Banding in shadows with GFX100s

I have some GFX100s images at ISO500 with 16 bit raw that were deliberately underexposed to preserve highlights. The exposure is then pushed 3.5 stops with the result that concentric circles of banding become visible in the sky.

All corrections are off except for the exposure correction (so that, for example, the vignetting and lens distortion corrections are disabled). The same image pushed equivalently in Capture One does not show banding.

Any suggestions how to avoid this?

I don’t see anything wrong.

George

I can’t see anything obvious but, instead of pushing the exposure, try using a little Smart Lighting and bending the tone curve to lift just the shadows…

I can see it and wouldn’t find it acceptable, radiating out and up from the brighter centre. Is there a supported lens module? Did you apply Deep Prime on export? Something is going on

Please share the original RAW file with wetransfer.com or any other suitable service. We can then check what’s in the file and if/how the issue can be avoided.

1 Like

And in addition to the RAW file please include your DOP file.

1 Like

This should not be happening, especially with the GFX100. I do not shoot with this camera, but it is a 100mp sensor, and assuming you converted your RAW file at 16-bit (which you say you did), AND saved in the ProPhoto color space, something ain’t right. Looks like you saved and posted the image as a JPG which is probably the main reason this is happening. Is it happening as a 16-bit TIF in ProPhoto? Lots to dissect here, we definitely don’t have enough information to solve this riddle. sRGB and Adobe RGB will give you this banding, but not ProPhoto, especially with a 16-bit RAW converted to a TIF. I have heard nothing but good things about the GFX100, so something is missing in your processing workflow.

I’ve had the effect in a few images myself and never did the tests to identify the cause. A number of suggestions across the web on why this happens which boil down to either Newton’s rings caused by optical surfaces (e.g. using filters) or software errors (e.g. lens correction).
My own untested theory is that perhaps some mirrorless cameras are susceptible to banding issues due to the close distance between their sensor and the rear lens element.

An example discussion:

I think that’s obvious on its face, but I do not have a problem in my workflow with banding because I am doing it correctly and DxO PL processes my images perfectly every time. Again, this banding problem is mostly correctable with a change in the workflow and export settings.

I’ve been looking dozens of time but I don’t see any banding. Maybe somebody will show me?

George

Hi George. It’s subtle, but it’s there.

Screenshot at 100% magnification of OP’s posted JPG with exposure and contrast increased.

Thanks. Now I see it too. But I’ve to know it’s there.

George

1 Like

This was processed with all corrections off except for a 3.5 stop exposure push - which is fine with this camera at this ISO. I can use smart lighting, clear view etc to make the banding (rings) more visible, but they are visible even with these turned off. They are not the result of the monitor/display bit depth.

Using either Capture One or Affinity Photo no rings are visible, so it is something specific to PL6.

Just curious. Did you use wide gamut or classic gamut?

George

It happens with both wide and classic gamuts, although the rendering changes slightly.

Would you be willing to share the raw file with us?

Mark

1 Like

And the DOP file as well. You will probably need to use a sharing service such as Dropbox and then post the link here.

This kind of banding typically occurs when using sub-sampling (8 bit instead of 16 bit) and it has sometimes happened with the original nikcollection
This also means that we have enough data in the chosen luminance range, and that in this part we should, use the whole luminance range, and then evolve it, and
we come to the intrinsic limit of Dxo without the layers and the associated mode a sophisticated noise processing, can not solve everything.
I use ETTR for this, give me more data in the shadows
not enough data plus the under sampling causes posterization…

If you have Affinity, trial version ?, compare a process, in 16 bit and 8 Bit, you can make your own your own idea