Basic Banding/Posterization Question

[Arthur Morris]

When photographing bright sunrise and sunrise images with the sun in or close to the edge of the frame, what causes the banding/posterization issues in the RAW files with pretty much all camera bodies? Is the any way to avoid it? To deal with it?

[Roger Clark]

When photographing bright sunrise and sunrise images with the sun in or close to the edge of the frame, what causes the banding/posterization issues in the RAW files with pretty much all camera bodies? Is the any way to avoid it? To deal with it?

Hi Artie,

Yes, I have seen this too and it can be quite annoying as some of my favorite images are sunrises and sunsets with the sun in the frame. Because digital cameras are linear off the sensor, such posterization should not be there, especially considering the limits of photon noise. I believe it is most likely due to the raw converters with their integer conversions, especially if they apply the characteristic curve using integers. Next time I encounter this effect, I'll try a conversion using 64-bit floating point and see if it makes a difference which would enforce the idea that it is integer roundoff.

The way I mitigate the posterization is to first convert to 16-bit tiff then apply some blur on the posterized area.

Roger

[Don Thompson]

Could one of you post examples for those of us who don't know what you are referring to. Thanks

[Jon Rista]

Hi, Don. Take a look at this entry from Art's blog: So Simple, So Sweet. I think the photos there demonstrate the problem rather well.

@Art: I would be curious to know how compressed those JPEGs were. I've experienced similar things myself, and I always thought it was due to the amount of JPEG compression. In the past, if I encountered banding, blocking and posterization, reducing the amount of compression (or eliminating it entirely...JPEG compression scale 100 as it were) seemed to remedy the problem. You obviously end up with larger, possibly much larger image files...but the posterization either goes away entirely, or is greatly mitigated. A number of years ago, I found this helpful article, which might provide some useful insight: Jpeg Quality.

If your JPEG images are already at maximum quality, I have no further insight to offer. Hopefully Roger or another insightful mind can help. :)

[Arthur Morris]

Jon, I have seen the banding in my RAW files as mentioned in Pane 1 :)

[Arthur Morris]

Hi, Don. Take a look at this entry from Art's blog: So Simple, So Sweet. I think the photos there demonstrate the problem rather well.

@Art: I would be curious to know how compressed those JPEGs were. I've experienced similar things myself, and I always thought it was due to the amount of JPEG compression. In the past, if I encountered banding, blocking and posterization, reducing the amount of compression (or eliminating it entirely...JPEG compression scale 100 as it were) seemed to remedy the problem. You obviously end up with larger, possibly much larger image files...but the posterization either goes away entirely, or is greatly mitigated. A number of years ago, I found this helpful article, which might provide some useful insight: Jpeg Quality.

If your JPEG images are already at maximum quality, I have no further insight to offer. Hopefully Roger or another insightful mind can help. :)

Sorry again Jon. In the skimmer image there was no banding at all in the RAW file. That one was a result of JPEG compression or something else. I will try to find a file that illustrates my point and post it here soon.

[Jon Rista]

Ok. I wasn't sure if you were actually seeing it in the raw, or only in the output files. I'd seen the occasions in the past where it exhibited in your samples on your blog, so I thought it might just be due to JPEG compression (and with that creamy-smooth boke you always get, I know it can be hard to avoid even at a very low compression.)

[Roger Clark]

Could one of you post examples for those of us who don't know what you are referring to. Thanks

Don,
I've been upgrading my computer and reorganizing my images, so as soon as that is complete, I'll see what I can find to illustrate the problem.

Roger

[Arthur Morris]

Ok. I wasn't sure if you were actually seeing it in the raw, or only in the output files. I'd seen the occasions in the past where it exhibited in your samples on your blog, so I thought it might just be due to JPEG compression (and with that creamy-smooth boke you always get, I know it can be hard to avoid even at a very low compression.)

Jon, Not to be a pain, but you actually gotta read the post :). This from Pane 1: "what causes the banding/posterization issues in the RAW files with pretty much all camera bodies? Is the any way to avoid it? To deal with it?"

[Jon Rista]

I did read it. You mentioned that it occurs with nearly all camera bodies. I have certainly not taken as many photos as you, however in the last year I've accumulated about 50,000 shots. I've never experienced the banding issue before myself, and if it was common (your not the first to mention it, I've read about it and seen example photos on other forums, where it usually boiled down to JPEG compression), I figure I'd have experienced the issue myself by now at least once. So I thought the issue might potentially be something other than the RAW itself. I was just trying to offer assistance. Apologies for any offense.

[John Chardine]

I will press the point a little about what is meant by "in the RAW files". What is needed is to demonstrate that it occurs in Adobe Camera Raw or DPP. A screen capture at that stage of the processing would be useful.

I took a look at the banding on the skimmer image from Artie's blog and I have to say I have never ever seen this in the RAW development stage in ACR or DPP.

[Arthur Morris]

Good suggestion John. My question has nothing to do with JPEGs .

[Arthur Morris]

When viewing the RAW file in BreezeBrowser (actually the JPEG of course), this image showed horrific banding. I processed it in DPP, saved the highlights in part by reducing the Saturation, processed it in Photoshop, boosted the color with Vibrance after noting that increasing the Saturation brought back the banding. It looks great--no banding at all, but do see the image in the next Pane.

[Arthur Morris]

For this one I increased the Saturation 30 points; the banding resulted.

[Roger Clark]

Art,
Are you making 8 or 16-bit files out of the raw conversion?

Roger

[Arthur Morris]

16-bit conversions.

[Jon Rista]

Art,

Just to make sure I am seeing what I should be seeing. In pane #14, on both my work screen and my home screen, I definitely see the boost to saturation, which has made gradients and transitions harsher, however I can't say that I see any posterization or banding. It is entirely possible I'm not looking in the right place, or I might have my screen brightness too high or low. Any chance you could outline or highlight the regions in the photo from Pane #14 that are exhibiting the issues your seeing?

T/Y!

[Michael Gerald-Yamasaki]

Artie,

Greetings. I see posterization in both panes #13 & #14. I'm inclined to think that the issue is with the representation of color in 8 bit color per channel. Unless you are working with a 10 bit monitor-graphics board-cable-software display all images you see on the screen are 8-bits of color per channel (irregardless of 16 conversion, storage or otherwise). Darker reds and blues tend to be more problematic (due to how the software maps luma to rgb in conversions and such).

The posterization issues are IMO entirely due to 8-bit color display and for screen viewing are avoidable by going with a higher bit monitor-graphics board-etc. setup (of course that doesn't stop others with 8-bit displays from seeing the posterization). Printing with at 16 bit workflow should not exhibit the posterization. I'd recommend not worrying about it.

Jon, the posterization is most easily seen toward the edges of the images in the transition from brighter red/orange to dark.

Cheers,

-Michael-

[Arthur Morris]

Just above the sun inside the black boundary lines :)

[Jon Rista]

Hmm. I might have to take a closer look at home. I have an Apple CinemaDisplay 30" at home, and I did not notice anything when I looked there before I headed into work (although admittedly I did not scrutinize either image for long as I was short on time). At work, I use two Dell 24" screens (Model: P2411Hb). On one of them, if I angle my head the right way in relation to the screen, I see some very slight posterization in #14 in that transition area from the brighter orange around the sun to the darker brown around the periphery, although it is pretty faint. On the other, its too difficult to tell...every so often I think I can, but in general I don't really see much posterization.

Is the effect really just very slight, or should I be seeing something more pronounced? I'm inclined to agree with @Michael, that the issue could very well be related to the bit depth (and possibly quality) of your video card and computer screen. A lot of screens are simpler 6-bit screens, which support much faster response times but at the cost of color fidelity. The only screens I really know could eliminate the problem are ones with hardware LUTs, such as a high end Eizo or possibly a LaCie. LaCie used to make one of the most amazing screen's I had ever set eyes on, a full 10-bit RGB LED display that supported 123% of the AdobeRGB gamut...I assume that with such a large gamut and bit depth posterization would never be an issue. They seem to be slowly exiting the display market, however, and I am not sure that beautiful gem of a screen is still sold. Eizo, or possibly a high-end NEC, might be the last two options for full 10-bit displays or displays with high quality LUTs that could eliminate posteriaztion issues (assuming it is a hardware-related problem, and further assuming you guys don't already own a nice high-end display like that.)

I am curious to see if the posterization exhibits on my home screen, though. On my one screen here, I can sort of see it, however it is incredibly faint.

[Michael Gerald-Yamasaki]

Jon,

The posterization is subtle. I put a curve to the right side of the image (panel #13) to show the pattern that you can then pick out in the original. One thing of interest to note is the image with the curve applied (right side) is may appear less posterized than the original due to noise, which helps break down posterization (something akin to dithering).




Once you see it, it's hard not to see it.

Cheers,

-Michael-

[Jon Rista]

Ok, I'm a believer now. I see posterization in both #13 and #14. The ambient lighting is a lot lower at this time of day (cloudy day just before sunset), so the photo and its nature stands out a lot better than it did earlier. I use a calibrated screen at home, calibrated with a DataColor Spyder3 and the latest software. There is a handy color profile switching tool that comes with the DataColor calibration device. I have a bunch of profiles that I've accumulated, for a variety of different white points (D50, D52, D55, D65).

On a whim, I started switching color profiles. I decided to check because of @Michael's original comments about color lookup. Depending on which one I had selected, the nature of the posterization changed. In none of them did it become entirely eliminated, however at D55, gamma 2.2, 110mcd posterization was hardly visible at all...I have to scrutinize to see any. At D50 70-80mcd, even when the screen brightness was properly set, posterization was really bad. I could see it clearly, and in the outer regions of the brown there were several bands of a sickly green. At D55 90mcd, it was not as bad, but definitely more visible than at D55 110mcd. I have a profile somewhere (I must have saved it in the wrong location) for 200mcd brightness. The Apple CinemaDisplay is a 400mcd screen at full brightness, and I wonder if improperly calibrating at an improper white point and preferred screen brightness will bring out posterization more. On my screen, lower brightness at D55 (my preferred white point) definitely results in more posterization. I am not really sure if a D65 white point results in more or not...color shifts a fair bit between D55 and D65, so it may simply be that viewing my screen with a white point I am not used to makes the posterization stand out more.

I have not calibrated my screen for a couple months. I'll run through a few calibrations with different settings, and see if there is any particular one that eliminates the posterization, or at least minimizes it to the point where it isn't an issue. It was rather remarkable though, with my profiles for a dimmer screen brightness, how pronounced the posterization was. I'll see if I can take a few photos of my screen for comparison. If anyone has a screen calibrator and can run through some trials of their own, perhaps we can prove or disprove the hypothesis that screen calibration is either the cause, or at the very least can greatly exacerbate posterization present in a RAW file.

P.S. As a point of reference, viewing on my uncalibrated Dell screens at work, I was unable to observe any posterization at all. The screens are all 100% default factory settings, and in all honesty both samples of the sun uploaded by Art looked great.

[Tom Graham]

I would submit that any graduated in luminance 8 bit image will show what is seen in panel #21 - if "expanded/stretched" (curves etc) enough in contrast.
In the meantime, I'll wait for Rogers Clark's analysis of this.
Tom

[Michael Gerald-Yamasaki]

I would submit that any graduated in luminance 8 bit image will show what is seen in panel #21 - if "expanded/stretched" (curves etc) enough in contrast.
In the meantime, I'll wait for Rogers Clark's analysis of this.
Tom

Tom,

You might try it on some random jpeg you have for grins.

Otherwise, you miss the point. The image with the curve applied is just to show the posterization pattern which is (once you see the pattern) readily seen in the originals in panels #13 & #14.

Cheers,

-Michael-

[Michael Gerald-Yamasaki]

perhaps we can prove or disprove the hypothesis that screen calibration is either the cause, or at the very least can greatly exacerbate posterization present in a RAW file.

Jon,

Er, the thing is the posterization is most likely not in the RAW file. Remember anything you see on the screen is reduced to 8 bits (from the 12 or 14 bits commonly in RAW) the posterization is introduced at that point and is really an artifact of the reduction to 8 bits. This is true even as you work with ACR or DPP or CNX, all you see on the screen is 8 bits (unless you have the hardware and software to display 10 bit images).

The important thing is in order not to actually maintain the posterization in your images is to not drop down to 8 bits in your workflow. Even as the display is in 8 bits, the saved files and processing in PS, for instance, can be in an always 16 bit workflow.

Artie, So there is the answer to the OP, always work in 16 bits even if you see posterization it is most likely not present in the 16 bit file that you are working. The posterization can be seen as an artifact of the display that is not contained in the image file and as such should print just fine as a 16 bit image without any posterization.

The way I see it... ymmv

Cheers,

-Michael-

[Jon Rista]

I understand its not "in" the RAW file, however Art mentioned he was seeing the problem when viewing RAW images in his editor, and that it wasn't simply a JPEG issue. I said "in the RAW" simply to differentiate between RAW and JPEG, since my previous comments referred to the problem exhibiting in JPEGs with higher compression. I'll be more precise in future posts to alleviate confusion.

Personally, I work in all 16-bit, ProPhotoRGB gamut, for my entire workflow until I export for some kind of final output (be it for viewing on screen via web, or for print.) Only at that point might I convert to 8-bit or sRGB.

I do believe that certain calibration settings can exacerbate how much posterization exhibits when viewing any image (including a RAW). I used to work with my screen calibrated for an 80mcd brightness, however all of my screen profiles at or around that brightness exhibit very pronounced posterization in panes #13 & #14 (as well as pane #21). At brighter calibrations, posterization also exhibits more, however to a much lesser degree. It seems a calibration between 100mcd and 200mcd produces the best results, with minimal posterization (although it is still present.) I recently moved to using CFL bulbs in my home, with 4100K bulbs in my workstation area (I like the whiter light better for reviewing my prints). It also appears that when calibrating under the CFL light results in a greener tint to my screen calibration than before when I used tungsten bulbs or daylight. I assume there is a certain amount of magenta in the new CFLs. All of my profiles calibrated under CFL ambient exhibit more posterization than those calibrated either under sunlight or tungsten ambient, as certain bands take on a slightly green tinge that makes them stand out more.

Currently, with a previous D55 Gamma 2.2 100mcd calibration created during the daytime, I get the most natural color with the least posterization. While reduction to 8 bit color from 14 bit color RAW or 16 bit TIFF is the fundamental issue, I think that with careful calibration in the right way at in the right ambient light can minimize the negative effects of posterization, as a lot of how color is converted between source data and target data or rendered on screen is color management software, gamut, and how colors are looked up.

I'll be interested in hearing what Roger has to say about it all, though.

[Roger Clark]

Jon,

Er, the thing is the posterization is most likely not in the RAW file. Remember anything you see on the screen is reduced to 8 bits (from the 12 or 14 bits commonly in RAW) the posterization is introduced at that point and is really an artifact of the reduction to 8 bits. This is true even as you work with ACR or DPP or CNX, all you see on the screen is 8 bits (unless you have the hardware and software to display 10 bit images).

The important thing is in order not to actually maintain the posterization in your images is to not drop down to 8 bits in your workflow. Even as the display is in 8 bits, the saved files and processing in PS, for instance, can be in an always 16 bit workflow.

Artie, So there is the answer to the OP, always work in 16 bits even if you see posterization it is most likely not present in the 16 bit file that you are working. The posterization can be seen as an artifact of the display that is not contained in the image file and as such should print just fine as a 16 bit image without any posterization.

The way I see it... ymmv

Cheers,

-Michael-

If the posterization was a visual artifact of 8-bit displays, then running the curves tool would not enhance it. If one can do this on the 16-bit file (as opposed to the posted jpegs), then it would be 1) either an artifact of raw conversion, or 2) in the raw data. The raw data are linear so should not actually be in the raw data. However, the tone curve applied greatly compresses highlights into fewer levels, and that will cause posterization. The function is a variable gamma function. I have the equation somewhere and will have to dig it out if anyone is interested.

Roger

[Roger Clark]

Artie,

I know I've seen this issue before, but I have looked through my images but I could not find one that showed the banding. I may have deleted those images as bad exposure. If you send me the raw file (and if you used photoshop ACR, the xmp file), I'll investigate the cause. I have tools to look at the raw data, including 32 bit, linear tone curve, floating point conversion. My email can handle a raw file rnclark at my usual clarkvision.com.

Roger

[Michael Gerald-Yamasaki]

If the posterization was a visual artifact of 8-bit displays, then running the curves tool would not enhance it.

Agree. Just to be clear, I ran the curves on an 8 bit image (the jpegs Artie posted) for the purpose of pointing out to Jon where the posterization was in panels #13 & #14... not to show that how posterization is a visual artifact of 8 -bit displays (one would need the original Raw file or a 16-bit Tiff to do that).

One way to indicate if the posterization due to the reduction to 8-bits: Force the software that is converting to 8-bit to display several versions of the image (zooming at different magnifications is usually handy) and see if the posterization changes or is in precisely the same location. If the posterization changes then it is an 8-bit reduction artifact. If the posterization stays precisely the same, it is most likely in the original data.

Oh, actually, as you say Roger, one could also run curves on the or any other tone changing method, and if the posterization changes... it's a 8-bit reduction artifact. The change will look like a change in the number of levels of posterization, the number of steps in the image.

Cheers,

-Michael-

[Jon Rista]

Oh, actually, as you say Roger, one could also run curves on the or any other tone changing method, and if the posterization changes... it's a 8-bit reduction artifact. The change will look like a change in the number of levels of posterization, the number of steps in the image.

Cheers,

-Michael-

Changing tones (or white & black points and the mapping of levels between them, along with changing gamut) is part of what ICM does. What you see on-screen when using a tool like Lightroom or Photoshop is a heavily processed image that has been converted from one color space (the source space) to the working space (Photoshop/Lightroom) and finally to the destination space (the screen). Along with the color gamut processing (which with either relative colorimetric or perceptual rendering intent could cause some posterization if the gamuts vary widely) is also the bit depth conversion from source to working to destination depth. I believe that can be emperically demonstrated as well by calibrating a screen to different settings and comparing the output of each profile. I am probably unable to eliminate all posterization in #13 & #14 with this method because some of the posterization is baked into the image as a result of JPEG compression. I'd need a RAW version to test any further.

ICM is probably not the entire cause, either. Most RAW images are not edited in a linear space anyway. Some kind of base tone curve (image style, picture style, camera setting) is applied to every RAW on import into most RAW editors. That would apply a non-linear tone curve, which could introduce additional posterization dependent on the application being used to edit.

[Michael Gerald-Yamasaki]

Changing tones (or white & black points and the mapping of levels between them, along with changing gamut) is part of what ICM does. What you see on-screen when using a tool like Lightroom or Photoshop is a heavily processed image that has been converted from one color space (the source space) to the working space (Photoshop/Lightroom) and finally to the destination space (the screen). Along with the color gamut processing (which with either relative colorimetric or perceptual rendering intent could cause some posterization if the gamuts vary widely) is also the bit depth conversion from source to working to destination depth. I believe that can be emperically demonstrated as well by calibrating a screen to different settings and comparing the output of each profile. I am probably unable to eliminate all posterization in #13 & #14 with this method because some of the posterization is baked into the image as a result of JPEG compression. I'd need a RAW version to test any further.

ICM is probably not the entire cause, either. Most RAW images are not edited in a linear space anyway. Some kind of base tone curve (image style, picture style, camera setting) is applied to every RAW on import into most RAW editors. That would apply a non-linear tone curve, which could introduce additional posterization dependent on the application being used to edit.

While color management is an important issue in the appearance of an image on the screen, posterization is a distinct issue aside from color management. It's true by changing the color profile of a device the appearance of posterization can be minimized but color profiles do not change the underlying RGB values in which the posterization is encoded. As such, changing the color profile will not change the basic pattern of the posterization levels but rather the visual variance between the steps in the pattern (much in the way the curve applied to the 8 bit jpeg in panel #21 didn't change the pattern but made it more obvious).

I would caution against profiling your monitor to make differences in tone indistinquishable. The posterization in panels #13 and #14 is burned into the 8 bit jpegs and is not an artifact of the screen rendering of the jpeg. The posterization was introduced as the jpeg was created and so is burned in there. The rendering of the jpegs in panels #13 & #14 should show posterization because it is now in the 8 bit jpeg data itself.

Cheers,

-Michael-

[Arthur Morris]

Jon,

Er, the thing is the posterization is most likely not in the RAW file. Remember anything you see on the screen is reduced to 8 bits (from the 12 or 14 bits commonly in RAW) the posterization is introduced at that point and is really an artifact of the reduction to 8 bits. This is true even as you work with ACR or DPP or CNX, all you see on the screen is 8 bits (unless you have the hardware and software to display 10 bit images).

The important thing is in order not to actually maintain the posterization in your images is to not drop down to 8 bits in your workflow. Even as the display is in 8 bits, the saved files and processing in PS, for instance, can be in an always 16 bit workflow.

Artie, So there is the answer to the OP, always work in 16 bits even if you see posterization it is most likely not present in the 16 bit file that you are working. The posterization can be seen as an artifact of the display that is not contained in the image file and as such should print just fine as a 16 bit image without any posterization.

The way I see it... ymmv

Cheers,

-Michael-

Thanks Michael for your excellent contributions to this thread!

[Roger Clark]

but color profiles do not change the underlying RGB values in which the posterization is encoded.

Michael,
That is not quite true. I have had to stop using photoshop for some scientific applications because simply reading a file into photoshop changes image values. The changes seem to vary depending on the color profile in use.

Roger

[Jon Rista]

While color management is an important issue in the appearance of an image on the screen, posterization is a distinct issue aside from color management. It's true by changing the color profile of a device the appearance of posterization can be minimized but color profiles do not change the underlying RGB values in which the posterization is encoded. As such, changing the color profile will not change the basic pattern of the posterization levels but rather the visual variance between the steps in the pattern (much in the way the curve applied to the 8 bit jpeg in panel #21 didn't change the pattern but made it more obvious).

I would caution against profiling your monitor to make differences in tone indistinquishable. The posterization in panels #13 and #14 is burned into the 8 bit jpegs and is not an artifact of the screen rendering of the jpeg. The posterization was introduced as the jpeg was created and so is burned in there. The rendering of the jpegs in panels #13 & #14 should show posterization because it is now in the 8 bit jpeg data itself.

Cheers,

-Michael-

You're saying the pretty much the same thing I'm saying, just in a different way, although we may be arguing from different angles. I understand that the sample images posted are 8-bit JPEG and "burn in" the effects tone curves and the like may have. I believe I said that very same thing myself in an earlier post. Thats not my point, though. Art originally stated (and made it very clear to me when I originally mentioned JPEG) that he was seeing the problem when working his RAW images, and that the problem was not due to 8-bit conversion, JPEG compression, or anything like that. Given that the original data in a RAW image is linear, I'm skeptical that the problem is actually something that has been "encoded". However there is plenty of other processing going on when you work a RAW, from color management to camera profiles to potentially your own tweaks to tone curves and camera profile settings that could affect how smoothly tonal grades present, and introduce posterization artfacts into your workflow that is not otherwise present in the original data (assuming you ARE working a RAW image, and not an 8-bit TIFF or JPEG.)

I agree, posterization exhibits due to the degree of difference between tonal "steps" or levels, and when the difference between neighboring levels is increased beyond some acceptable norm, a harsher transition is visible. Changing color profiles does not change the pattern of the posterization, however it can indeed change how it exhibits. As I mentioned in my previous posts, calibrating my display outside of the 100-200mcd range (180mcd is the recommended calibration for my screen), posterization exhibits more harshly, and with a greater number of bands (which are probably already there to begin with, just now with harsher transitions between them). Additionally, calibrating under artificial light rather than full-spectrum daylight also introduces a color cast that changes how certain bands of posterization exhibit. At a calibration of D55, Gamma 2.2, 110mcd (generated under daylight ambient) the posterization in the JPEG's from #13 and #14 is minimal, and many of the steps between bands disappear entirely. At a similar calibration, only for a lower brightness, posterization became extreme and many more bands exhibit, some of them in a visibly (and incorrect) greener tint.

While you caution against profiling to minimize posterization, if it is something that bothers you while editing, why not do something to improve the situation? While I am not certain what changes might be needed for every screen, I would say calibrating to a higher brightness than you generally do, while keeping white point and gamut the same, is a small price to pay to change bad or terrible posterization into hardly noticeable posterization. I would also argue that an ugly presentation of posterization has a worse effect on processing your RAW's than tuning the screen for a higher brightness (which, BTW, doesn't necessarily mean you have to keep it that bright, you can still drop the brightness setting a bit) , as it can present in such a way as to affect color accuracy.

I would post samples of the effect on my own screen, but its proving difficult to take a photograph of my screen that will actually demonstrate the color shifts. I'll post something if I manage to get some working photos. I don't seem to have any photos of my own that exhibit posterization during any part of my own workflow, either in Lightroom or Photoshop (I'm using LR 4.1 and PS 6, not sure if either of those may affect whether or how posterization presents.) Anyway, I believe it is possible to visibly demonstrate to yourself how different screen calibrations can exacerbate problems with posterization if you're interested (and have a calibrator to experiment with), and have any interest in trying to rectify the situation on your own systems.

Now, to be frank, while I may be new here, I do know what I'm talking about. I'm getting a little tired of being treated like I don't, and Art seems happy with an answer provided by Michael, so I'm out. Best of luck trying to figure out any further issues with posterization if they arise.

[Michael Gerald-Yamasaki]

Now, to be frank, while I may be new here, I do know what I'm talking about. I'm getting a little tired of being treated like I don't, and Art seems happy with an answer provided by Michael, so I'm out. Best of luck trying to figure out any further issues with posterization if they arise.

Hey, Jon... sorry if you think I'm treating you like you don't know what you're talking about. I think we just have an honest disagreement about what's important regarding posterization. Don't mean to be off putting.

I must admit to being a bit confused at the recommendation to change calibration of one device (one's monitor) for the purposes of minimizing the impact of an artifact. I thought the purpose behind calibration was to make a particular device respond in the same way as other devices so that when you made changes to an image you could gain some assurance that other devices would see the change in the same way (other devices such as printers, other monitors). It puts everyone on the same page. Altering one's monitor to minimize posterization seems antithetical to the purpose of calibrating in the first place.

How I see it, could be wrong.

Cheers,

-Michael-

[Michael Gerald-Yamasaki]

Thanks Michael for your excellent contributions to this thread!

Artie... YAW

Cheers,

-Michael-

[Jon Rista]

I guess I don't see calibrating ones screen to normalize its response to the Camera. I don't even think that is really possible, given all the variables. I see calibration as normalizing the response of the screen to produce valid, accurate output in a given viewing context. At least, thats how must commercial calibration devices, such as those from iOne, DataColor, Pantone, etc. function. You calibrate to a specified configuration, such as a 5000 kelvin (D50) or 6500 kelvin (D65) white point, a Gamma of 2.2 (in the case of most screens) or 1.8 (in the case of some Mac setups), and to a certain screen brightness setting. I'm not really sure, given the common screen calibration options, how that normalizes the response of the screen to any given camera. (It might be that some see Gamma correction, which is also done in profiling, as an explicit means of normalizing device response. When it comes to film, there is indeed a difference in gamma between film and photographic paper/scanner/screen. Since digital cameras are linear devices...I am not sure that gamma correction plays a direct role when editing RAW....at least not at the time of screen calibration.)

If you need to tune how a RAW image is rendered on your screen (assuming the screen is properly calibrated in the first place), which I guess could be considered normalizing the response of the camera and screen, you would need to first photograph something like a Gretag Macbeth color checker card under a known (or at least understood) illuminant . Once you have that, its a rather painstaking process to tune the colors as they appear on screen to match the colors on the card, under the same illuminant. Usually you only try to correct colors that have a certain deviation from norm (i.e. the ones that stand out on screen) to produce "checked color". Thats more of a studio thing, though, and I don't know of anyone who does that for nature photography.

Anyway, I see tuning the calibration of a screen as an entirely valid way to correct something that is obviously incorrect when viewed on screen. I wouldn't have guessed that calibration could cause posterization to exhibit in the first place (remember my original skepticism?) until you mentioned something in one of your earlier posts that indicated such might indeed be the problem, and I actually tested a variety of my existing screen profiles against Arts images. However seeing the difference in the exhibition of posterization in all of the images in this thread under different profiles (and even profiles generated by my Spyder3 under different types of ambient illumination), I believe they do indeed have an impact...and in some cases it is significant.

[Michael Gerald-Yamasaki]

Michael,
That is not quite true. I have had to stop using photoshop for some scientific applications because simply reading a file into photoshop changes image values. The changes seem to vary depending on the color profile in use.

Roger

Roger,

That's surprising (unless PS was set up to convert to the working space profile). A test for the impact of assigning a color profile is to put a selection of color samplers in your image and then assign an alternate color profile. The colors in the image will change, sometimes dramatically, depending on the profile chosen. Yet the rgb values will stay the same.

Converting color profiles would change the rgb values, but that's not what I was talking about.

Cheers,

-Michael-