other cameras (rolling shutter). The read interval/frequency is likely different whether you are in RF or LV mode.

AFAIK, M11 is the first camera to meter from the sensor without LV.

Edited January 28, 2023 by SrMi