Can you fix a heavy banding problem in my exposure?

Taken handheld (on a knee) at 1/4sec, ISO 3200, f/2, exposed for the highlights.

Leica M240 is not a forgiving camera when it comes to highlights clipping and the image will also degrade considerably at ISO 3200.

I will leave you with a creative carte blanche with this file. I pushed the exposure of the water by 2.00 stops using a gradient and I also used tone equalizer with “-6 EV” band pushed +1.5 EV, other bands accordingly.

This kind of heavy processing leaves the file with very noticeable horizontal banding (although now vertical since I took it in portrait orientation). I played with GMIC, Rawtherapee, Darktable. Couldn’t really achieve a result that’d be any better than just leaving the banding there.

MP003322.DNG (25.0 MB)

That’s a tough one, I would also be curious about how to deal with this defect … If I understand correctly, in the first place it’s due to the high iso exposure but I guess this kind of issue shared by many (all ?) cameras.
The real solution would be to have dedicated tool to autodetect and remove banding in our raw processing softwares ? I’m curious to know how popular non open source alternative are dealing with that ?

Recent versions of Rawtherapee have a “line noise filter” that is suited for certain patterns of banding noise that on-sensor-PDAF sensors may exhibit when pushed. But that kind of noise is better-behaved than what this Leica is doing. I did not have success with the line noise filter on this file.

RT would be to take a series of shots to enable auto select of dark frames

is a method of dealing with “fixed pattern noise”, please report back on success or failure for other users.

1 Like

Difficult! darktable plus a little “destripe” in gimp.

MP003322.DNG.xmp (8.9 KB)

1 Like

I made some dark frames just here and now using the same settings I did on the field.

But darkframe subtraction doesn’t effectively improve the issue. Above, Dark-Frame disabled, below, enabled.

Cool take, very smoky/hazy feel to it. :slight_smile:

Yeah, looks and feels also a little like a painting.

My original/first processing also brings an oil painting to mind. The banding resembles the not fully flat canvas behind the paint. The crude red noise artifacts add a touch towards the same. :smiley:

The hazy/smoky appearence comes from the heavy use of denoising. In general I try to avoid that. But in this case it gives a nice touch without reducing to much (meaningful) details

Where did you take that shot?.

Taken abt 2 weeks ago in Helsinki, Finland.

I’m totally not a tripod person but I’m afraid I have to start considering carrying one…

Here is a RawTherapee effort.
MP003322.jpg.out.pp3 (12.2 KB)


There’s a 7 year old discussion here:

with a steve huff example.

And here is my take on this using darktable 3.3.0+1993~ga5829cf6c

MP003322.DNG.xmp (19,3 KB)

1 Like

MP003322.DNG.xmp (11.5 KB) dt 3.3.0

1 Like

What banding, where?

In the water, I can see vertical bands of dark and light. These are caused by reflections of distant lights and the sky.

There may be some fine-grained vertical banding in the sky at the right, above the horizon. This could be caused by rain.

I suspect that if you re-took the photo, rotating the camera 45 degrees around the lens axis, the entire picture including the banding would also rotate.

@snibgo There might be a confusion regarding the term ‘banding’ here? Or a serious screen calibration issue on your side maybe.

Regarding this picture, there is a visible line-amplifier inhomogenity present. With the portrait orientation of the picture those are vertical lines of gain variations (most likely offset problems, but gain imhogenity might be on top as well). Those vertical lines are a sensor artefact and more pronounced in the shadows but still visible throughout all gradations, just a bit less so the more light the sensor receives.

This is not light reflections on water and not rain (the lens perspective would impact the parallelity of those rain streaks towards the corners, especially top left and right corners, as the rain would be falling towards the lens slightly missing it and not like seen here perfectly parallel to sensor orientation).

@mike3996 took dark frames for subtraction, if in doubt I bet he would be willing to upload those too, so you can see that this is purely from the sensor, not from reflections or rain. This is a common problem of sensors and takes some circuitry to counteract. I know this from Canon sensors and apparently this generation of Leica sensors has a similar artefact. Calling it ‘banding’ is a bit deceiving as this usually means something slightly different though, so you might be looking for something else? Also: this artefact is practically jumping into my face on my (calibrated) monitor, so maybe there is a viewing problem which compounds properly detecting it on your side? Not sure.

Edit: I replied to Alan Gibsons post, but it showed up as a non-linked post, so I put Alans handle at the front. Usually the forum does this right…weird.

1 Like

Thanks, @PhotoPhysicsGuy.

Yes, maybe I have a monitor/eyesight/understanding problem. That’s why I asked: where is the visible banding?

I could see vertical bands/streaks/areas of light and dark in the water. These are quite wide, 20 pixels or more. But they stopped at the far shore, and were not evident in the distant shore or sky.

However, after developing with dcraw and ImageMagick “-equalize rotate 90” I can now clearly see what are now colourful bands spreading horizontally, all the way across, most visible where the image is dark. These are more narrow, around 5 pixels wide. I show a JPEG of that image, just for interest.

("-equalize" makes an ugly result, but helps me see what is really there.)

This banding isn’t fixed by dividing this horizontal image by twice the average column, so the problem isn’t consistent under/over readings of rows.

I suppose they are caused by in-camera noise that has some consistency within rows.

1 Like

Here’s one dark frame.

ISO 3200, 1/4 sec

All kinds of stripes can be observed when pushed enough. A sample from one of my darkframes, not necessarily the one I uploaded here:

MP003436.DNG (18.9 MB)

1 Like

I have shot 30k+ frames with my M240 over the last two years, very familiar with this issue. But this is the first time I wanted to ask around if anyone knows a trick or two help combat this problem in PP.

All cameras will suffer some banding when pushed excessively but M240 and many Leicas perform very poorly in this regard.

The cure to banding is to avoid heavy underexposure. Even modern Nikons and Sonys can show some banding if the file has to be pushed 5+ stops. Remain within sane limits and you won’t have a problem. Sadly for Leica, two stops at higher ISOs can already be too much to handle.

If one wants to push the shadows by two stops in this scene the exposure should be done at ISO 800 preferably. (Visible banding will start to occur whenever the “effective ISO” approaches 6400 and above.) Then the file would remain pretty good in all likelihood. But handholding gets me only so far, so one-second portrait exposures are off the table. I’m frankly surprised I managed to get a sharp shot at 1/4s!

1 Like