PP Upgrade Conundrum

[CBImages]

I am quite certain the banding you are seeing is due to your monitor not being able to display the entire gamut of your image.

1) If you shoot in RAW, process in Lightroom you can then go on to make quite a few image adjustments in Photoshop without introducing banding in your image - assuming you use 16 bit versions of Photoshop. If, as you mention you use Elements then you can still make some adjustments but your scope is much more limited.

2) If you are shooting JPG's then there is much more image information available to you and banding will start appear in your images during fairly minimal adjustments.

Often image artefacts - banding and noise while noticable on your computer screen will not be evident once you print the image. This obviously is different for every image/monitor/printer set up and the more you experiment the more you will be able to recognise when banding will appear in your images.

Obviously to reduce the possibility of banding:
1) Shoot in RAW
2) Use 16 bit processing

If you do see any and you employ the above then it might be worth investing in a new computer screen.

[Stagecoach]

Thanks for the info Chris.


An update ................................................

Problem a) The occasional banding I get in CERTAIN SKIES that limits how far I can push them (if I want to).

Numerous tests tonight using an image where I reach limitations in pushing that sky area in 8 bit Elements are not overcome to any degree that I can measure visually by using 16 bit CC in any colour space. In other words changing to 16 bit working or even upgrading to the newer ACR or LR does not help something I considered it may.

This is not to say in any way that I'm suggesting 16 bit working is not an advantage, or upgrading, simply some things can not be pushed further.


Problem b) The banding I see in a 'Gradient' mask as per this example.

I found a Gradient test chart http://www.lagom.nl/lcd-test/gradient.php saved and opened it in PSCC and clearly see similar banding to what I have been experiencing.

I'll progress this one tomorrow, a swop of graphics cards (I have a spare) as a test, and I saw a nice new monitor last week

[ajohnw]

Are you by any chance producing aRGB jpg's Grahame? Maybe without an included colour profile? The profile might be something adobe leave out of elements.

I ask because I'm pretty certain when I have had banding appear when I adjust some one else's posted jpg they are aRGB jpg's. Or had curves applied that just can't be taken any further. That will happen based on the raw file at any bit depth at some point.

I just did 3 gradients out of curiosity. These might not contain colour profiles but are sRGB.

All enhancements off



Dithering on



Dithering on and adaptive sub sampling, which I suspect adds nothing unless there is an image underneath it. Not sure




I sometimes feel I can see steps in these but suspect it's an optical illusion / looking too hard.

This is the grey scale response of my monitor which I believe meets Enzio's more critical spec easily.



If nothing else people can compare the gradients with yours.

John
-

[ajohnw]

This is a GIMP tell no lies historgam of the image originally posted



Against that this is the same for the all effects off one I posted.



Looks a little dithered to me but no where near as much as the other one appears to be. It has to dither a little due to the format limitations. It's 8 bit so can only show that resolution.

John
-

[ajohnw]

Just to show the other problem stretching past the bit depth capability - what ever it is. This i the histogram of one that doesn't go from black to white.



The top is smoother because less tonal range is covered.

If I then stretch this to near black to near white this is what happens and poor old software has no choice as there is nothing for it to fill the gaps with. I do have a package with a debanding slider but doubt if it would cope with this.



Banding can be seen in that.

John
-

[Mike Buckley]

Grahame,

I have occasionally but very rarely experienced this issue and have attributed it to working in the sRGB space. I haven't changed my workflow to test that theory because I have so rarely experienced the issue and have usually (not always) been able to solve it by applying either gaussian blur or noise reduction.

[Mike Buckley]

Grahame,

I have occasionally but very rarely experienced this issue and have attributed it to working in the sRGB space. I haven't changed my workflow to test that theory because I have so rarely experienced the issue and have usually (not always) been able to solve it by applying either gaussian blur or noise reduction.

Best of luck coming to an acceptable compromise if not a solution.

[xpatUSA]

Are you by any chance producing aRGB jpg's Grahame? Maybe without an included colour profile? The profile might be something adobe leave out of elements.
John
-

I can confirm that Elements 6 embeds profiles into aRGB and sRGB edited images. It also embeds ProPhoto but only if the image had it when opened. Any other profile gets stripped during the space conversion and is replaced by the only two choices for conversion.

As we all know, aRGB 'stretches' some of those 255 levels, compared to sRGB but not a lot in the blue (see any of the one kajillion CIE gamut diagrams out there).

[Brownbear]

Hi Grahame,

I've seen banding in a few of my images, typically images that have a lot of bright blue sky or skies with mist and fog, and typically when I've post processed an image in ACR Lightroom, and then opened in Photoshop CC and tried to apply a curves adjustment say to lower the highlights in a sky.

Until most recently all the images I have been posting were in the colour space Adobe RGB. (John, you were indeed correct) It is something that I manage to fix just a month or two ago, although I'm not sure what I did differently this time around - I just know that my colour space for posting here is finally correct.

When I read your post I revisited an old image that I have shared here before that had banding in the clouds/sky... I tried post processing the image again. First in Lightroom but when I opened the image in Adobe Photoshop CC to sharpen and apply a curves adjustment the highlights in the red channel were clipped. (clipping not seen in LR and I've experienced this before)

So I post processed the image again, this time using CC Adobe Photoshop, opened the raw file, applied a curves adjustment to the image and sharpened. Absolutely no banding seen in the image. I'm sharing because perhaps it is due to the colour space.

I've never used a gradient mask before but I applied one (in Photoshop CC) and it made my image black and white. Again absolutely no banding seen at full size.

As I can't explain "why" the difference with this particular image, I thought it might be helpful if I shared simply because you could try the same and see if that this is the cause?

I can see the banding in your image and my monitor is just a regular one.

[xpatUSA]

Problem b) The banding I see in a 'Gradient' mask as per this example.

I found a Gradient test chart http://www.lagom.nl/lcd-test/gradient.php saved and opened it in PSCC and clearly see similar banding to what I have been experiencing.

FYI, I downloaded that .png image, opened it in ImageJ, took a screen shot and saved it as a PNG (no JPEG obfuscation):



As expected, perfectly even steps and banding, if any exists, is very hard to see.

Now, this is an 8-bpc PNG image. How CiC and your system shows it will likely differ from mine.

Not suggesting a fix (too many variables) but hope this info helps . .

[ajohnw]

Gimp shows that one as dead flat in it's histogram. RT shows each of the steps.

John
-

[Stagecoach]

Are you by any chance producing aRGB jpg's Grahame? Maybe without an included colour profile? The profile might be something adobe leave out of elements.

I ask because I'm pretty certain when I have had banding appear when I adjust some one else's posted jpg they are aRGB jpg's. Or had curves applied that just can't be taken any further. That will happen based on the raw file at any bit depth at some point.

John
-

John, wrt to 'banding' of certain sky type areas these are always from my own NEFs (RAW) well exposed files.

The three B&W gradients you have posted I can clearly see banding whether viewed through my browser or with the file opened in PSCC or any of my other editors.

This is a GIMP tell no lies historgam of the image originally posted

Against that this is the same for the all effects off one I posted.

Looks a little dithered to me but no where near as much as the other one appears to be. It has to dither a little due to the format limitations. It's 8 bit so can only show that resolution.

John
-

Those two tell no lies histograms clearly show a difference.

Just to show the other problem stretching past the bit depth capability - what ever it is. This i the histogram of one that doesn't go from black to white.

The top is smoother because less tonal range is covered.

If I then stretch this to near black to near white this is what happens and poor old software has no choice as there is nothing for it to fill the gaps with. I do have a package with a debanding slider but doubt if it would cope with this.

Banding can be seen in that.

John
-

I have seen this happen before when going to extremes but it is not evident in the histograms of the sky areas I have been investigating.

Grahame,

I have occasionally but very rarely experienced this issue and have attributed it to working in the sRGB space. I haven't changed my workflow to test that theory because I have so rarely experienced the issue and have usually (not always) been able to solve it by applying either gaussian blur or noise reduction.

Best of luck coming to an acceptable compromise if not a solution.

It's the occasional ones Mike that are always the worst to confirm the reason for.

Hi Grahame,

I've seen banding in a few of my images, typically images that have a lot of bright blue sky or skies with mist and fog, and typically when I've post processed an image in ACR Lightroom, and then opened in Photoshop CC and tried to apply a curves adjustment say to lower the highlights in a sky.

Until most recently all the images I have been posting were in the colour space Adobe RGB. (John, you were indeed correct) It is something that I manage to fix just a month or two ago, although I'm not sure what I did differently this time around - I just know that my colour space for posting here is finally correct.

When I read your post I revisited an old image that I have shared here before that had banding in the clouds/sky... I tried post processing the image again. First in Lightroom but when I opened the image in Adobe Photoshop CC to sharpen and apply a curves adjustment the highlights in the red channel were clipped. (clipping not seen in LR and I've experienced this before)

So I post processed the image again, this time using CC Adobe Photoshop, opened the raw file, applied a curves adjustment to the image and sharpened. Absolutely no banding seen in the image. I'm sharing because perhaps it is due to the colour space.

Hi Christina

The clipping difference is quite noticeable in using different packages and workflows I have found but is to be expected and I have always addressed it by not pushing the histogram fully to the right at the initial stages of PP. All PP is done so that at the very last stage there is scope to move the histogram to full right if wanted.

I try to do my 'pushing' if required in ACR or Elements 16bit but have been limited when it comes to such areas as blend modes where only 8bit was available. With my limited experience of PSCC (almost two days now) I see that not only do I have a wealth of additional tools in ACR but my much loved gradients are available to be used at that stage, fantastic !

I've never used a gradient mask before but I applied one (in Photoshop CC) and it made my image black and white. Again absolutely no banding seen at full size.

As I can't explain "why" the difference with this particular image, I thought it might be helpful if I shared simply because you could try the same and see if that this is the cause?

I can see the banding in your image and my monitor is just a regular one.

Yes it is very helpful that you shared this You produced a black and white gradient in PSCC of which you say you saw absolutely no banding, I can't

FYI, I downloaded that .png image, opened it in ImageJ, took a screen shot and saved it as a PNG (no JPEG obfuscation):



As expected, perfectly even steps and banding, if any exists, is very hard to see.

Now, this is an 8-bpc PNG image. How CiC and your system shows it will likely differ from mine.

Not suggesting a fix (too many variables) but hope this info helps . .

Good one Ted. What you have confirmed is that the test image is an evenly stepped image of which I would hope to see on my monitor when open in PSCC (or any other good software package) as either a smooth gradient or evenly stepped.

What I see is a randomly unevenly stepped image.


Grahame

[Stagecoach]

Well I have not managed to progress the 'gradient uneven stepping' much due to time, taking pictures of biros and playing with CC.

What I have come across is a setting on my graphics card which gives me two choices and I am unsure of which one to select;

For the option Choose How Colour Is Set

a) other applications control colour settings
b) use NVIDIA settings



It was set on 'other settings' but I'm unsure of where I should have this set.

I'm not suggesting this is the answer to my uneven gradient problem but it's something I'm unsure of and worth getting right.

Grahame

[ajohnw]

Pity about the biro's thread. I was hoping to see lots from different gear. The thread has gone awol now.

One thing you haven't told us Grahame is if say the web site test gradient or any of the ones I have posted look to be banded.

The histogram I posted of yours - 1st post you made - will show bands irrespective of card if it's reasonably linear but that is some what dependent on the monitors grey scale behaviour. I can see the bands but they aren't what I would call strong but I feel would be more noticeable in an image.

Later I posted histograms of what happens if a grey scale is stretched past the available bit depth underneath. That showed banding with bells on. That will happen eventually what ever the bit depth of the original raw file or for that matter jpg is.

On the post you have just made my Nvidia driver allows several things to be changed which I leave all at 1 = do nothing and the calibration corrects things. Brightness and contrast setting and rgb balance are set with the monitor controls. In your case Other Applications Control Settings leaves things as they should be, corrections made to the monitor settings and the calibration will then correct the rest as well as it can. As set calibration will try and correct anything that the other settings cause which interfere with an ideal response. Might be good or bad. Impossible to say.

If you want to see the actual response of your monitor install this and plug in your calibrator and select current setting (default when it powers up) and then tools - verify calibration. A window pops up allowing various test charts to be selected. The default is suitable.

http://dispcalgui.hoech.net/

You will also need to install this and tell dispcal where it is

http://argyllcms.com/

You can also use this for calibration but might not want to. It currently doesn't support V4 ICC files. If that ever gets to be a problem I'd guess it will. Must admit I have only used it for verification against a profile it has installed but it should work as I have indicated.

Linux comes rather well equipped for judging monitors visually. Normally these images are separate so when viewing dark grey it may be best to cover the others up. They are also extreme particularly the botton of dark grey and the top end of light grey. All open source and down to KDE.



Might help. Top one should be easy. Steps should look even but dark and light ends of the others are likely to fade out in some cases. Mid grey is rather important.

John
-

[Stagecoach]

John,

Thanks for the below;

Pity about the biro's thread. I was hoping to see lots from different gear. The thread has gone awol now.

I agree, it's a pity.

One thing you haven't told us Grahame is if say the web site test gradient or any of the ones I have posted look to be banded.

Thought I had covered that in post 32 but yes they do appear banded to me but not too bad.

The histogram I posted of yours - 1st post you made - will show bands irrespective of card if it's reasonably linear but that is some what dependent on the monitors grey scale behaviour. I can see the bands but they aren't what I would call strong but I feel would be more noticeable in an image.

I suppose the problem here is what one calls 'noticeable'

Later I posted histograms of what happens if a grey scale is stretched past the available bit depth underneath. That showed banding with bells on. That will happen eventually what ever the bit depth of the original raw file or for that matter jpg is.

I think I'm clear on this one but with respect to simply producing a plain image or mask that is filled with white and dragging a gradient across it, this would not be stretching things?

On the post you have just made my Nvidia driver allows several things to be changed which I leave all at 1 = do nothing and the calibration corrects things. Brightness and contrast setting and rgb balance are set with the monitor controls. In your case Other Applications Control Settings leaves things as they should be, corrections made to the monitor settings and the calibration will then correct the rest as well as it can. As set calibration will try and correct anything that the other settings cause which interfere with an ideal response. Might be good or bad. Impossible to say.

So I will leave set to a) other applications control colour settings.

If you want to see the actual response of your monitor install this and plug in your calibrator and select current setting (default when it powers up) and then tools - verify calibration. A window pops up allowing various test charts to be selected. The default is suitable.

http://dispcalgui.hoech.net/

You will also need to install this and tell dispcal where it is

http://argyllcms.com/

You can also use this for calibration but might not want to. It currently doesn't support V4 ICC files. If that ever gets to be a problem I'd guess it will. Must admit I have only used it for verification against a profile it has installed but it should work as I have indicated.

Linux comes rather well equipped for judging monitors visually. Normally these images are separate so when viewing dark grey it may be best to cover the others up. They are also extreme particularly the botton of dark grey and the top end of light grey. All open source and down to KDE.



Might help. Top one should be easy. Steps should look even but dark and light ends of the others are likely to fade out in some cases. Mid grey is rather important.

John
-

I do not use a calibrator John, and if I wanted to can not get one here. I have always relied upon a visual calibration simply comparing against the charts available similar to the one you have posted above. Yes I know some will jump up and down in amazement but to me the ultimate accuracy of the presentation of what I see on screen is not a priority.

For info, I can see all steps in each section of that chart, BUT, this is dependent upon whether they are placed at the top or bottom of the monitor and my viewing angle

Grahame

[IzzieK]

Grahame, we have the same settings on the Nvidia card and I chose the same as yours then press the apply button at the bottom of the page so it keeps using it but then I have 2 monitors so that my tools in Photoshop goes to the other monitor. I used to attached an old, old monitor with my main one and it still works with my system. Now that monitors here are cheaper, it is easier to buy a bigger one to serve as a dumping station for instructions on the other side while using the main one for PP. Nvidia is very good graphics card.

[Stagecoach]

Grahame, we have the same settings on the Nvidia card and I chose the same as yours then press the apply button at the bottom of the page so it keeps using it but then I have 2 monitors so that my tools in Photoshop goes to the other monitor. I used to attached an old, old monitor with my main one and it still works with my system. Now that monitors here are cheaper, it is easier to buy a bigger one to serve as a dumping station for instructions on the other side while using the main one for PP. Nvidia is very good graphics card.

Hi Izzie,

Bought a new monitor today and have been busy setting that up and seem to be digging myself a deeper hole

I believe it best to set the NVIDIA settings to OFF and use the first choice 'Other applications control colour settings'.

My theory being that otherwise you have two sets of controls that can adjust your monitor, the NVIDIA adjustments AND the monitor (built in) adjustments.

I have also got another glitch in that when opening PSCC (or Elements) now they show an error message that the monitor display driver has an error and on continuing and opening a New blank file that would normally be white its beige

I HATE COMPUTERS AND AM PACKING UP FOR THE DAY !!!!

Grahame

[ajohnw]

Pc's are a pain Grahame - when they don't do what is wanted. Sorry about comments about the web link gradient - did read but forgot.

Sort of starting again.

It doesn't surprise me that the gradient of your own shows banding as gaps can be seen in the histogram. As some one else pointed out that might be down to jpg quality level. I always save at 95%. On the other hand I recently posted some shots at 65% as they were a bit big. Can't say that I noticed anything but didn't look that hard. Seems 65% is not an unusual value for posting if some one wants to keep bandwidth down from web sites. So problems with this one aren't down to the monitor or display gear. More how it was generated. The ones I posted were saved at 95%. Maybe they are better? The histogram does show some ripple though but no gaps.

The one in the web link is a png and I assume has zero compression. No banding should be visible but if some one stares at something long enough ???? The problem is that viewing angles, glasses, lighting and etc might cause some odd effects. It might pay to only look at things centre screen with it square on to your line of sight. That is needed on some panel types and people who manipulate colour all day reckon the limit is still only a few degrees even with the best available. I reckon that is a bit extreme but I don't do that sort of thing all day n days a week.

There are likely to be some problems with the test bars I posted due to backlighting. These are png and the bars themselves are probably compressed. Max seems to be 9%. Problems will show up at the dark ends. On the rgb bars for instance I might be able to see a tinge of colour in the last squares and the ones before aren't exactly distinct. particularly on red. I have tried to use these to set up colour manually. Not something I would recommend.. The separate grey scales are more useful. No square centre of mid grey. Bright end of light grey barely visible, last one is near hard to see. That is the last 2. Several of the dark grey at the dark end might not be visible. In my case that more or less is 3 of them. Room lighting does have an effect on that. The Y looks more like a gradient than steps at the dark end. Yellow is a tough colour for cameras too. C and M are ok.

If I was on my machine when you mentioned getting another monitor I would have posted this link. My machine might be on line all day but I'm not.

http://www.tftcentral.co.uk/articles/icc_profiles.htm

This page contains calibration profiles and settings. The reviews are also the only ones I am aware of that are really suitable for choosing a monitor. Another review site also posts profiles on there but often the monitor settings are missing - no use at all. The Tftcentral ones are to be preferred really. Using these is a decent way of avoiding the need for a colorimeter. Some one on here did a few months ago and was rather pleased with the results. Others have been too. The other way is to buy a pre calibrated display. I'd guess downloading one of the icc files and using the settings provided will give better results really especially as these days the displays come way too bright to make the dynamic range look better. The brightness and contrast settings they give will sort that out and that is the aspect that will give better grey scales. I'm running at 125 cm/m^2 at the moment but intend to go back to 150 as it will improve the dark end a little. I tried it as some use 100 but I'd guess that goes back to crt's.

I reckon you could get a colorimeter if you wanted. The answer is Ebay and or Paypal. I'd suggest you sign up to the nearest real Ebay. Not sure if the Fiji one is. There should be a link to all of their sites some where - in the UK it's currently at the bottom of the Today page. I buy items from all over the place at times even just using Paypal sometimes. More and more shops offer that option directly now as it simplifies currency conversion. There are UK ebay members that don't live here so I'd guess that applies to all of their sites. Only way to find out is to try. I checked shipping costs to Fiji from the UK via Royal Mail, value £250 weight 999gm and it comes out at £20.20, signed for and takes 5-7 working days. At 2kg it goes up to around £30 and past that they don't want to know. It seems to go up as the weight increases. Past that other carriers have to be used and the price rockets. Much the same is true of the USA. If sellers use USPS they offer 2 services. Global priority mail is surprisingly cheap. Expedited more expensive and quicker. FEDEX etc insane. The catch sometimes with the USA is that some sellers charge for filling in a form as well as the shipping cost or use one of the standard boxes. Many though are extremely helpful. Chinese sellers will generally ship anywhere. One aspect to watch is import duty. A real hit in the uk but the real downer is what they charge for collecting it. Chinese sellers often don't state value at all or seriously understate it. Japanese sellers can also be helpful but their shipping costs are higher. Most people have very few problems using ebay. Problems are usually solved by sending the item back which may mean a dispute and probably forking out for return shipping costs. Amazon though provide a return postal label and collection - pass on Fiji in that respect. Only way to find out is to sign up to one and find out. They tell people shipping costs eventually before the item is bought but keep it till last. People I know in the USA have also bought things for me and shipped them on. Some sellers/shops wont export full stop. Paypal using the fee paid option is an easy way of arranging this sort of thing. I once sold a rather heavy refracting telescope to some one far away. His family collected it and sent it on to him. If you buy from overseas the main thing is to ask about shipping costs first even if there is no indication that people will ship to you. A countries domestic shipping outfit USPS, Royal Mail etc are usually the cheapest option. I suspect this is because they have an arrangement which means they don't pay for delivery at the other end.

John
-

[Mike Buckley]

an error message that the monitor display driver has an error

Even though your monitor is new, it's possible that an updated driver is required. It's also possible that the driver has become corrupted.

[xpatUSA]

Grahame,

Here's shipping costs from USA to Fiji based on medium-small package:

Priority Mail Medium Flat-rate Box $61.75
Priority Mail Medium Small-rate Box $24.75
International First-class Parcels up to 8lb $12.75

Info as of Jan 26, 2014.

I would be happy to buy an item and ship it to you, should you feel the need . . .

. . . but in June I shipped a camera sensor to a friend in Oz (first-class parcel) and it never left the USA. Came back from Chicago in September going on about incorrect address format, glurk. Priority mail would be my choice anyway.