Banding from Olympus EM5ii

Started Oct 7, 2020 | Discussions
VinceC Senior Member • Posts: 1,296
Banding from Olympus EM5ii

I'm posting below two versions of a recent photo shot in RAW and processed in Lightroom Classic. The finished photo, converted to sepia and quite heavily processed (see screenshot for history) shows quite distinct banding in the lighter area of the couuds. I've included the "original" colour image, imported into Lightroom Classic and saved as a JPEG.

I've been shooting with the M5ii for about 8 months and never seen this banding effect before. The image has been pushed quite hard in LR, but the banding must already be there in the original. Shooting conditions are shown on the LR screenshot. Any suggestions as to what I can do to avoid it (in this kind of image)?

Thanks!

-- hide signature --

All the best!
Vince C

 VinceC's gear list:VinceC's gear list
Olympus E-M5 III Nikon Z5 Panasonic 100-300mm F4-5.6 II Olympus M.Zuiko Digital 45mm F1.8 Nikon Z 24-200mm F4-6.3 VR +3 more
Tom Axford Veteran Member • Posts: 9,020
Re: Banding from Olympus EM5ii

Interesting!  I've never seen anything quite like that before.

victorav Senior Member • Posts: 2,235
Re: Banding from Olympus EM5ii

Still a good shot even with banding!

Digital Dick Senior Member • Posts: 2,302
Re: Banding from Olympus EM5ii
4

The vertical banding in the B&W version appears to follow the gradient filter. Even in the B&W version I'm not seeing the banding beyond the end of the gradient filter. Maybe caused by the range settings you used when applying the gradient filter??

Never seen it before in any of my color conversions to B&W.

Dick

 Digital Dick's gear list:Digital Dick's gear list
Olympus OM-D E-M5 Olympus E-M5 II
knickerhawk Veteran Member • Posts: 7,350
The likely cause is...
10

Notice how the bands get progressively wider as you move from left to right? That's a telltale sign of the application of a gradient at a less than ideal bit depth. Indeed, the history indicates that you used the graduated filter effect, which is a type of gradient. The real question is whether you converted the original raw to 8-bit or otherwise saved it in 8-bit along the way prior to the application of the graduated filter effect. If you did, then it's very likely the case that you wouldn't see that problem if you had done all of the processing in 16-bit. If all processing took place in the 16-bit domain, I'm rather surprised at these results, but they're still almost certainly related to the grad filter effect.

knickerhawk Veteran Member • Posts: 7,350
Re: Banding from Olympus EM5ii

Digital Dick wrote:

The vertical banding in the B&W version appears to follow the gradient filter. Even in the B&W version I'm not seeing the banding beyond the end of the gradient filter. Maybe caused by the range settings you used when applying the gradient filter??

Never seen it before in any of my color conversions to B&W.

Dick

Damn! You beat me by a minute!  Let's see what the OP says about my 8-bit speculation.

kenw
kenw Veteran Member • Posts: 6,926
Re: The likely cause is...

knickerhawk wrote:

Notice how the bands get progressively wider as you move from left to right? That's a telltale sign of the application of a gradient at a less than ideal bit depth. Indeed, the history indicates that you used the graduated filter effect, which is a type of gradient. The real question is whether you converted the original raw to 8-bit or otherwise saved it in 8-bit along the way prior to the application of the graduated filter effect. If you did, then it's very likely the case that you wouldn't see that problem if you had done all of the processing in 16-bit. If all processing took place in the 16-bit domain, I'm rather surprised at these results, but they're still almost certainly related to the grad filter effect.

His screen shot seems to show all the edits are on the original ORF file so this doesn't seem to be the likely cause in this case.

-- hide signature --

Ken W
See profile for equipment list

 kenw's gear list:kenw's gear list
Panasonic Lumix DMC-GM1 Nikon Z7 Panasonic Leica DG Summilux 15mm F1.7 ASPH Nikon Z 14-30mm F4 Nikon Z 24-200mm F4-6.3 VR +42 more
kenw
kenw Veteran Member • Posts: 6,926
Re: Banding from Olympus EM5ii
1

Digital Dick wrote:

The vertical banding in the B&W version appears to follow the gradient filter. Even in the B&W version I'm not seeing the banding beyond the end of the gradient filter. Maybe caused by the range settings you used when applying the gradient filter??

It appears the gradient is likely increasing contrast and potentially clarity, both of which would increase the visibility of any sort of banding.  So yes, I think you are right about those more extreme settings making it visible, whatever the source.

-- hide signature --

Ken W
See profile for equipment list

 kenw's gear list:kenw's gear list
Panasonic Lumix DMC-GM1 Nikon Z7 Panasonic Leica DG Summilux 15mm F1.7 ASPH Nikon Z 14-30mm F4 Nikon Z 24-200mm F4-6.3 VR +42 more
kenw
kenw Veteran Member • Posts: 6,926
Re: The likely cause is...

knickerhawk wrote:

Notice how the bands get progressively wider as you move from left to right? That's a telltale sign of the application of a gradient at a less than ideal bit depth. Indeed, the history indicates that you used the graduated filter effect, which is a type of gradient. The real question is whether you converted the original raw to 8-bit or otherwise saved it in 8-bit along the way prior to the application of the graduated filter effect. If you did, then it's very likely the case that you wouldn't see that problem if you had done all of the processing in 16-bit. If all processing took place in the 16-bit domain, I'm rather surprised at these results, but they're still almost certainly related to the grad filter effect.

So looking at the edit history more, I wonder if even if processed in 16-bit as this appears to be if you are still on the right track.  There are some huge saturation moves and such there which depending on how the B&W conversion was done could result in forms of gamut clipping that might have similar effects to low-bit processing.  People have seen similar unexpected posterization even working from RAW once odd edge of gamut effects come into play.

OP, can you describe what you've done with the editing as far as sliders go?  Or perhaps actually without changing anything else in the settings create a color version?  If it looks really wonky in color that could cause gamut issues that look like luminance banding in the B&W conversion.

Mostly guesswork here!

-- hide signature --

Ken W
See profile for equipment list

 kenw's gear list:kenw's gear list
Panasonic Lumix DMC-GM1 Nikon Z7 Panasonic Leica DG Summilux 15mm F1.7 ASPH Nikon Z 14-30mm F4 Nikon Z 24-200mm F4-6.3 VR +42 more
OP VinceC Senior Member • Posts: 1,296
Re: The likely cause is...

I'm not aware that anything I have done converted the file of 8 bit. It didn't take any trips to PS.

Looking carefully at the steps the banding was present before I applied the gradient.

-- hide signature --

All the best!
Vince C

 VinceC's gear list:VinceC's gear list
Olympus E-M5 III Nikon Z5 Panasonic 100-300mm F4-5.6 II Olympus M.Zuiko Digital 45mm F1.8 Nikon Z 24-200mm F4-6.3 VR +3 more
OP VinceC Senior Member • Posts: 1,296
Re: The likely cause is...

Or perhaps actually without changing anything else in the settings create a color version? If it looks really wonky in color that could cause gamut issues that look like luminance banding in the B&W conversion.

Mostly guesswork here!

Virtual copy converted to colour:

-- hide signature --

All the best!
Vince C

 VinceC's gear list:VinceC's gear list
Olympus E-M5 III Nikon Z5 Panasonic 100-300mm F4-5.6 II Olympus M.Zuiko Digital 45mm F1.8 Nikon Z 24-200mm F4-6.3 VR +3 more
kenw
kenw Veteran Member • Posts: 6,926
Re: The likely cause is...

VinceC wrote:

Or perhaps actually without changing anything else in the settings create a color version? If it looks really wonky in color that could cause gamut issues that look like luminance banding in the B&W conversion.

Mostly guesswork here!

Virtual copy converted to colour:

Thanks Vince!  That really doesn't look severe at all color wise.  So I doubt this is related to gamut clipping at all then.  And as you confirmed this is all RAW editing doesn't seem like it could be related low bit processing posterization either.

I've seen banding patterns similar to this before from nearby electrical interference but that was usually on older sensors that used off chip ADCs.  Also this banding is rather "clean" in comparison.

Hmmm.....

-- hide signature --

Ken W
See profile for equipment list

 kenw's gear list:kenw's gear list
Panasonic Lumix DMC-GM1 Nikon Z7 Panasonic Leica DG Summilux 15mm F1.7 ASPH Nikon Z 14-30mm F4 Nikon Z 24-200mm F4-6.3 VR +42 more
Lichtspiel
Lichtspiel Veteran Member • Posts: 3,192
Re: Banding from Olympus EM5ii
1

Reminds me of the pattern you get sometimes with scanned images.

If you want an independent verification, post a link to the ORF original. I processed your color "original" and got the same banding when increasing contrast, in all color channels. So yes, it is already present in the color version (JPG).

If you don't get an answer here, consider posting over at the Retouching forum, where a lot of artifact experts lurk.

 Lichtspiel's gear list:Lichtspiel's gear list
Sony a7C Sony FE 20mm F1.8G Tamron 70-300 F4.5-6.3 Di RXD III Samyang AF 35mm F1.8 FE Voigtlander 15mm F4.5 Super Wide Heliar +5 more
knickerhawk Veteran Member • Posts: 7,350
Re: Banding from Olympus EM5ii

Lichtspiel wrote:

Reminds me of the pattern you get sometimes with scanned images.

If you want an independent verification, post a link to the ORF original. I processed your color "original" and got the same banding when increasing contrast, in all color channels. So yes, it is already present in the color version (JPG).

Which, to me, is just further confirmation that this is 8-bit processing issue. However, the mystery remains as to how it got into 8-bit mode.

If you don't get an answer here, consider posting over at the Retouching forum, where a lot of artifact experts lurk.

knickerhawk Veteran Member • Posts: 7,350
Re: The likely cause is...
1

VinceC wrote:

I'm not aware that anything I have done converted the file of 8 bit. It didn't take any trips to PS.

Looking carefully at the steps the banding was present before I applied the gradient.

I'm surprised to learn this. However, I don't personally use LR (I'm an ACR/PS guy), so I'm stumbling around a bit to give you a good explanation. One thing to confirm: you didn't run this image through any external tools such as one of the Topaz tools, did you?

knickerhawk Veteran Member • Posts: 7,350
Re: The likely cause is...
2

kenw wrote:

VinceC wrote:

Or perhaps actually without changing anything else in the settings create a color version? If it looks really wonky in color that could cause gamut issues that look like luminance banding in the B&W conversion.

Mostly guesswork here!

Virtual copy converted to colour:

Thanks Vince! That really doesn't look severe at all color wise. So I doubt this is related to gamut clipping at all then. And as you confirmed this is all RAW editing doesn't seem like it could be related low bit processing posterization either.

I've seen banding patterns similar to this before from nearby electrical interference but that was usually on older sensors that used off chip ADCs. Also this banding is rather "clean" in comparison.

Hmmm.....

Hmmmm, for sure. This has the classic look of 8-bit banding introduced during processing. I'm not surprised at all that it's not gamut related. It's a cloudy sky! Not much color there.

Lichtspiel
Lichtspiel Veteran Member • Posts: 3,192
Workaround

This won't save this image, but in future, you can avoid excessive "pushing" by doing a few exposure brackets, and merge them afterwards. Doesn't always have to result in a garish maximum HDR effect...

 Lichtspiel's gear list:Lichtspiel's gear list
Sony a7C Sony FE 20mm F1.8G Tamron 70-300 F4.5-6.3 Di RXD III Samyang AF 35mm F1.8 FE Voigtlander 15mm F4.5 Super Wide Heliar +5 more
JakeJY Veteran Member • Posts: 5,314
Re: Banding from Olympus EM5ii
8

I threw your color JPEG into Photoshop and was able to get banding to show by playing with levels, so it has nothing to do with the B&W filter.

A quick google found that this is likely related to EFCS ("anti-shock"). It causes banding along the read direction (so banding is horizontal in landscape and vertical in portrait). This effect is most visible at 1/320-1/350, which your photo is at.

See the following thread:

https://www.mu-43.com/threads/anti-shock-banding-on-e-m5-ii.87933/

 JakeJY's gear list:JakeJY's gear list
Nikon Coolpix S9300 Nikon D5000 Panasonic Lumix DMC-GX85 Nikon AF-S DX Nikkor 18-55mm f/3.5-5.6G VR Nikon AF-S DX Nikkor 55-200mm f/4-5.6G VR +6 more
kenw
kenw Veteran Member • Posts: 6,926
Re: Banding from Olympus EM5ii
1

JakeJY wrote:

I threw your color JPEG into Photoshop and was able to get banding to show by playing with levels, so it has nothing to do with the B&W filter.

A quick google found that this is likely related to EFCS ("anti-shock"). It causes banding along the read direction (so banding is horizontal in landscape and vertical in portrait). This effect is most visible at 1/320-1/350, which your photo is at.

See the following thread:

https://www.mu-43.com/threads/anti-shock-banding-on-e-m5-ii.87933/

Excellent find! That's it for sure. Same change in period of the banding which makes perfect sense for EFCS which uses essentially a time varying reset and read process. It doesn't exactly follow the mechanical shutter, it lags it a bit at one end of the exposure and leads it a bit on the other. This is why you can get exposure variation across the frame when using very fast shutter speeds on EFCS if the camera doesn't have perfect knowledge of the exit pupil of the lens. It fails to use the correct lead/lag timing to make up for the difference in distance of the electronic and mechanical shutters from the surface (the electronic one being zero of course).

But a side effect is probably that as it varies the read/reset cadence it results in very slight differences in exposure for each row because these reads/resets happen in discrete exposure increments (usually a row read time is the minimum increment). And that would perfectly explain what we see here and in the link.

-- hide signature --

Ken W
See profile for equipment list

 kenw's gear list:kenw's gear list
Panasonic Lumix DMC-GM1 Nikon Z7 Panasonic Leica DG Summilux 15mm F1.7 ASPH Nikon Z 14-30mm F4 Nikon Z 24-200mm F4-6.3 VR +42 more
OP VinceC Senior Member • Posts: 1,296
Re: Banding from Olympus EM5ii
2

Many thanks, I suspect you have identified the cause.  Something to be aware of in future.  Now, where’s thar sky replacement tutorial?

-- hide signature --

All the best!
Vince C

 VinceC's gear list:VinceC's gear list
Olympus E-M5 III Nikon Z5 Panasonic 100-300mm F4-5.6 II Olympus M.Zuiko Digital 45mm F1.8 Nikon Z 24-200mm F4-6.3 VR +3 more
Keyboard shortcuts:
FForum MMy threads