Banding issues at low ISOs with E-M5

[ecw]

I have heard of high iso banding issues, and banding issues with the 20mm lens, but my problem seems to be different.

Im getting banding between ISO 200-500. At ISO 800+, there is no banding. The light source doesn't seem to matter, as the same problems occur in daylight.

Has anyone else experienced this? I'm thinking its a defect with my camera.

I did a simple test, taking the same photo with different ISOs:
All taken with the 25mm f/1.8

(full album here: https://www.flickr.com/photos/127648879@N03/sets/72157648146029478/ )

View attachment 389080 20141008-PA080088.jpg by ecw11, on Flickr

View attachment 389081 20141008-PA080089.jpg by ecw11, on Flickr

View attachment 389082 20141008-PA080090.jpg by ecw11, on Flickr

View attachment 389083 20141008-PA080091.jpg by ecw11, on Flickr

View attachment 389084 20141008-PA080093.jpg by ecw11, on Flickr

View attachment 389085 20141008-PA080095.jpg by ecw11, on Flickr

View attachment 389086 20141008-PA080097.jpg by ecw11, on Flickr

 

[HarryS]

Agreed. Looks like defect. Are these JPG out of camera?

 

[eteless]

what it reminds me of is an unstable voltage on the amplifier/reset circuits, it switches up each full stop of ISO on the amp (which is why noise doesn't increase as much by increasing ISO if you reach the next full stop (320 to 400 or 640 to 800) for example (which is why it's almost never worth using third stops of ISO, they are essentially the lower stop digitally pushed rather than an analogue gain (unless the E-M5 has changed since the E-520/E-3/E-5, I don't actually know if it's changed for the newer bodies).

At 800 it might shift to a different power supply which has better smoothing thus removing the jitter between lines (as each line is read one at a time). If you remember earlier bodies would get heavy banding - this was partially as you need to keep the charge voltage consistent because small changes result in large differences after amplification, a change in voltage might not drain the pixel or might over amp it once drained, etc.

Really though, this is all just guesswork - I have no clue of the inner design of the sensor and I'm just guessing on the vague design behind CMOS sensors that I can remember, I would guess it's to do with the power supply to the discharge(read) or amplification stage though. Does it change on a fully charged battery vs an empty one?

 

[kwalsh]

That certainly isn't right. And disappearing at ISO800 is strange.

Are these shot RAW or out of camera JPEGs? And if RAW any adjustments to the shadows or just a straight "default" kind of conversion?

I've pushed ISO200 shadows by quite a bit (shooting RAW) and never seen banding like this on my E-M5.

 

[fortwodriver]

Have you got a really noisy power transformer nearby? Anything that uses a "wall wart" or power cube? Does it look better or worse in different rooms of the house?

 

[ecw]

These were all jpgs, but the same thing happens with the raw files.

I don't think its caused by interference from another source, as this still happens when i'm hiking in the middle of no where.

 

[HarryS]

Whatever the cause, it needs fixing, Can't shoot above ISO 800 all the time. Is the camera still under warranty?

 

[ecw]

Luckily, its still under olympus's warranty and I have adorama's extended warranty

 

[Reflector]

what it reminds me of is an unstable voltage on the amplifier/reset circuits, it switches up each full stop of ISO on the amp (which is why noise doesn't increase as much by increasing ISO if you reach the next full stop (320 to 400 or 640 to 800) for example (which is why it's almost never worth using third stops of ISO, they are essentially the lower stop digitally pushed rather than an analogue gain (unless the E-M5 has changed since the E-520/E-3/E-5, I don't actually know if it's changed for the newer bodies).

At 800 it might shift to a different power supply which has better smoothing thus removing the jitter between lines (as each line is read one at a time). If you remember earlier bodies would get heavy banding - this was partially as you need to keep the charge voltage consistent because small changes result in large differences after amplification, a change in voltage might not drain the pixel or might over amp it once drained, etc.

Really though, this is all just guesswork - I have no clue of the inner design of the sensor and I'm just guessing on the vague design behind CMOS sensors that I can remember, I would guess it's to do with the power supply to the discharge(read) or amplification stage though. Does it change on a fully charged battery vs an empty one?

http://home.comcast.net/~nikond70/Charts/RN_e.htm#D3S_14,OM-D E-M5_12

For the E-M5 that occurs around ISO 400. I left the D3s in the chart for the enjoyment of anyone else who wants to see the multiple stages that occur at higher ISOs.

 

[eteless]

Yeap, the plot sort of shows what I meant (less so for the lower ISOs). If you look at 1600 the ones after it are flat, then the same for 3200 and 6400... it actually looks like they're pulling the higher stop rather than pushing the lower stop now though which is interesting.

It's fun to look at how a few different brands do it though, thanks for the link.
Edit: The following link is a perfect example of what I meant, it's almost an ideal graph to show it. http://home.comcast.net/~nikond70/Charts/RN_e.htm#EOS 5D Mark III_14

 

[fortwodriver]

Just out of curiosity. What brand/model SD card are you using?
The banding on these images is very tight... It doesn't look like channel-read noise to me.

 

[klee]

pretty much happens to every one with an EM5+20mm combo. I don't think anyone has successfully gotten around this, including Olympus or Panasonic.

does anybody get this with an EM1 or EM10 body? I really miss the 20mm. amazing glass. terrible AF on the EM5.

 

[barry]

pretty much happens to every one with an EM5+20mm combo. I don't think anyone has successfully gotten around this, including Olympus or Panasonic.

does anybody get this with an EM1 or EM10 body? I really miss the 20mm. amazing glass. terrible AF on the EM5.

He's getting it on the 25mm.

Barry