Tests of the banding behaviour with the Panasonic 20mm f/1.7 II lens

Machi

Mu-43 Top Veteran
Joined
May 23, 2015
Messages
629
I promised some time ago in different thread that I'll look to the banding problem of Panasonic 20mm f/1.7 lens,
or more precisely II version which I own.
I did some quick tests already month ago but results were rather inconclusive.
They showed some banding issues which were not present with fully manual ("unelectronic") lens Canon 50mm f/1.4 (my comparison lens) but differences was relatively mild.
This was something unexpected as I have lots of photos which show much more pronounced banding. Since then I wanted to do more thorough testing and I finally did it.

Tests were done with Olympus E-M10II + Panasonic 20mm f/1.7 II lens at ~23°C room temperature. Testing scene was very dark as I wanted to see behaviour under low light conditions (signal is mostly noise from the sensor) but it contains weak red light as I wanted to do majority of tests with active autofocus.

Inserted images demonstrates major points of my comments on the tests.

Subscribe to see EXIF info for this image (if available)


First was test of dependence of banding on shutter speed between 1/4000s up to 1sec for f/1.7 at ISO400 (both electronic and mechanical shutter) and ISO1600 (electronic shutter only). One sequence at ISO400 and mechanical shutter was with noise reduction ON as I thought (based on my older tests) that it can have some effect.
I used Canon FD 50mm f/1.4 with Camdiox focal reducer for comparison and I did it only for scenario where I expected worst results (for this manual lens) - ISO1600, electronic shutter.
Results showed banding in almost all Panasonic images.
Shutter speed has no significant effect on banding.
Images taken at ISO1600 shows much worse banding than those at ISO400.
Images taken with Canon 50mm at ISO1600 don't show any banding at all.
Noise reduction has no visible effect on banding.

Subscribe to see EXIF info for this image (if available)


Another sequence of photos was intended to test effect of electronic shutter, mechanical shutter and mechanical shutter with anti-shake (0sec).
Those were taken at f/1.7 with ISO200, 400 and 1600 and with shutter speeds 1/100s, 1/50s and 1sec.
Banding shows no dependence on type of shutter.
Banding at ISO200 and ISO400 was comparable, ISO1600 was clearly worst of them.

Subscribe to see EXIF info for this image (if available)


Then there was a test about dependence of banding on ISO settings (all analog settings 200 up to 3200 + 6400). Here the results were most interesting.
Based on previous tests with ISO200, 400 and 1600, I expected more banding with higher ISO.
That wasn't exactly what happended.
From ISO200 up to ISO640 results were similar with clear but weak banding.
At ISO800 banding was very bad. The same happens with ISO1000, 1250 and 1600 (as it was already observed in previous tests).
Surprisingly at ISO 2000, 2500, 3200 and 6400, banding was effectively nonexistent (comparable to the Canon lens).

Subscribe to see EXIF info for this image (if available)


Last two tests was only quick ones.
First was about dependence on aperture settings. I did it only at ISO1600 and 1/50sec at f/1.7, 2.8, 4.0, 5.6 and 8.
Changing aperture has no effect on results (they were all similarly bad as expected for ISO1600).
Second one was about different AF settings again for ISO1600, 1/50sec and at f/1.7.
Only three images were taken with S-AF, C-AF and MF settings. No visible difference in results.

My conclusions for now are that banding is strongly dependent on ISO settings where at low ISO (200-640) banding is visible but not significant for most practical cases.
At ISO800 up to 1600 banding is strong and it probably affects photos.
For high ISO settings (2000-3200, 6400) banding isn't a problem.
Other settings, which I tested (shutter type, shutter speed, type of focus), have lower (if any) impact on banding.

Next I want to repeat tests with my other Olympus camera (E-PM2) and I want to test banding behaviour at shutter speeds >1 sec.

Also if someone want to add more data to this thread, it would be helpful and I'll appreciate it as all my results could be caused by my copy of lens (or camera) and behaviour could be different for the other copies.
It's easy to test some basic conclusion of my tests.
It only takes 4 images (RAWs) at 1/50s f/1.7 with lens cover on at ISO640, ISO800, ISO1600 and ISO2000 ("dark frames"). According to my results ISO640 will show weak banding, ISO800 and ISO1600 will show strong banding and ISO2000 will show no banding.
 
Last edited:

Egregius V

Mu-43 Top Veteran
Joined
Jun 14, 2015
Messages
652
Location
Massachusetts, USA
Real Name
Rev. Gregory Vozzo
A while back, I tested banding under rather different conditions. I noticed that someone reproduced it reliably with sunlit venetian blinds in the picture, so I did that and took some other high-contrast photos exposed normally (using the histogram as a guide). Banding showed up in the shadow areas at all ISOs and was more severe at smaller apertures (barely visible at f/1.7). Testing with the E-M5 II, I found that there was zero banding using the electronic shutter (JPEG and RAW). So I'm rather surprised at your test results. I can't tell what's going on - all of your photos are labeled 1/50s shutter speed - but am suspicious of your light source, being it's artificial and therefore subject to flickering. That will contaminate your test results. Any artificial light in front of the camera (even if outside the picture) will introduce banding at certain shutter speeds, especially with the electronic shutter.

Another thing I observed is that the banding pattern varies from camera to camera, leaving a band-free area in some part of the frame. I tested an E-PL5, E-M10 (original), and E-M5 II. Your test images show banding across the whole image. I don't think it's the same as what I tested for. (Look here for a sample.)
 

Egregius V

Mu-43 Top Veteran
Joined
Jun 14, 2015
Messages
652
Location
Massachusetts, USA
Real Name
Rev. Gregory Vozzo
When there is high contrast, and a reliable banding-free approach can't be utilized, merging bracketed shots into an HDR image can provide a banding-free result. Here is an example:

Low-light image with overexposed parts:

Subscribe to see EXIF info for this image (if available)


Underexposed image showing some banding:
Subscribe to see EXIF info for this image (if available)


A more underexposed image to merge with the first:
Subscribe to see EXIF info for this image (if available)

The result, using Photomatix Pro:
Subscribe to see EXIF info for this image (if available)
 

Machi

Mu-43 Top Veteran
Joined
May 23, 2015
Messages
629
I can't tell what's going on - all of your photos are labeled 1/50s shutter speed - but am suspicious of your light source, being it's artificial and therefore subject to flickering. That will contaminate your test results. Any artificial light in front of the camera (even if outside the picture) will introduce banding at certain shutter speeds, especially with the electronic shutter.
From practical reasons I showed only images at 1/50s (it's favorite automatic shutter speed of my camera with P20 at low light conditions).
But in the first test I tested other shutter speeds (1/4000 up to 1s) and it has no significant effect on results.
Used red light looks almost flicker free as there is no banding in the images taken with Canon (or with P20 at ISO2000+).
I also now tested what I recommended in the last paragraph and results are the same even without the light.

Did you try setting the 20mm using manual focus as opposed to AF? I wonder if the AF itself effects banding.
Yes, I tried it but only quickly. Three images at ISO1600 with MF, S-AF and C-AF are in the last composite image.
That's another thing which I want to more cover in other tests (different AF settings with ISO different than 1600).
 
Last edited:

robcee

Mu-43 Top Veteran
Joined
Jan 10, 2016
Messages
567
Location
New Brunswick, Canada
Real Name
Rob Campbell
Did you try setting the 20mm using manual focus as opposed to AF? I wonder if the AF itself effects banding.
having owned a P20 and tested it on both the EM5mk1 and EM1mk1, I theorized that the older focus mechanism on that lens was causing a greater power draw which manifested in the electronics in the sensor. I don't recommend that lens for users of Olympus cameras for this reason. The banding (and poor autofocus) makes that lens not really suitable for any kind of light-challenged shooting.

Always nice to see more science though! I'm really amazed Panasonic hasn't updated this with a more modern autofocus motor / design given its popularity.
 
Joined
Sep 10, 2016
Messages
1,209
When there is high contrast, and a reliable banding-free approach can't be utilized, merging bracketed shots into an HDR image can provide a banding-free result. Here is an example:

Low-light image with overexposed parts:

View attachment 545632

Underexposed image showing some banding:
View attachment 545633

A more underexposed image to merge with the first:
View attachment 545635
The result, using Photomatix Pro:
View attachment 545636
Were these pictures taken manually by taking multiple pictures, or was it automatic using in camera bracketing?
 

hazwing

Mu-43 Hall of Famer
Joined
Nov 25, 2012
Messages
2,136
Location
Australia
Thanks, appreciate these tests, even though I don't have a 20mm 1.7

Just to clarify, iso 400 vs 1600... are you pushing the exposure the ISO 400 to match be similar to the ISO 1600. Are you pushing the exposure of the 1600 as well, and if so how many stops?
 

Egregius V

Mu-43 Top Veteran
Joined
Jun 14, 2015
Messages
652
Location
Massachusetts, USA
Real Name
Rev. Gregory Vozzo
Were these pictures taken manually by taking multiple pictures, or was it automatic using in camera bracketing?
Bracketing, hand-held.

This was a really dark scene. The only light in the room was coming from the Virgin Mary's crown and a few candles.

Machi, you've clearly found a way to bring out the worst in your 20mm lens.:th_salute: I, too, would like to know more about how you're exposing these shots and what else might be going on. (The banding doesn't normally disappear with high ISOs; if anything, it becomes more noticeable, unless you're changing something else like exposure as ISO increases.)

How much banding are you seeing with real-world usage of this camera and lens? It shouldn't be this bad.
 
Last edited:

Machi

Mu-43 Top Veteran
Joined
May 23, 2015
Messages
629
having owned a P20 and tested it on both the EM5mk1 and EM1mk1, I theorized that the older focus mechanism on that lens was causing a greater power draw which manifested in the electronics in the sensor. I don't recommend that lens for users of Olympus cameras for this reason. The banding (and poor autofocus) makes that lens not really suitable for any kind of light-challenged shooting.
I cannot agree. One of the major reason why I did those tests was to find conditions under which banding isn't problem.
And I think that I found them.
P20 is good lens but it has few bugs. :)

Always nice to see more science though! I'm really amazed Panasonic hasn't updated this with a more modern autofocus motor / design given its popularity.
Thanks!
BTW, I think that for Panasonic it's relatively negligible problem.
 

Machi

Mu-43 Top Veteran
Joined
May 23, 2015
Messages
629
Thanks, appreciate these tests, even though I don't have a 20mm 1.7

Just to clarify, iso 400 vs 1600... are you pushing the exposure the ISO 400 to match be similar to the ISO 1600. Are you pushing the exposure of the 1600 as well, and if so how many stops?
Yes, ISO400 is matched with ISO1600.
Exposure compensation is +7 for the ISO400 and +5 for the ISO1600.
It effectively shows read-out noise from the sensor.
 

Machi

Mu-43 Top Veteran
Joined
May 23, 2015
Messages
629
Machi, you've clearly found a way to bring out the worst in your 20mm lens.:th_salute: I, too, would like to know more about how you're exposing these shots and what else might be going on. (The banding doesn't normally disappear with high ISOs; if anything, it becomes more noticeable, unless you're changing something else like exposure as ISO increases.)
All results are with very high exposure compensation (+~5 up to +~7).
So they shows mostly readout noise from the sensor (which is affected by banding).
In normal usage, with proper exposure, weak banding would not be visible.
Strong banding between ISO800 and ISO1600 is visible under right circumstances (see below).

How much banding are you seeing with real-world usage of this camera and lens? It shouldn't be this bad.
I found that in real world usage I have no problem at low ISO settings.
I found many cases between ISO800 and ISO1600 where banding is visible in darker, flat parts of photos (as expected for the read-out noise effect).
Major problem with it is that it's a much harder nut to crack for the denoising algorithms than nice homogenous noise,
normally produced by the most sensors.

But I really don't want to scare someone. Properly exposed photo taken with P20 will be alright under most circumstances.
I think that slow AF is much bigger problem of this lens than banding (and even with slow AF it's a nice lens).
All tests which I have done are about pushing the boundaries of photographic technique.
 

SojiOkita

Mu-43 All-Pro
Joined
Feb 23, 2014
Messages
1,424
Location
France
On my sample, there is no banding if I use f/1.7.
As soon as I change the aperture (f/1.8), banding occurs (more visible at high ISO).
I have one example of banding at f/1.7, though, but only one over a large number of photos.

It does not match your observation, so it's highly possible that the banding phenomenon is different among samples.
I don't think it depends on the sensor sample, because I tested my lens on several bodies with similar sensors (my E-M10, E-M10 II, E-M5 II) with the same results.
Of course, on bodies like E-M1 or GM5, no banding issue.
 

Impulse

Mu-43 Regular
Joined
Apr 18, 2014
Messages
83
Location
Puerto Rico
Real Name
Frank
If you can make it work, good on you. It is a great lens optically, and no, pannies don't suffer the same problem.
Unless you happen to have a GH3 or some other Pana body the used the same kinda Sony sensor... :p
 

stiblee

Mu-43 Regular
Joined
Dec 11, 2016
Messages
44
This banding issue has been around since the Olympus E-M5 and Panasonic 20mm f1.7 version 1 (version 2 just has new cosmetics):

Olympus acknowledges E-M5 / 20mm banding and is working on fix


Olympus acknowledges E-M5 / 20mm banding and is working on fix

After checking every possible combination of a body and a lens, we found the phenomena only with this combination (OM-D, E-M5 coupled with the Panasonic 20mm pancake lens). We are continuing to study how we can eliminate this and we recommend for our customers using E-M5 with Panasonic 20mm pancake lens to keep a low ISO to avoid this problem for the time being.

In the Panasonic GH3 manual they also warn about banding with the 20mm f1.7.

Here is a post by one of the dpreview staff about it:

Re: 20 mm lens banding on EM1 and GX7?: Micro Four Thirds Talk Forum: Digital Photography Review

Here is just one of the many long threads that fully document the banding problem:

Panasonic Lumix G 20mm F1.7 II (H-H020A) - Banding: Micro Four Thirds Talk Forum: Digital Photography Review

Pay particular attention to the tsi and Anders W posts. You will see that tsi repeatedly disassembled his 20mm, removed a part, reassembled, and tested. Each time he would then replace the part and remove a different part. He finally isolated what seemed to be causing the banding for his 20mm and camera body combination.
 

MarkRyan

Instagram: @MRSallee
Joined
May 3, 2013
Messages
856
Location
California
Pay particular attention to the tsi and Anders W posts. You will see that tsi repeatedly disassembled his 20mm, removed a part, reassembled, and tested. Each time he would then replace the part and remove a different part. He finally isolated what seemed to be causing the banding for his 20mm and camera body combination.
Thanks for the link. Bravo to those two, both for the diligence in tracking down the issue and for admirably putting up with some truly idiot "dispute" from someone with zero experience with the scenario they were testing.

I don't shoot over ISO400 very often, but did get one shot spoiled by banding with the 20/1.7 at ISO200. I believe it was on E-P5. I haven't used the 20/1.7 much on my E-M5ii.
 
Links on this page may be to our affiliates. Sales through affiliate links may benefit this site.
Mu-43 is a fan site and not associated with Olympus, Panasonic, or other manufacturers mentioned on this site.
Forum post reactions by Twemoji: https://github.com/twitter/twemoji
Copyright © 2009-2019 Amin Forums, LLC
Top Bottom