Sony A9 has random banding issues at ISO 100?!

[peterzuehlke]

just anecdotally, I've been shooting performance for over 10 years, the last 4 with a Canon 6D, usually 300 to 500 shots per performance and I have never seen this banding or other stripes in any of my images. When I saw the pattern with the Sony a6500, in silent shutter mode, I reduced the shutter speed because of my experience with shooting video under LED light and the aliasing beat (usually around 1/60, but different shutter speeds depending on dimmer setting) you sometimes get with these lights.

 

[SecureGSM]

you get it right. here is a pretty reasonable write up on the issue by DPR:

https://www.dpreview.com/articles/7370859353/sony-a9-banding-reported-by-fro-fact-or-fiction

it all makes sense to me.

just anecdotally, I've been shooting performance for over 10 years, the last 4 with a Canon 6D, usually 300 to 500 shots per performance and I have never seen this banding or other stripes in any of my images. When I saw the pattern with the Sony a6500, in silent shutter mode, I reduced the shutter speed because of my experience with shooting video under LED light and the aliasing beat (usually around 1/60, but different shutter speeds depending on dimmer setting) you sometimes get with these lights.

 

[neuroanatomist]

you get it right. here is a pretty reasonable right up on the issue by DPR:

https://www.dpreview.com/articles/7370859353/sony-a9-banding-reported-by-fro-fact-or-fiction

it all makes sense to me.

I wonder what DPR would have concluded had such banding been reported on a Canon 1-series body? :

 

[peterzuehlke]

Maybe the DP Review people will find a new test protocol, maybe a repeating strobe flash that is synced to the readout bands in Sony sensors when in silent shutter mode, that shows this to be a positive feature, probably resulting in lower shadow noise and greater dynamic range.

 

[Jack Douglas]

That's my laugh for the day! Tony already has demonstrated the symptom. He wants so badly to believe it's perfect (can't really blame him considering the up side of it is so great). As Neuro and others have said, innovation and early implementation has it's up side and down side; high stakes gambling that Canon avoids.

Jack

 

[Jack Douglas]

"Granted, as a camera designed (partly) to satisfy the needs of sports photographers, the a9 is probably going to be found shooting in situations with LED signboards around where, after-hours, they might account for a significant portion of light on your sideline subjects. If that describes the situations you'll be routinely shooting under, and you're concerned about the 2% banding rate in sideline action, this may be something to add to your 'cons' list when considering this camera."

How polite of DPR. "May be something ..." I wonder if "partly" relates to the 20 fps they brag about?

Jack

 

[ahsanford]

"Granted, as a camera designed (partly) to satisfy the needs of sports photographers, the a9 is probably going to be found shooting in situations with LED signboards around where, after-hours, they might account for a significant portion of light on your sideline subjects. If that describes the situations you'll be routinely shooting under, and you're concerned about the 2% banding rate in sideline action, this may be something to add to your 'cons' list when considering this camera."

How polite of DPR. "May be something ..." I wonder if "partly" relates to the 20 fps they brag about?

Jack

A number of photography websites -- DPR included -- seem to be defending this as a nitpicky little oddity and not a show stopper of a problem. That's certainly possible, but they are not really offering a defense. They are simply dismissing JP as a non-sports photog making a big deal about a very rare occurrence.

All Sony needs to mitigate this is any one of the following:

• A statement that this indeed is a confirmed issue, but it is firmware correctable and that firmware update is in progress.
  
  
• They have their sports photogs state that mechanically shuttered cameras suffer from this as well and that it's not a unique problem to the A9. The narrative then becomes 'Sports guys need to judiciously manage their shutter speeds under certain lighting' instead of 'the A9 has problems'.
  
  
• A statement that this only happens to a very small number of cameras and they will be serviced/replacedon request.
  
  
• Confirm the electronic shutter + some artificial lighting uniquely generates this problem, and give guidelines for max shutter speed under such lighting.

...but in absence of that, we don't know what we don't know. If the electronic shutter will globally exhibit this performance in artificially lit sports/action work, folks will wonder why they ponied up $4500 for a rig that occasionally needs to be 'downshifted' to a 5 fps max mechanical shutter.

- A

 

[peterzuehlke]

generally, the venues i shoot in are so dim (and light by dmx addressed LED stage lights) i am lucky to get up to 1/100 sec shutter speeds, at f/2.0 and 6400 to 10,000 iso (thank you lighting designers) and yet they manage to get the sound up to 100 dB plus, so I probably won't see it again. I got the Sony for the silent shutter, small size and image stabilized primes, plus still shot IBIS (hello Canon). Waiting to see some real shots out of the 6dII. was tempted by the 5DIV, and love the color and tonality out of a friend's 5DS. hopefully be able to make a decision soon, before i go totally nuts. (i just bought a folding 6x9 cm film camera.)

 

[Jack Douglas]

Agreed on how to handle this situation. It's kind of sad when anyone puts their heart and soul (myself included) into a project and unanticipated problems surface, especially when it's high stakes gambling. Not that I feel sorry for Sony but still ...

Jack

 

[peterzuehlke]

I am not sure why Sony is marketing this camera as a sports camera. Canon has owned the sidelines with those big white lenses for a long time. (and many of us think lenses are the important part of the kit) I got the sony in large part due to IS (whatever Sony calls it) primes specifically for small dark clubs. was thinking of just getting the new 85mm IS, but i am afraid it will be the size and weight of a 70-200 2.8 and pointing big lenses at musicians is a problem. Actors light up, musicians not so much.

 

[Don Haines]

I rather doubt that this is a problem that is restricted to the Sony A9, and I think that it is a problem that will get bigger.

One has to understand how LED lighting panels work..... At full power, the LEDs are constantly on. To dim them, they pulse them on and off. For example, at half brightness, the LED will be turned on half of the time, and turned off half of the time. When you do that with an incandescent bulb, it takes tens of milliseconds to turn on and up to 100 milliseconds more to turn off, resulting as if you pulse it at 60hz the dimming is not noticeable to the human eye or to a camera.

LEDs turn on and off in Nanoseconds, so if you pulsed them at 60 hz, it would be a very noticeable flicker. The solution is to pulse them at a faster rate, say 1Khz.... At 1Khz, the human eye will not detect the flickering and the light levels will appear to be stable, but if you have a camera with a reasonably fast shutter, it will.

The same thing happens with florescent lighting.... older lighting flickers, some of the newer lighting is at a higher frequency and the flicker is too fast for the eye to detect...... BUT A CAMERA STILL CAN!

This is why Canon has been putting the anti-flicker software into every camera released since the 7D2. They realize that this is a problem and have taken steps to prevent it. I would suspect that other manufacturers will start giving the option.....

 

[peterzuehlke]

i have heard that the dmx dimmed led stage lights actually don't just change dwell (duty cycle) they change frequency at different dimmer settings. with video sometimes you have to pick the best frame rate, no ideal one. i am always happy to see an old tungsten halogen ellipsoidal light.

 

[foo]

We need someone with the physics background in light sources to detail how what's on the market works and then with the detailed operation of the shutter it should be possible to determine how the interaction takes place. I'm guessing that it's not going to be a simple fix.

That's likely to be a practical impossibility. There will be many different lights that operate differently, even the same type of light with a different controller may have a different effect on the result. Multiplied by how many lighting manufacturers who all have a different rate of progress in their technologies..

Simplistically this looks like an interference pattern created by the scanning frequency of the sensor vs the frequency the light source is operating at.

The line from the DPR article 'typical artifical lighting, which tends to flicker at 60 or 120 Hz' is far too simplistic as it's assuming US style 60Hz mains, so with most of the rest of the world using 50Hz there may be different effects.

Increasingly flourescent lighting is using more efficient high frequency electronic ballasts, so may be operating in the 100's of kilohertz, but likely not all of these operate at the same frequency.

Then you have PWM controlled LED's etc. What happens if some lighting somewhere uses PFM, or even just PWM with pulse skipping?

If we assume that it is an interference pattern between the two frequencies, you're going to get odd things like 50kHz lighting is really bad, while 50.1kHz may show nothing.

Whatever the outcome, it could be that the issue is just an inherent limitation of the architecture of the sensor and as such can't be fully mitigated. Suspect Sony won't be up for giving out those sorts of details just so that they can be roasted on various forums by a bunch of keyboard warriors though..

 

[ahsanford]

Here is an attempt at defending this as a non-issue: https://sonymirrorlesspro.com/sony-a9-banding-problem/

Which is at least an attempt to explain things.

"As a refresher, the Sony a9 operates primarily in electronic shutter mode, an operation where instead of a physical, mechanical shutter ending each frame, the camera controls the shutter duration by powering on and off the imaging sensor.

...

With a mechanical shutter, the whole sensor is on when the shutter goes up, and it’s still on when the shutter goes down. This makes a mechanical shutter camera less prone to the effects of flicker because the whole sensor is exposed at once. An electronic shutter (at least a progressively read electronic shutter) reads lines of information from the sensor from top to bottom or from side to side. Because it’s reading while the light is flickering, it’s possible for the sensor to record that flicker, and the result is banding."

But my take home from this (pls read the entire article) is:

• He doesn't want to call it banding, but later refers to the phenomenon as banding anyway. ??? What you call it doesn't mitigate that is happening, and the output is unacceptable.
  
  
• He seems to believe this is absolutely electronic shutter related, and seems to chalk it up to 'the price of admission' for such a rig. You get 20 fps, but you also may get banding in some circumstances -- use your noggin to avoid this. But how to avoid this remains undefined: is there a magical max shutter speed threshold to stay under or a multiple of 30/60 Hz that will prevent this? Must you downshift to 5 fps mechanical shutter to truly rid yourself of this?
  

If this is a consistent electronic shutter problem for all the A9s out there in certain lighting, and the only fix is 'please slow down your shutter or consider 5 fps mode' is the only recourse, this is a big deal. This should scare the hell out of one group this rig is squarely aimed at: sports folks.

But if this is just a nutty nutty rare phenomenon, perhaps this isn't a big deal.

I'm just not seeing any context to the scale/likelihood of this happening from those that are dismissing this, and as such, they aren't dismissing it very effectively.

- A

 

[SecureGSM]

Don,

Yes, problem seems to be not brand restricted. Canon anti-flicker won't help though as we deal here with multiple light sources flickering at multiple frequencies, sadly.
here is a similar case:

https://www.dpreview.com/forums/thread/4133297

p.s. I suspect that application of CPL may help to isolate frequencies in theory?

I rather doubt that this is a problem that is restricted to the Sony A9, and I think that it is a problem that will get bigger.

One has to understand how LED lighting panels work..... At full power, the LEDs are constantly on. To dim them, they pulse them on and off. For example, at half brightness, the LED will be turned on half of the time, and turned off half of the time. When you do that with an incandescent bulb, it takes tens of milliseconds to turn on and up to 100 milliseconds more to turn off, resulting as if you pulse it at 60hz the dimming is not noticeable to the human eye or to a camera.

LEDs turn on and off in Nanoseconds, so if you pulsed them at 60 hz, it would be a very noticeable flicker. The solution is to pulse them at a faster rate, say 1Khz.... At 1Khz, the human eye will not detect the flickering and the light levels will appear to be stable, but if you have a camera with a reasonably fast shutter, it will.

The same thing happens with florescent lighting.... older lighting flickers, some of the newer lighting is at a higher frequency and the flicker is too fast for the eye to detect...... BUT A CAMERA STILL CAN!

This is why Canon has been putting the anti-flicker software into every camera released since the 7D2. They realize that this is a problem and have taken steps to prevent it. I would suspect that other manufacturers will start giving the option.....

 

[foo]

i have heard that the dmx dimmed led stage lights actually don't just change dwell (duty cycle) they change frequency at different dimmer settings.

It's quite common for stuff that's nominally fixed frequency PWM to do pulse skipping at lower levels to increase efficiency. This will essentially give interesting effects as it may not simply skip every other cycle, but could do every third.. The decsison to skip or not is also probably re-evaluated every cycle, so around the transition points could be particularly problematic for the camera to predict.

Have to be careful with lighting though as at some point the flicker will become visible to the human eye. Peripheral vision is much more sensitive to flicker, so it's not just about what's in the center of the stage as you may see flicker from the same light as it reflects off things in the periphery.

 

[foo]

I rather doubt that this is a problem that is restricted to the Sony A9, and I think that it is a problem that will get bigger.

Definately, more likely it's a general thing for electronic shutters. Given time I expect that software strategies or different sensor architectures will get better results though.

 

[SecureGSM]

what a bunch of clowns. what about shutter speeds in excess of X-Sync??? bang... shooting at speed exceeding the camera x-sync speed becoming potentially problematic even with mechanical shutter. hence shooting at 1/250s with 1DX shoud not pose an issue and that seems can be easily verified..

Here is an attempt at defending this as a non-issue: https://sonymirrorlesspro.com/sony-a9-banding-problem/

Which is at least an attempt to explain things.

"As a refresher, the Sony a9 operates primarily in electronic shutter mode, an operation where instead of a physical, mechanical shutter ending each frame, the camera controls the shutter duration by powering on and off the imaging sensor.
...
With a mechanical shutter, the whole sensor is on when the shutter goes up, and it’s still on when the shutter goes down. This makes a mechanical shutter camera less prone to the effects of flicker because the whole sensor is exposed at once.

 

[neuroanatomist]

what a bunch of clowns. what about shutter speeds in excess of X-Sync??? bang... shooting at speed exceeding the camera x-sync speed becoming potentially problematic even with mechanical shutter. hence shooting at 1/250s with 1DX shoud not pose an issue and that seems can be easily verified..

Here is an attempt at defending this as a non-issue: https://sonymirrorlesspro.com/sony-a9-banding-problem/

Which is at least an attempt to explain things.

"As a refresher, the Sony a9 operates primarily in electronic shutter mode, an operation where instead of a physical, mechanical shutter ending each frame, the camera controls the shutter duration by powering on and off the imaging sensor.
...
With a mechanical shutter, the whole sensor is on when the shutter goes up, and it’s still on when the shutter goes down. This makes a mechanical shutter camera less prone to the effects of flicker because the whole sensor is exposed at once.

Oh, c'mon. It's not like sports shooters shoot using shutter speeds faster than 1/250 s, right?

 

[foo]

Here is an attempt at defending this as a non-issue:

for someone critisizing Fro for a lack of understanding, he's demonstrating his own misunderstandings too.

But if this is just a nutty nutty rare phenomenon, perhaps this isn't a big deal.

If it was, it wouldn't show up so easily.

Like most of these things it'll affect some more than others, there'll be people who blow it out of proportion and the die hard defenders. Whatever, for some people it's going to be a real issue and the reasons behind what's happening are almost irrelevant.