Pages: [1]   Go Down

Author Topic: Banding /Posterization in sky's  (Read 17377 times)

shaunw

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 396
    • Shaun Walby Landscape Photography collections
Banding /Posterization in sky's
« on: November 21, 2011, 10:56:11 am »

Iam having the above ruin some of my more recent images. I understand there can be a few causes of the banding....working in 8 bit as opposed to 16 bit incompatible work/colour space flow....i use adobe 98 in camera...prohoto in RAW/photoshop and convert to sRGB for web...also you have to have a have a half decent monitor which i think mine is.


It seems to occure wherever the is little detail present like the sky of long exposure water shots?

anyone help with this very annoying problem?

Regards
Shaun W
Logged
Canon 5D mk II Sigma 10-20, Canon 17-40mm L, Canon 24-105mm L, Canon 70-200 L, Lee Filters, Manfrotto geared head/tripod.

''Twelve significant photographs in any one year is a good crop''. – Ansel Adams
http://www.shaunwalbyphotography.com

degrub

  • Guest
Re: Banding /Posterization in sky's
« Reply #1 on: November 21, 2011, 11:05:08 am »

an example  or two might help us .....
Are you making any large curve moves in PS ?
When is the first time in your workflow that you see the banding ?

Frank
Logged

shaunw

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 396
    • Shaun Walby Landscape Photography collections
Re: Banding /Posterization in sky's
« Reply #2 on: November 21, 2011, 02:19:09 pm »

Hi Frank

I notice it very slightly in the canon CR2 file it gets worse when i convert to DNG to open in RAW and worse still when i open into photoshop CS4...

possible causes

1. ive recently re-calibrated my monitor...could i have made an error somewhere?
2. switched off noise reduction in camera re long re-write times...doubt its this?

other than those nothing new to my work flow


ill post one of the worst from a long expo night shoot


cheers shaun
Logged
Canon 5D mk II Sigma 10-20, Canon 17-40mm L, Canon 24-105mm L, Canon 70-200 L, Lee Filters, Manfrotto geared head/tripod.

''Twelve significant photographs in any one year is a good crop''. – Ansel Adams
http://www.shaunwalbyphotography.com

shaunw

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 396
    • Shaun Walby Landscape Photography collections
Re: Banding /Posterization in sky's
« Reply #3 on: November 21, 2011, 02:22:00 pm »



banding /posterization horror show
Logged
Canon 5D mk II Sigma 10-20, Canon 17-40mm L, Canon 24-105mm L, Canon 70-200 L, Lee Filters, Manfrotto geared head/tripod.

''Twelve significant photographs in any one year is a good crop''. – Ansel Adams
http://www.shaunwalbyphotography.com

degrub

  • Guest
Re: Banding /Posterization in sky's
« Reply #4 on: November 21, 2011, 03:41:42 pm »

"switched off noise reduction in camera re long re-write times"

maybe since you see it in the raw /jpg straight from the camera.
Logged

Sussex Landscapes

  • Newbie
  • *
  • Offline Offline
  • Posts: 28
    • Sussex Landscape Photography
Re: Banding /Posterization in sky's
« Reply #5 on: November 22, 2011, 11:37:58 am »

i have had the same issue with a few of my shots with extremes, however i found that by zooming in to the right %, ie 25% /50% or 100% made this banding go away. 33% and other zoom ratios amplify this.

if thats not it, well your guess is as good as mine.

simon
Logged

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4388
    • Pieter Kers
Re: Banding /Posterization in sky's
« Reply #6 on: November 22, 2011, 01:08:07 pm »

Hi Frank

I notice it very slightly in the canon CR2 file it gets worse when i convert to DNG to open in RAW and worse still when i open into photoshop CS4...

possible causes

1. ive recently re-calibrated my monitor...could i have made an error somewhere?
2. switched off noise reduction in camera re long re-write times...doubt its this?

other than those nothing new to my work flow


ill post one of the worst from a long expo night shoot


cheers shaun



1 If you can see banding at 100% in Photoshop it is probably real and not a monitor flaw
2 no-

-

Your problem is a basic digital problem. Gradual dark skies are difficult to get without banding. The image is built up with coarser steps in the dark regions- thats is why they say expose to the right..

I would say-

go back tot raw and change as little as possible to the original and save it as a 16 bit RGB tiff.
See if it has banding in photoshop at 100%.
if it already does show banding go back to the raw converter and work in sRGB -and do the same.
If that does not help find another raw converter.

PS
it could also well be the black and white transition that is the problem- then change that process- the black and white layer in photoshop often introduces banding
PS2- even if your photo does not have banding ( anymore) it might turn op with the print - caused by the printer)

« Last Edit: November 22, 2011, 01:10:33 pm by kers »
Logged
Pieter Kers
www.beeld.nu/la

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2005
    • http://www.guillermoluijk.com
Re: Banding /Posterization in sky's
« Reply #7 on: November 22, 2011, 03:52:44 pm »

Most of the visible jumps are of just 1 level (for example from gray (120,120,120) to gray (121,121,121), so they cannot be made softer. In absence of noise (which is the case) this means visible posterization, and it cannot be avoided.

This is a good example of the 'more levels' falacy many people insist when talking about ETTR.

ETTR means better SNR, i.e. less visible noise. If a plain area of a scene (e.g. your sky) is sufficiently clean of noise because of ETTR, once converted to an 8-bit output (e.g. JPEG) the bands will be visible like in this case. The RAW file had a lot of levels thanks to ETTR, but far from preventing posterization, the low noise achieved through ETTR produced banding on the 8-bit output.

Contrarily to what many times is said, posterization usually comes from insuficient levels on well exposed areas in 8-bit, never from insufficient levels in the RAW file since noise always dithers captured information, no matter the exposure achieved.

A possible solution is to add noise in the sky to make the bands invisible.

Regards

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: Banding /Posterization in sky's
« Reply #8 on: November 22, 2011, 06:41:57 pm »

Most of the visible jumps are of just 1 level (for example from gray (120,120,120) to gray (121,121,121), so they cannot be made softer. In absence of noise (which is the case) this means visible posterization, and it cannot be avoided.

This is a good example of the 'more levels' falacy many people insist when talking about ETTR.

ETTR means better SNR, i.e. less visible noise. If a plain area of a scene (e.g. your sky) is sufficiently clean of noise because of ETTR, once converted to an 8-bit output (e.g. JPEG) the bands will be visible like in this case. The RAW file had a lot of levels thanks to ETTR, but far from preventing posterization, the low noise achieved through ETTR produced banding on the 8-bit output.

Contrarily to what many times is said, posterization usually comes from insuficient levels on well exposed areas in 8-bit, never from insufficient levels in the RAW file since noise always dithers captured information, no matter the exposure achieved.

A possible solution is to add noise in the sky to make the bands invisible.

I agree with Guillermo's analysis. In some areas, the banding seems to take an abrupt jump rather than occurring in a more uniform pattern. I looked at one such area with ImageJ (a freeware program--do a Google search if you are interested), and plotted the pixel profile along a vertical line as shown. One stretch of pixels have a value of 58 and then jumps to 60. The L* for an 8 bit pixel value with a gamma of 2.2 is 24.15 for the 8 bit pixel value of 60 and 23.66 for the value of 59, a change of 2% in luminance and 4% for a 2 pixel value change (58-60).

According to the Weber-Fechner law (see Norman Koren), a 1% difference in luminance is visible to the eye at higher luminances and the eye may be less sensitive at lower levels.

Working in 16 bit might help if your printer can work in that bit depth. I understand that 16 bit printer output is available only on a few printers and only on the Mac at present, since the PC offers only an 8 bit path to the printer. Most video output is also only 8 bits per channel (less if you perform adjustments in the video card LUT), so you would have to examine for banding in the print rather than relying on the monitor.

As Guillermo pointed out, banding would be dithered by noise, so ETTR wouldn't help here, since it would reduce noise.

By the way, a very interesting image. I like the mood.

Regards,

Bill

Logged

EgillBjarki

  • Guest
Re: Banding /Posterization in sky's
« Reply #9 on: November 23, 2011, 03:57:25 am »

I had problems with this as well on my old camera setup. My solution was to ADD NOISE, you can give that a try. It works better on a larger file, here is what I could do with this file.

Logged

EduPerez

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 700
    • Edu Pérez
Re: Banding /Posterization in sky's
« Reply #10 on: November 23, 2011, 05:20:51 am »

Could it be that this posterization is caused during the conversion from the 16bits of the RAW file
  • to the 8bits of the JPEG file?

Instead of adding noise, shouldn't the program that does the conversion add some dithering?

  • I know most RAWs are only 14 or even bits, and even PS only uses 15 bits, but you get the point.
Logged

Graystar

  • Guest
Re: Banding /Posterization in sky's
« Reply #11 on: November 23, 2011, 05:32:21 am »

I would suggest checking the 16-bit values to see if the posterization is in the data or simply due to display limitations.

In Photoshop, in the Info Palette you should be able to click the eyedropper and select 16bpc.  Zoom in to 100% to make the bands large and move the pointer from one transition to another while watching the 16-bit RGB values.  If the values stay nearly the same until you reach a transition point then it’s the image data that’s faulty.  However, if the values increase smoothly then the problem is the display, not the data.  In that case, while the image looks bad on the screen, it should print nicely if you have your 16-bit printing worked out.
Logged

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2005
    • http://www.guillermoluijk.com
Re: Banding /Posterization in sky's
« Reply #12 on: November 23, 2011, 12:45:25 pm »

As long as the 8-bit data changes mainly in steps of 1 level (this is the maximum softness for a gradation), we don't need to look for the reasons of posterization in the RAW file, in the post processing nor in the 16 to 8 bit conversion.

It is a fact that if you can see a gradation with the maximum softness, we are just being victims of the limitations intrinsic to the 8-bit display. So there was nothing wrong in the process, and the only solution is to add noise to dither the bands.
« Last Edit: November 23, 2011, 12:48:36 pm by Guillermo Luijk »
Logged

Graystar

  • Guest
Re: Banding /Posterization in sky's
« Reply #13 on: November 23, 2011, 01:24:27 pm »

But if the bands are simply a result of 8-bit rendering, then you don't need to add noise for 16-bit printing.  You only need the noise if a computer monitor is the intended form of display.  Even then, a 10-bit system may display the image without banding.  If the posterization is in the data, then the banding will always appear regardless of how the image is displayed.  That's what checking the data will tell you.

If the data is good and you intend to print the image, then there's no need to add noise.
Logged

shaunw

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 396
    • Shaun Walby Landscape Photography collections
Re: Banding /Posterization in sky's
« Reply #14 on: November 23, 2011, 02:30:32 pm »

Dear All


Thanks for the input thus far...iam working my way through all the suggestions.

Regards Shaun
Logged
Canon 5D mk II Sigma 10-20, Canon 17-40mm L, Canon 24-105mm L, Canon 70-200 L, Lee Filters, Manfrotto geared head/tripod.

''Twelve significant photographs in any one year is a good crop''. – Ansel Adams
http://www.shaunwalbyphotography.com

Lost

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 71
    • Flickr snapshots
Re: Banding /Posterization in sky's
« Reply #15 on: November 23, 2011, 02:38:30 pm »

It is a fact that if you can see a gradation with the maximum softness, we are just being victims of the limitations intrinsic to the 8-bit display. So there was nothing wrong in the process, and the only solution is to add noise to dither the bands.

I fully support this. I have seen similar effects when fixing flare problems in sky areas using Photoshop Elements. This forces the image to 8 bit, and usually results in banding artefacts - particularly if trying to use a filter to smooth out the corrected area. As noted, the only 'fix' I have found for this is to add noise, to help conceal the banding.

As an aside, I am considering upgrading to full Photoshop to avoid this. The 8 bit limitations of PSE are frustrating - though the cost of upgrading simply for 16 bit is somewhat unpalatable. The alternative is to stick with Lightroom, which doesn't give artefacts, but which suffers from huge catalogue files if you do a lot of cloning/brushing (=> problems for backup etc).

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: Banding /Posterization in sky's
« Reply #16 on: November 24, 2011, 04:55:24 am »

This is where the, often overlooked, relationship between bit depth and image resolution comes into play...

If you have a 10x10px image sampled on 2 bits (4 possible values), you will have banding in many cases. if you code the same image on 4 bits (16 possible values), then the chance of being able to manage gradients in such a way that no groups of adjacent pixels share the same value is greatly increased.

Now there will always be some gradients that will generate banding if no noise is present, but the frequency of occurence will depend on the relaitionship between resolution and bit depth. In other words, the higher the resolution, the higher the chance of having some form of banding.

This is in fact a major limiting factor to the increase of resolution in digital images, regardless of the size of the sensor.

Now, these issues are of couse often only visible on screen because of the projection on an 8 bits space. Occurences in prints are reduced by the usage of a 16bits pipe and also by the fact that printers essentially model gradients with noise patterns.

Cheers,
Bernard
« Last Edit: November 24, 2011, 04:57:07 am by BernardLanguillier »
Logged

RazorTM

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 54
    • http://www.flickr.com/photos/razortm/
Re: Banding /Posterization in sky's
« Reply #17 on: November 24, 2011, 09:53:49 am »

I had problems with this as well on my old camera setup. My solution was to ADD NOISE, you can give that a try. It works better on a larger file, here is what I could do with this file.



Clever solution.  I never would have thought of that!
Logged
Pages: [1]   Go Up