Banding in sky

Started May 25, 2013 | Discussions
OP CLYoung Regular Member • Posts: 204
Re: Banding in sky

Thanks - the exposure is definitely the difference between the two, obviously at the time I realised I'd got it wrong and corrected but I overlooked that completely when reviewing them.

 CLYoung's gear list:CLYoung's gear list
NEX5R Nikon D610 Nikon AF-S Nikkor 70-300mm f/4.5-5.6G VR Sony E 16mm F2.8 Pancake Nikon AF-S Nikkor 50mm f/1.8G +7 more
OP CLYoung Regular Member • Posts: 204
Re: Banding in sky

Thanks, I will give those a go.

 CLYoung's gear list:CLYoung's gear list
NEX5R Nikon D610 Nikon AF-S Nikkor 70-300mm f/4.5-5.6G VR Sony E 16mm F2.8 Pancake Nikon AF-S Nikkor 50mm f/1.8G +7 more
LifeIsAVerb Senior Member • Posts: 1,773
Re: Banding in sky

blue_skies wrote:

In the vertical shots I can see 'bands' that follow the horizon roughly, I count maybe 6 or 7 or so.

This is not a JPG artifact, even thought the bands follow the sky gradation towards darker.

It could be a reflection of sorts inside the lens, or between the sensor and the lens.

Perhaps the lens surface or the sensor surface has streaks? (It is definitely time to clean the sensor too )

The horizontal shot does not show this which hints at some directionality.

It's more subtle, but i can see it in the horizontal shot, too (and i have an ancient iMac). Shows up better for me when viewed at an angle (from below).

LifeIsAVerb Senior Member • Posts: 1,773
Re: Banding in sky

twald wrote:

In think this is called posterization, after posters. The smooth transition between gray to white in the sky is reflected, i.e., recorded by just a few data levels and jpep attempts to get this slow progression compressed and messes up..
in LR you can add grain manually in special effects and that dithers the data levels and does not allow jpeg to make these line like distinctions that you call bands.

hHappens for many same (nearly so) light and color large images.

I think you are correct; it is posterization caused by jpeg compression. It is not present in the second photo because the exposure is different.

But it is present in the second (horizontal) photo, it's just less pronounced. Try viewing the image at an angle (from below).

GaryR60 Contributing Member • Posts: 661
Re: Vertical huh

CLYoung wrote:

Thanks, yes, it's circular banding. And how I missed before that I'd gone from +1 compensation to 0 I don't know. I obviously noticed at the time and changed it (I'd been taking hot air balloons against the sky). So nothing to do with vertical vs horizontal.

Were you using the NEX-7, by any chance? I've heard it's easy to bump the two knobs and accidentally change settings. I tend to notice the banding in bright skies, myself, but it usually doesn't become evident unless I really crank up the Structure slider in either HDR Efex Pro or Silver Efex Pro in post.

OP CLYoung Regular Member • Posts: 204
Re: Vertical huh

No, a 5R, but it's still quite easy to slip up. In this case though I think it had been deliberate, I just took a shot before remembering to change back.

 CLYoung's gear list:CLYoung's gear list
NEX5R Nikon D610 Nikon AF-S Nikkor 70-300mm f/4.5-5.6G VR Sony E 16mm F2.8 Pancake Nikon AF-S Nikkor 50mm f/1.8G +7 more
D Cox Forum Pro • Posts: 22,953
Re: Banding in sky

chromnd wrote:

D Cox wrote:

This banding is not visible on my analog monitor (a good quality Iiyama).

I downloaded the "original" and made a quick A4 print. The banding doesn't show on the print either.

I wonder if this is a monitor problem. How many bits per channel are you sending to your monitor ? Banding basically occurs when a gradient between closely similar colours is not rendered with enough bits.

I did not see it as well, but when you "push" the original a little bit, a structure in the sky becomes visible:

http://i.imgur.com/OzSbibL.jpg

I experience such problems sometimes, when I upload stuff to facebook - it's really annoying. I guess that's a JPEG engine problem. Personally I only shoot RAW and never experienced such a problem.

Clearly not enough bits to give a smooth gradation. Maybe the JPEG engine is calculating at only 8 bits per channel. (JPG images are normally only 8 bits per channel.)

RGBaker Contributing Member • Posts: 962
Re: Banding in sky

LifeIsAVerb wrote:

It's more subtle, but i can see it in the horizontal shot, too (and i have an ancient iMac). Shows up better for me when viewed at an angle (from below).

It is difficult to make comments on 'banding' particularly for reasons like this -- older iMacs have a tendency to banding; many displays, even high-grade image displays, have a tendency to banding when over sharpened or too bright a gamma; resized images -- i.e. images not viewed 1:1 on a display -- have a tendency to banding ...

In other words, having us 'look' for banding can be an exercise in frustration -- some viewers can find banding in almost any image, and that can be laid at their own monitor or setting choices.

Cheers,
GB

 RGBaker's gear list:RGBaker's gear list
Sony Alpha NEX-7 Sony E 18-55mm F3.5-5.6 OSS Sony Carl Zeiss Sonnar T* E 24mm F1.8 ZA Sony E 50mm F1.8 OSS Sony E 55-210mm F4.5-6.3 OSS +1 more
Ed B
Ed B Veteran Member • Posts: 9,777
Re: Banding in sky

D Cox wrote:

This banding is not visible on my analog monitor (a good quality Iiyama).

I downloaded the "original" and made a quick A4 print. The banding doesn't show on the print either.

I wonder if this is a monitor problem. How many bits per channel are you sending to your monitor ? Banding basically occurs when a gradient between closely similar colours is not rendered with enough bits.

I'm using a high dollar IPS monitor and don't see any banding either.

Metering of the sky didn't work out very well but no banding. Just a shot that's improperly exposed.

The OP cranked in a +1 exposure compensation as compared to +0 for the second shot. That much compensation for an outdoor picture with sky showing will give some poor results.

Sorry for being so blunt but over exposure is the main problem here.

GaryW Veteran Member • Posts: 8,933
Re: Banding in sky

Ed B wrote:

D Cox wrote:

This banding is not visible on my analog monitor (a good quality Iiyama).

I downloaded the "original" and made a quick A4 print. The banding doesn't show on the print either.

I wonder if this is a monitor problem. How many bits per channel are you sending to your monitor ? Banding basically occurs when a gradient between closely similar colours is not rendered with enough bits.

I'm using a high dollar IPS monitor and don't see any banding either.

Yeah, this is a good point.  A lot of normal monitors are going to show banding in general.

Metering of the sky didn't work out very well but no banding. Just a shot that's improperly exposed.

The OP cranked in a +1 exposure compensation as compared to +0 for the second shot. That much compensation for an outdoor picture with sky showing will give some poor results.

Sorry for being so blunt but over exposure is the main problem here.

If you're shooting in JPEG and edit the exposure later, you are looking at loss of data when you look at the banding.  Sure, you can edit JPEGs, and if all I have is a JPEG I'll do so, but there is definitely a limit to the data.  Minor edits will not be noticeable, but 1 stop might be such as in this example.

-- hide signature --

Gary W.

 GaryW's gear list:GaryW's gear list
Sony E 16-50mm F3.5-5.6 PZ OSS Sony E PZ 18-105mm F4 G OSS Sony Alpha DSLR-A100 Sony Alpha NEX-5 Sony Alpha NEX-6 +9 more
captura Forum Pro • Posts: 26,980
Re: Banding in sky

"I wonder if this is a monitor problem."

I see no banding whatsoever on my monitor.

 captura's gear list:captura's gear list
Sony Alpha NEX-7 Canon EOS Rebel SL1 Sony a7 Olympus OM-D E-M10 Canon EF 85mm f/1.8 USM +3 more
OP CLYoung Regular Member • Posts: 204
Re: Banding in sky

Don't apologise for being blunt, you're quite right I overexposed the first one. I'd had +1 dialled in because I was shooting balloons against the sky and simply forgot to correct it for the first shot. At the time I clearly spotted my mistake and corrected it, but somehow I overlooked it when reviewing the pictures.

Still, it's been an interesting discussion, among other things I was prompted to clean my sensor for the first time (scary but it worked and I now feel better about changing lenses) and I am now shooting jpg+RAW.

 CLYoung's gear list:CLYoung's gear list
NEX5R Nikon D610 Nikon AF-S Nikkor 70-300mm f/4.5-5.6G VR Sony E 16mm F2.8 Pancake Nikon AF-S Nikkor 50mm f/1.8G +7 more
D Cox Forum Pro • Posts: 22,953
Re: Banding in sky

GaryW wrote:

Ed B wrote:

D Cox wrote:

This banding is not visible on my analog monitor (a good quality Iiyama).

I downloaded the "original" and made a quick A4 print. The banding doesn't show on the print either.

I wonder if this is a monitor problem. How many bits per channel are you sending to your monitor ? Banding basically occurs when a gradient between closely similar colours is not rendered with enough bits.

I'm using a high dollar IPS monitor and don't see any banding either.

Yeah, this is a good point.  A lot of normal monitors are going to show banding in general.

Metering of the sky didn't work out very well but no banding. Just a shot that's improperly exposed.

The OP cranked in a +1 exposure compensation as compared to +0 for the second shot. That much compensation for an outdoor picture with sky showing will give some poor results.

Sorry for being so blunt but over exposure is the main problem here.

If you're shooting in JPEG and edit the exposure later, you are looking at loss of data when you look at the banding.  Sure, you can edit JPEGs, and if all I have is a JPEG I'll do so, but there is definitely a limit to the data.  Minor edits will not be noticeable, but 1 stop might be such as in this example.

If you are going to edit exposure, white balance, levels, curves etc on a JPG, it is safest to convert the image to 16 bits per channel first. If you are worried about storage space, you can always convert back down to 8 bits after processing.

This particularly applies to shots that contain smooth gradients such as sky.

GaryR60 Contributing Member • Posts: 661
Re: Vertical huh

CLYoung wrote:

No, a 5R, but it's still quite easy to slip up. In this case though I think it had been deliberate, I just took a shot before remembering to change back.

I see.

OP CLYoung Regular Member • Posts: 204
Re: Banding in sky

Thanks, this discussion has been really helpful.

 CLYoung's gear list:CLYoung's gear list
NEX5R Nikon D610 Nikon AF-S Nikkor 70-300mm f/4.5-5.6G VR Sony E 16mm F2.8 Pancake Nikon AF-S Nikkor 50mm f/1.8G +7 more
Ed B
Ed B Veteran Member • Posts: 9,777
Re: Banding in sky

CLYoung wrote:

Don't apologise for being blunt, you're quite right I overexposed the first one. I'd had +1 dialled in because I was shooting balloons against the sky and simply forgot to correct it for the first shot. At the time I clearly spotted my mistake and corrected it, but somehow I overlooked it when reviewing the pictures.

Still, it's been an interesting discussion, among other things I was prompted to clean my sensor for the first time (scary but it worked and I now feel better about changing lenses) and I am now shooting jpg+RAW.

I can't count the times I've forgot to change settings, when using the 5n.

I'm use to cameras that have dials and more obvious visual indications of settings. 100% my fault and it always happens when I'm in a hurry.

OP CLYoung Regular Member • Posts: 204
Re: Banding in sky
1

Ed B wrote:

CLYoung wrote:

Don't apologise for being blunt, you're quite right I overexposed the first one. I'd had +1 dialled in because I was shooting balloons against the sky and simply forgot to correct it for the first shot. At the time I clearly spotted my mistake and corrected it, but somehow I overlooked it when reviewing the pictures.

Still, it's been an interesting discussion, among other things I was prompted to clean my sensor for the first time (scary but it worked and I now feel better about changing lenses) and I am now shooting jpg+RAW.

I can't count the times I've forgot to change settings, when using the 5n.

I'm use to cameras that have dials and more obvious visual indications of settings. 100% my fault and it always happens when I'm in a hurry.

Same here - I do this far less with my D80.

 CLYoung's gear list:CLYoung's gear list
NEX5R Nikon D610 Nikon AF-S Nikkor 70-300mm f/4.5-5.6G VR Sony E 16mm F2.8 Pancake Nikon AF-S Nikkor 50mm f/1.8G +7 more
Ed B
Ed B Veteran Member • Posts: 9,777
Re: Banding in sky

CLYoung wrote:

Ed B wrote:

CLYoung wrote:

Don't apologise for being blunt, you're quite right I overexposed the first one. I'd had +1 dialled in because I was shooting balloons against the sky and simply forgot to correct it for the first shot. At the time I clearly spotted my mistake and corrected it, but somehow I overlooked it when reviewing the pictures.

Still, it's been an interesting discussion, among other things I was prompted to clean my sensor for the first time (scary but it worked and I now feel better about changing lenses) and I am now shooting jpg+RAW.

I can't count the times I've forgot to change settings, when using the 5n.

I'm use to cameras that have dials and more obvious visual indications of settings. 100% my fault and it always happens when I'm in a hurry.

Same here - I do this far less with my D80.

Before I forget to tell you, I looked at your Flickr page and think you have some very good pictures posted.

OP CLYoung Regular Member • Posts: 204
Re: Banding in sky

Thanks.

 CLYoung's gear list:CLYoung's gear list
NEX5R Nikon D610 Nikon AF-S Nikkor 70-300mm f/4.5-5.6G VR Sony E 16mm F2.8 Pancake Nikon AF-S Nikkor 50mm f/1.8G +7 more
chuxter Forum Pro • Posts: 21,714
Re: Banding in sky

CLYoung wrote:

Thanks, this discussion has been really helpful.

First, I apologize for resurrecting this thread after 2 years. I recently looked on line for an image as an example for "banding" of JPEG images. Your shot popped out as a great example! I used it in a Power Point presentation on "Dynamic Range and Granularity" that I'm writing. I hope you don't mind if I use your images. I say images, because I went to the DPR thread to find out more and located your original image [before you cropped and darkened it]. I also used that image to show how to read a histogram. I didn't used your name or give a link the this thread.

I didn't read every entry in this thread, but I did read a LOT of them and came away thinking that you ended up w/ no clue about what really happened. That's not to say your helpers were not helpful, just that they, IMO, missed the important points. And I noticed that you came up w/ some superstitions about the cause.

I'm not a Sony guy, so I don't frequent this Forum. I do have an old R-1 , but don't use it much anymore.

If you want to see my "take" on your image, you can go here:

http://mavpca.org/photos/PhotographyLessons/PPTfiles/

and click on "DynamicRange.pptx.

There is also a PDF version at:

http://mavpca.org/photos/PhotographyLessons/PDFfiles/

For those disinclined to explore, I'll summarize:

Look at the histogram of the original image:

This is what you don’t want to see! Notice the big group of bright values jammed up on the right side! You blew out those bright highlights and then could not successfully recover them via editing.

Most cameras use the sRGB color space, which apples a "0.45 gamma curve" before saving a JPEG. This curve assigns more brightness values to the dark parts than to the light parts. This was done because humans don't notice large steps in bright parts of an image as much as in dark parts. Note that in your original image, nobody noticed the banding in the sky! But it was there.

Thus, when you took the 8-bit JPEG image encoded w/ a 0.45 gamma curve and tried to darken the image, those existing "bands" or large steps in brightness in the almost white sky became obvious to everyone. That is simply the way human vision works! Large brightness steps in a bright part of an image are ignored, but the same size steps in a darker part become very obvious.

BTW, there is a complimentary "2.2 gamma curve" used to decode JPEG files so that they can be displayed on a monitor. But this decoding doesn't undo the "damage" to the data; the bright parts will always have less granularity than the dark parts.

In every case that I have researched, banding happens when someone edits an overexposed JPEG image to make it darker. It never happens when they make it brighter.

As several of your helpers suggested, a better approach is to save RAW files. RAW files have greater bit-depth and there is no gamma curve applied; the "brightness curve" is almost linear. Thus, when you darken a RAW image, you don't normally get banding in the sky.

HTH...

 chuxter's gear list:chuxter's gear list
Nikon D810 Nikon D500 Tamron SP 90mm F2.8 Di VC USD 1:1 Macro (F004) Sigma 24-105mm F4 DG OS HSM Tamron SP 150-600mm F/5-6.3 Di VC USD +2 more
Keyboard shortcuts:
FForum MMy threads