proof colors in PS RAW editor?

[marlinspike]

Is there a way to use proof colors in the RAW conversion editor of PS CS2? It's kinda a pain doing the guess and check routine (my proof colors are very different...I wish I could just get PS to use the same color profile as the rest of windows without having to press ctrl+y).

[arodney]

Is there a way to use proof colors in the RAW conversion editor of PS CS2?

Nope.

[marlinspike]

Well that's just genius on their part. Is there a way to make PS CS2 uncolormanage itself, i.e. just use whatever the rest of windows and IE and Firefox are using?

[arodney]

Well that's just genius on their part. Is there a way to make PS CS2 uncolormanage itself, i.e. just use whatever the rest of windows and IE and Firefox are using?

There's no way to turn off color management in Photoshop! You can setup a custom proof setup to simulate what an image would look like outside an ICC aware application (Monitor RGB, sRGB, Mac RGB) but to do so, you have to use color management!

[colourbox]

I wish I could just get PS to use the same color profile as the rest of windows without having to press ctrl+y).

Wait a sec...what profile is "the rest of Windows" using? I seriously doubt that there is just one, or any, that you could latch on to. The problem is not for Adobe to dumb down to chaos, but for Windows to smarten up to consistency. The closest you'll get is to calibrate your monitor, set Camera Raw to sRGB, and get back to work.

[arodney]

Wait a sec...what profile is "the rest of Windows" using.

Its not. It simply sending the RGB numbers to the display.

[marlinspike]

Wait a sec...what profile is "the rest of Windows" using? I seriously doubt that there is just one, or any, that you could latch on to. The problem is not for Adobe to dumb down to chaos, but for Windows to smarten up to consistency. The closest you'll get is to calibrate your monitor, set Camera Raw to sRGB, and get back to work.

Well, the thing is this, I can't find my eye-one display 2 (presumably I left it back at my parents' home, but I don't know when I'll next return, probably not until next Christmas), and the rest of windows, perhaps by chances, matches the prints from my color profiled printer almost exactly (my screen is ever so slightly brighter than the prints, but the colors are dead on). On the other hand, whatever profile PS is using is grossly inaccurate. Just for a sample, 0R 0G 255B is purple in PS.

Inaccurate PS colors are corrected if I do proof colors with the proof setup set to monitor rgb. When I go to windows (this is Vista Business FWIW) color management, there is an icc profile loaded called "wide viewing angle & High density FlexView Display (default)". (I have one of the last laptop flexview displays, w00t). "Use my settings for this device" is NOT checked.

So, what do I need to do to make PS look like chaos which just happens to be fairly true?

[arodney]

Inaccurate PS colors are corrected if I do proof colors with the proof setup set to monitor rgb.

You need to change your thinking here. The previews are NOT inaccurate, they are accurate. Using Monitor RGB just makes them inaccurate to match the other inaccurate previews. It solves nothing. Just as cranking up the luminance control on your display while viewing a dark image does nothing to correct the problem.

[marlinspike]

You need to change your thinking here. The previews are NOT inaccurate, they are accurate. Using Monitor RGB just makes them inaccurate to match the other inaccurate previews. It solves nothing. Just as cranking up the luminance control on your display while viewing a dark image does nothing to correct the problem.

No, see, the non-profiled world of my computer matches almost exactly (the only difference being that my monitor shows things just a hair lighter) the prints made by my printer which is color profiled. The same is true of PS when I have proof colors selected. When proof colors are not selected in PS, the image I get is not even close. Blue is purple, yellow is green, up is down, it's borderline surreal.

So, someone please tell me how to make PS like the rest of my computer's world, or offer to lend me their eye one display 2 since I can't find my own (lol).

I know it's not ideal to go around with no profile, but I need something to carry me over until the next time I return to my parents' home. It just so happens by luck that the non-profiled world of my computer is very very close to accurate. I want to get rid of this darn Salavador Dali color profile that PS is apparently using.

[arodney]

No, see, the non-profiled world of my computer matches almost exactly (the only difference being that my monitor shows things just a hair lighter) the prints made by my printer which is color profiled.

Maybe today. But a display is an unstable device that requires regular calibration and profiling, the profile is used by ICC aware applications. Plus you're not using Photoshop correctly nor at its capacity because you're not soft proofing (and you can't without color management).

[dmmattix]

Sounds like a lot of work and thinking. You don't say where your parents live but I suspect they would have access to some sort of shipping to send you your Eye-One. Now would not that be simpler than trying to match odd PS colors to what your printer is outputting?

Or am I missing something here?

[marlinspike]

Well, for one, I have no idea where at home it is - last time I was there, I looked and couldn't find it, but I didn't look seriously because I figured it was at my place, but it turns out it's not.

Keep in mind that families can be crazy and just trust me when I say having it shipped to me is not an option.


So, rather than lecture me on the things I already know about why I should have a profiled monitor, can someone just tell me what to do to make PS as unprofiled as the rest of Windows?

[jfriend]

Well, for one, I have no idea where at home it is - last time I was there, I looked and couldn't find it, but I didn't look seriously because I figured it was at my place, but it turns out it's not.

Keep in mind that families can be crazy and just trust me when I say having it shipped to me is not an option.


So, rather than lecture me on the things I already know about why I should have a profiled monitor, can someone just tell me what to do to make PS as unprofiled as the rest of Windows?

While I'm in the camp that thinks you should find a better answer to this dilemna, I'll attempt to answer your question anyway.

You cannot directly turn color management off in Photoshop. It will correct your colors according to the profile it finds in your system for your monitor. Further, Photoshop will look at the profile of your source document and pay attention to that. Non-color-managed apps ignore the profile of the source document completely. They just dump raw RGB values to the screen. So, there is no screen profile that will make all source profiles convert to the same thing as the non-color-managed apps. But, you could pick one source profile (say sRGB) and get a monitor profile that essentially told Photoshop that your monitor was already perfect sRGB so the amount of correction needed for the monitor was zero. It would call this a "null sRGB monitor profile".

Therefore, if you want the Photoshop display to look the same as apps that are doing no color management for sRGB images, you need something I this "null sRGB monitor profile" and you then need to tell Windows that that is your correct monitor profile. Photoshop will find it, use it and when it goes to correct your sRGB colors for proper display on your monitor, the correction needed will be zero. Wala, Photoshop color now looks as unmanaged as the rest of the apps in Windows.

So, where does one get a "null sRGB monitor profile"? That I don't know. Maybe some Google searching would turn something up.

[arodney]

Well that's just genius on their part. Is there a way to make PS CS2 uncolormanage itself, i.e. just use whatever the rest of windows and IE and Firefox are using?

No. It IS showing you the RGB values correctly (like Photoshop) and like Photoshop, there's no way not to use color management.

[marlinspike]

Ok, well then, if I set my working space to Monitor RGB or to the profile I made the last time I used my eye-one display 2 (which seemed not to have any impact), then the colors are accurate (actually, using the profile I made as opposed to monitor rgb gets rid of the images on screen are darker than they appear problem). So, what went wrong? There is apparently a proper color profile made by my i1d2, why can't I implement it (i.e. why are the colors so wrong if the working space is srgb but so right if the working space is my i1d2 made profile)?

[arodney]

Ok, well then, if I set my working space to Monitor RGB or to the profile I made the last time I used my eye-one display 2 (which seemed not to have any impact), then the colors are accurate (actually, using the profile I made as opposed to monitor rgb gets rid of the images on screen are darker than they appear problem). So, what went wrong?

What's wrong is that's not how ICC aware and non ICC aware applications work:

ICC Aware application. Examine the display profile and the document profile. Use Display Using Monitor Compensation to produce correct previews of numbers. This requires two profiles: Document and Display.

Non ICC Aware application. Send RGB numbers to the display. Application doesn't know the color space of the image (it sends the numbers without Display Using Monitor Compensation) nor does it see or understand the display profile. NO profiles being used.

There is a color look up table that's loaded by the profile and non ICC aware applications ARE affected in just these current settings and their effect on the display. But they don't understand the profiles in any other way. The previews are incorrect.

[marlinspike]

So...I have this profile that was made by an i1 display 2. What do I do with it to make it useful? I'm guessing it must be a valid profile since if I set my working space to it my colors are correct.

[arodney]

So...I have this profile that was made by an i1 display 2. What do I do with it to make it useful? I'm guessing it must be a valid profile since if I set my working space to it my colors are correct.

You don't have to do anything. Its a system display profile, all ICC aware applications will access and use it. You NEVER set the RGB working space with this profile!

[marlinspike]

You don't have to do anything. Its a system display profile, all ICC aware applications will access and use it. You NEVER set the RGB working space with this profile!

So what's the deal with my monitor displaying extremely inaccurate colors in PS, even when I set this profile to be the default in Windows Color Management?

[arodney]

So what's the deal with my monitor displaying extremely inaccurate colors in PS, even when I set this profile to be the default in Windows Color Management?

Photoshop looks at the display profile to build a preview. If you provide the wrong one, it might match another application that is not ICC color managed, doesn't mean its accurate.

IF you want to see accurate RGB numbers, you need an ICC aware application. It needs an accurate (well behaved) profile of your display that reflects its current behavior (displays are unstable devices). It also needs the scale of the RGB numbers (its color space).

The one way Photoshop could show you inaccurate colors is if the display profile is wrong (or bad).

[marlinspike]

The one way Photoshop could show you inaccurate colors is if the display profile is wrong (or bad).

So does the "test" of setting the display profile as the color space not prove that the profile is good? One thing I'm thinking is perhaps this indicates that I've double profiled (i.e. I perhaps made the profile on the screen that was already profiled and so that what I have to double up by making the profile the working space for it to be accurate?)

[arodney]

So does the "test" of setting the display profile as the color space not prove that the profile is good?

It does not. There are two operations going on. One is a LUT in the profile being applied which does control what you see in both ICC and non ICC aware applications (note, its often best for all but the high end LCD's to be calibrated such that this LUT is doing as little as possible since all you gain is banding using only 8-bits).

ICC aware applications care about what the profile tells it about the state of the display, LUT not withstanding. In fact that's HOW we can produce proper color appearance in ICC aware applications without altering the display with a LUT. The compensation is happening at 21 bit precision using the Adobe Color Engine (ACE). None of this is happening outside an ICC aware application.

There ARE ways to get an idea if the profile is good:
http://www.ppmag.com/reviews/200412_rodneycm.pdf

[marlinspike]

Ok, I found my color profiler (i1 Display 2). Installed the latest software and drivers, ran it, profiled the monitor, and still everything looks good except in PS CS2. In PS, if srgb is the color space, 0r 0g 0b looks purple instead of blue. If I set the profile as the color space, things look like they are supposed to. Ideas?

[marlinspike]

Ok, I found my color profiler (i1 Display 2). Installed the latest software and drivers, ran it, profiled the monitor, and still everything looks good except in PS CS2. In PS, if srgb is the color space, 0r 0g 255b looks purple instead of blue. If I set the monitor profile as the color space, things look like they are supposed to. Ideas?

Can anybody help me here? It's really annoying to have to set the monitor profile as the color space to get accurate colors when I plan to upload them to smugmug which IIRC uses sRGB. FWIW, when opening a file in CS2 output from Lightroom, it asks me what I want to do with the color space (use embedded srgb, convert to the working space, discard).

If I select use embedded, I get wrong colors.
If I select convert to working space, I get correct colors
If I select convert to working space and then set proof colors to sRGB with preserve RGB numbers checked, I get those same wrong colors I get when I select use embedded, but if I uncheck preserve RGB numbers and instead say relative colorimetric and use black point compression I get correct colors.

[arodney]

Can anybody help me here? It's really annoying to have to set the monitor profile as the color space to get accurate colors...

Stop right there. That's not a way to get "accurate" colors, ever. The very idea of modern ICC color management, seen throughout all Adobe products, is the display conditions are divorced from the editing space (working space). Your display profile is ONLY used for previews of documents to provide the correct previews of those color numbers.


when I plan to upload them to smugmug which IIRC uses sRGB. FWIW, when opening a file in CS2 output from Lightroom, it asks me what I want to do with the color space (use embedded srgb, convert to the working space, discard).

This tells me you exported the LR documents in sRGB (use embedded sRGB). The preview of that document in Photoshop should match Lightroom's.

What color space do you want to render copies out of Lightroom in?

[marlinspike]

Stop right there. That's not a way to get "accurate" colors, ever. The very idea of modern ICC color management, seen throughout all Adobe products, is the display conditions are divorced from the editing space (working space). Your display profile is ONLY used for previews of documents to provide the correct previews of those color numbers.




This tells me you exported the LR documents in sRGB (use embedded sRGB). The preview of that document in Photoshop should match Lightroom's.

What color space do you want to render copies out of Lightroom in?

I want my things to be rendered in sRGB, however, if I use sRGB as my color space, PS is CLEARLY showing incorrect colors. 0R 0G 255B will look purple. I'll edit it a photo and the colors will look weird on my laptop, but on other color calibrated computers the colors will look correct (as they do in prints I make with my calibrated printer). I can make the colors in PS CS2 LOOK correct by setting my monitor profile as my color space, but while this looks correct obviously it is a hopeless technique.

I know my monitor takes to calibration because I tried the LUT Tester and it worked. Also, when I calibrate my monitor it clearly makes a difference in how things look...except in PS CS2. It's as if within the PS CS2 window the calibration I make with my i1D2 is being ignored.

I'm in Vista Business 32-bit btw.

[marlinspike]

This may provide another hint as to what is wrong.

The colors of the thumbnails for my image files (i.e. the things I double click to open the image file in PS) are correct. However, if I right click that thumbnail and select "set as desktop" it will appear with incorrect colors as my desktop.

[arodney]

This may provide another hint as to what is wrong.

The colors of the thumbnails for my image files (i.e. the things I double click to open the image file in PS) are correct. However, if I right click that thumbnail and select "set as desktop" it will appear with incorrect colors as my desktop.

No, you think they are correct. Its far more likely they are not color managed. The one application that should, when setup correctly, preview the numbers correctly is Photoshop.

You either have a hosed display profile or something else that's off here. But assuming you've got a good display profile that Photoshop is seeing and using (something you can check in your Color Settings under RGB working space), then everything else may look better, but its wrong! Your desktop is not color managed, assuming you're not on a Mac. Vista is a color management nightmare FWIW.

[marlinspike]

No, you think they are correct. Its far more likely they are not color managed. The one application that should, when setup correctly, preview the numbers correctly is Photoshop.

You either have a hosed display profile or something else that's off here. But assuming you've got a good display profile that Photoshop is seeing and using (something you can check in your Color Settings under RGB working space), then everything else may look better, but its wrong! Your desktop is not color managed, assuming you're not on a Mac. Vista is a color management nightmare FWIW.

Ignore what I set about image colors when the image is set to desktop...I had done some strange things with the image I was using. I have made a new test image
OK, here is what I know:

-My printer is properly profiled.
-0R 0G 255B should look blue.
- my monitor responds to calibration as verified by the LUT tester
- I calibrated it using an eye-one d2
- If I make an image out of 0R 0G 255B, and the color space is sRGB, it will look purple in PS CS2, its thumbnail will look blue, and if I set it to be my desktop background my desktop will be blue
- If I take that same image but set the proof colors to be the monitor profile and turn on proof colors, it will now look blue in PS
- if I take that same original image and convert the color space to my monitor profile, nothing happens
- if I take that same original image and assign the profile to be my monitor profile, it will now look blue in PS

I also know that when I show an image here (say of a blue car) and ask people what color it is they say blue, on my desktop it looks blue, but in PS CS2 it looks purple.

[arodney]

Ignore what I set about image colors when the image is set to desktop...I had done some strange things with the image I was using. I have made a new test image
OK, here is what I know:

-My printer is properly profiled.
-0R 0G 255B should look blue.
- my monitor responds to calibration as verified by the LUT tester
- I calibrated it using an eye-one d2
- If I make an image out of 0R 0G 255B, and the color space is sRGB, it will look purple in PS CS2, its thumbnail will look blue, and if I set it to be my desktop background my desktop will be blue
- If I take that same image but set the proof colors to be the monitor profile and turn on proof colors, it will now look blue in PS
- if I take that same original image and convert the color space to my monitor profile, nothing happens
- if I take that same original image and assign the profile to be my monitor profile, it will now look blue in PS

I also know that when I show an image here (say of a blue car) and ask people what color it is they say blue, on my desktop it looks blue, but in PS CS2 it looks purple.

This is a classic example of the other app's showing you the numbers incorrectly but preferability. Photoshop is showing the current RGB numbers correctly (what those numbers are or should be in your case is still questionable). It sounds like the documents are not in sRGB or are tagged incorrectly. Something is not set correctly or there are some conversions or assignments being made somewhere in this chain.

The soft proofing you used illustrates that color management isn't being properly applied to the current RGB values with the assumed color space. When you applied your display profile, you kind of used Photoshop to show you what those numbers would look like untagged outside an ICC aware application.

[marlinspike]

This is a classic example of the other app's showing you the numbers incorrectly but preferability. Photoshop is showing the current RGB numbers correctly (what those numbers are or should be in your case is still questionable). It sounds like the documents are not in sRGB or are tagged incorrectly. Something is not set correctly or there are some conversions or assignments being made somewhere in this chain.

The soft proofing you used illustrates that color management isn't being properly applied to the current RGB values with the assumed color space. When you applied your display profile, you kind of used Photoshop to show you what those numbers would look like untagged outside an ICC aware application.


Hmmm...here's some more info that may help.

If I use the eye dropper in PS CS2 (when the image is in srgb color space) it verifies 0R 0G 255B (well...it says 254B for some reason) even though it looks purple. When I open the image in lightroom (where it looks purple too), it says 41.1%R 18.3%G and 93.8%B

I'm so very confused :cry

[jfriend]

Hmmm...here's some more info that may help.

If I use the eye dropper in PS CS2 (when the image is in srgb color space) it verifies 0R 0G 255B (well...it says 254B for some reason) even though it looks purple. When I open the image in lightroom (where it looks purple too), it says 41.1%R 18.3%G and 93.8%B

I'm so very confused :cry

My guess is you have double profiling where you have two separate processes correcting for your monitor profile. When you use color-managed software, you get double correcting, when you don't use color-managed software, you get a single correction and it looks right. If this is the case, the solution here is to get rid of the double profiling, not to turn off Photoshop's profiling.

Are you on XP or Vista? Do you have an LCD monitor or CRT monitor?

Can you look in your startup sequence and make sure that Adobe Gamma is not set to run? Is anything else running there that might have something to do with color or your monitor?

What kind of video card do you have? Do you have software in your startup sequence that comes with your video card that is loading? Is there are control panel app for your video card?

[marlinspike]

My guess is you have double profiling where you have two separate processes correcting for your monitor profile. When you use color-managed software, you get double correcting, when you don't use color-managed software, you get a single correction and it looks right. If this is the case, the solution here is to get rid of the double profiling, not to turn off Photoshop's profiling.

Are you on XP or Vista? Do you have an LCD monitor or CRT monitor?

Can you look in your startup sequence and make sure that Adobe Gamma is not set to run? Is anything else running there that might have something to do with color or your monitor?

What kind of video card do you have? Do you have software in your startup sequence that comes with your video card that is loading? Is there are control panel app for your video card?

I'm on Vista Business 32-bit. It's a laptop, but the screen is an S-IPS lcd. Video card is an ATI V5200. The ATI control center is currently set to deactivate ATI color controls and I don't have adobe gamma in the startup.

[skibum4]

What's wrong is that's not how ICC aware and non ICC aware applications work:

ICC Aware application. Examine the display profile and the document profile. Use Display Using Monitor Compensation to produce correct previews of numbers. This requires two profiles: Document and Display.

Non ICC Aware application. Send RGB numbers to the display. Application doesn't know the color space of the image (it sends the numbers without Display Using Monitor Compensation) nor does it see or understand the display profile. NO profiles being used.

There is a color look up table that's loaded by the profile and non ICC aware applications ARE affected in just these current settings and their effect on the display. But they don't understand the profiles in any other way. The previews are incorrect.


for sRGB images though shouldn't the LUT and profile loader on startup be all you need? It seems strange to me that IE or irfanview show simple sRGB profile jpgs differently than say CS4.

When the LUT table gets loaded in it does very much change how pictures look in IE and irfanview (grayscales, etc. everything look much better and gamma along the whole ramp changes, etc.) so shouldn't sRGB pics look fine in them since isn't the LUT supposed to matching sRGB to my monitor? And isn't the whole point of the loading up the profile to make the gamma and white point and color balance of my monitor closer to the exact sRGB
I know they wont make adobergb and non-sRGB stuff look right in non-ICC aware programs, but shouldn't sRGB pics display the same way in any program??

and waht is most bizarre is this:

and why would loading the LUT and then removing the monitor in windows color mangement profile before loading CS4 -OR- not loading the LUT at startup but leaving the monitor profile associated with the monitor make CS4 display things in the exact same way? Doesn't that imply that the LUT laoding at startup does the EXACT same thing CS4 does itself when it gets informed by windows that there is an ICC profile assigned to the monitor and it decides to take that into account itself?
And isn't using both then double profile? I.E. CS4 say does sRGB->Monitor Profile and then the video card has the sRGB->Monitor Profile loaded into it and it gets fed the monitor profiled info from CS4 thinking it is sRGB and then it applies the transform again? THis really really seems to be what is happening, since if I apply the transform twice it looks just like what CS4 shows when I load the LUT at start and windows informs it about a monitor profile.

and using just the LUT at start OR just having windows tell CS4 about the minitor profile but not both not only both make CS4 do the exact same thing but both make it show sRGB jpg files exactly the way that irfanview does if I have the LUT loaded into the video card. and again when I leave everything set then colro-managed stuff looks different, but oddly enough in EXACTLY the same way as if the LUT and matrix transform loaded into the video card where to be applied to data and then again applied to that data....

leaving the LUT loading and the profile assigned seems to do very much what it seems doing about twice of everything the LUT alone did.

I thought the LUT profile loader applied a matrix transform (as well as additionally adjustments for those that also have a LUT) in the graphics card which transforms from say sRGB primaries to what the actually are on the monitor, no? Isn't that jsut what gets done when you load the profile with the LUT loader? Why do you need CS4 to take advantage of that?
I thought ICC savvy for color-aware programs only meant they could ahndle adobeRGB, ProPhotoRGB and other non-sRGB spaces in additon to the sRGB that everything else can handle fine.


when I do load the LUT at start and do leave the profile coloreyes made assigned to the monitor photoshop alters the colors a bit AND even makes shadows a bit brighter compared to irfanview.... and it just seems very surprising that loading the LUT alone or just having it apply the profile alone should make it render with, as far as I can tell, the EXACT same look, why in the world should those two things end up shifting things the exact same way if they really are doing different things?

and if CS4 is not ending up with double profiles applied then why do prints that i make in CS4 look a lot more like the images shown in irfanview (or CS4 with only LUT or profile but not both)?

I also noted as did the OP that if I set proofing on and use sRGB as the workspace and set the target as my monitor profile then colors look exactly as they do in irfanview and such.

[skibum4]

It does not. There are two operations going on. One is a LUT in the profile being applied which does control what you see in both ICC and non ICC aware applications (note, its often best for all but the high end LCD's to be calibrated such that this LUT is doing as little as possible since all you gain is banding using only 8-bits).

ICC aware applications care about what the profile tells it about the state of the display, LUT not withstanding. In fact that's HOW we can produce proper color appearance in ICC aware applications without altering the display with a LUT. The compensation is happening at 21 bit precision using the Adobe Color Engine (ACE). None of this is happening outside an ICC aware application.

There ARE ways to get an idea if the profile is good:
http://www.ppmag.com/reviews/200412_rodneycm.pdf

but if the LUT loaded into a graphics card is doing its best to account from how the monitor differs from sRGB and if a color-aware program looking at the monitor profile does it's best to shift colors as they need be to give the right color on teh monitor to match the sRGB color and then if that gets sent to the card which is already using the LUT why does this not give double profiling?

Note that when I don't use the LUT loader but do have the monitor assigned in windows color management that the prints CS4 makes look closer than what CS4 shows when I also load the LUT as well and that loading one or the other makes CS4 show images the EXACT same way and that having both set at once appears to just shift colors and brightness by twice the amount as having only set at once.

so why does not imply that color-aware programs end up doing double profiling under windows vista when the LUT is also loaded at startup??

[skibum4]

Ok, I found my color profiler (i1 Display 2). Installed the latest software and drivers, ran it, profiled the monitor, and still everything looks good except in PS CS2. In PS, if srgb is the color space, 0r 0g 0b looks purple instead of blue. If I set the profile as the color space, things look like they are supposed to. Ideas?

same here (although I should note that what we think of blue primary actually is a bit more like what most people are used to calling purplish)

but that said I do notice a shift towards purple, twice as large when I use both the LUT and leave monitor profile assigned

[skibum4]

My guess is you have double profiling where you have two separate processes correcting for your monitor profile. When you use color-managed software, you get double correcting, when you don't use color-managed software, you get a single correction and it looks right. If this is the case, the solution here is to get rid of the double profiling, not to turn off Photoshop's profiling.

Are you on XP or Vista? Do you have an LCD monitor or CRT monitor?

Can you look in your startup sequence and make sure that Adobe Gamma is not set to run? Is anything else running there that might have something to do with color or your monitor?

What kind of video card do you have? Do you have software in your startup sequence that comes with your video card that is loading? Is there are control panel app for your video card?

see this is what I think too

everyone tells me no always trust CS4 and yes you msut have the software that profiled your machine use the LUT loader at start but i could swear this results in double profiling under Vista (or at least vista plus ATI graphics card that supports LUT and matrix)

and i see so many people talk abotu this and then decided that colro-managed stuff understands things so it must be correct and other stuff is wrong but I think it is the color-managed stuff that is wrong when you use a LUT loader since I think you get double profiling and I think tons of people are going around making messed up images since nobody seems to have looked into it enough to start thinking that it actualy might be double profiling.

i admit i may well be wrong though

[skibum4]

I'm on Vista Business 32-bit. It's a laptop, but the screen is an S-IPS lcd. Video card is an ATI V5200. The ATI control center is currently set to deactivate ATI color controls and I don't have adobe gamma in the startup.

tell me what happens when you do this:

bootup machine as normal (make sure you have the profile your profiling software made assigned to the monitor under vista color management!) with your LUT loading loading the profiel globally to the video card and then AFTER that has finished go to vista colro management again and uncheck the box (or maybe check it) whatever it is that makes it do the remove these settings from current monitor to get rid of the association of the profile with the monitor and THEN load CS2 (oh and make sure just for this simple test to have earlier set CS2 workspace to sRGB and to have turned off proofing) and then load an sRGB jpg int CS2 and tell how it looks then....

doing this, and only doing this, appears to make non-aware programs show sRGB files correctly AND colro-aware programs show sRGB as well as any other sort of colorspaced file show correctly. Anything else I do either makes non-aware stuff show uncalibrated colors and aware stuff work great or aware stuff appear to be using double monitor profiling and unaware stuff show sRGB stuff correctly. (well also I found that setting proof to my monitor also works, but just for CS4 and at that ONYL if the workspace is sRGB but not everyone really necessarily wants the workspae to be sRGB so i don't find this a great solution since I like to use ProPhotRGB as the workspace and many people like to use AdobeRGB)

[arodney]

for sRGB images though shouldn't the LUT and profile loader on startup be all you need? It seems strange to me that IE or irfanview show simple sRGB profile jpgs differently than say CS4.

IE and that other app (which I don't use) are not color managed. That's key. 255G/0R/0B is not the same color in sRGB as it is in Adobe RGB (1998). Non ICC aware applications don't know that since they have no idea what sRGB nor Adobe RGB nor any other color space is (they are ICC unaware). All they "see" is 255G/R0/B0 and send that to the screen (though the graphic system).

There's more to all this than just having a display profile (does the application know what to do with this profile? In non ICC aware applications, not at all).


When the LUT table gets loaded in it does very much change how pictures look in IE and irfanview (grayscales, etc. everything look much better and gamma along the whole ramp changes, etc.) so shouldn't sRGB pics look fine in them since isn't the LUT supposed to matching sRGB to my monitor?

Nope, not enough info to do the full job. What's missing is the Display Using Monitor Compensation architecture in ICC aware applications which looks at the data from the document taking account of its color space and building a document specific preview conversion for that space WITH the display profile. IE and that other app simply don't do that.

And isn't the whole point of the loading up the profile to make the gamma and white point and color balance of my monitor closer to the exact sRGB

Not at all! And you don't have an sRGB display unless you're running a circa 1993 CRT with P22 Phosphors . The sRGB color space is a theoretical color space based on an emissive display who's specifications (down to the ambient light this theoretical display resides in), are a set of simple mathematical spec's (gamma, white point and primary chromaticity of the phosphors in this case).

[skibum4]

IE and that other app (which I don't use) are not color managed. That's key. 255G/0R/0B is not the same color in sRGB as it is in Adobe RGB (1998). Non ICC aware applications don't know that since they have no idea what sRGB nor Adobe RGB nor any other color space is (they are ICC unaware). All they "see" is 255G/R0/B0 and send that to the screen (though the graphic system).

There's more to all this than just having a display profile (does the application know what to do with this profile? In non ICC aware applications, not at all).



Nope, not enough info to do the full job. What's missing is the Display Using Monitor Compensation architecture in ICC aware applications which looks at the data from the document taking account of its color space and building a document specific preview conversion for that space WITH the display profile. IE and that other app simply don't do that.



Not at all! And you don't have an sRGB display unless you're running a circa 1993 CRT with P22 Phosphors . The sRGB color space is a theoretical color space based on an emissive display who's specifications (down to the ambient light this theoretical display resides in), are a set of simple mathematical spec's (gamma, white point and primary chromaticity of the phosphors in this case).


yeah i know 0,34,56 in sRGB is not the same as 0,34,56 in AdobeRGB and unless my LCD is exactly sRGB then 0,34,56 RGB on it is not likely to be 0,34,56 in sRGB.

that is not the issue

i'm not expecting non-color managed stuff to ever display adobergb or prophoto or any of that stuff properly

and i'm not expecting my monitor, BEFORE profiling, to display sRGB properly, but shouldn't loading the LUT and/or matrix conversion into my gfx card then make my vid card adjust anything sent to it as sRGB to display properly on my monitor?

and the problem is that color-managed stuff on windows vista appears to convert whatever profile the pic is in to Monitor Profile but when the LUT is loaded into the video card it is already doing an sRGB->Monitor Profile so it is getting the output that has already been converted to monitor profile and then treating that as sRGB and then converting it again, at least this is what it really, really looks like is happening.

[arodney]

and i'm not expecting my monitor, BEFORE profiling, to display sRGB properly, but shouldn't loading the LUT and/or matrix conversion into my gfx card then make my vid card adjust anything sent to it as sRGB to display properly on my monitor?

No. Because the application doesn't know the document is in sRGB or for that matter what sRGB is.

Color managed application can only be color managed with certain info:

1. What are the RGB numbers? (lets stick to one color model although it doesn't matter).
2. What are the scale of the numbers (what color space)?
3. What's the color space of the display?

Convert sRGB to display RGB for preview purposes (Display Using Monitor Compensation).

[jfriend]

yeah i know 0,34,56 in sRGB is not the same as 0,34,56 in AdobeRGB and unless my LCD is exactly sRGB then 0,34,56 RGB on it is not likely to be 0,34,56 in sRGB.

that is not the issue

i'm not expecting non-color managed stuff to ever display adobergb or prophoto or any of that stuff properly

and i'm not expecting my monitor, BEFORE profiling, to display sRGB properly, but shouldn't loading the LUT and/or matrix conversion into my gfx card then make my vid card adjust anything sent to it as sRGB to display properly on my monitor?

and the problem is that color-managed stuff on windows vista appears to convert whatever profile the pic is in to Monitor Profile but when the LUT is loaded into the video card it is already doing an sRGB->Monitor Profile so it is getting the output that has already been converted to monitor profile and then treating that as sRGB and then converting it again, at least this is what it really, really looks like is happening.

I'm piping in here only because I went through the same process you are. When I first started calibrating my monitor, I thought that calibrating my monitor meant that it was going to make my monitor work like sRGB and therefore even non-color-managed apps should work OK if given sRGB input. As it turns out, that is just not the case. Let me repeat, that is just not the case and you shouldn't be trying to make it be that with LUT loaders and all that.

The calibration software may tweak some parameters on the monitor, but particularly with LCDs, it is NOT trying to change the monitor to simulate sRGB. It's main purpose in life is to create the monitor profile which describes how the monitor behaves, it does not change the monitor's behavior for non-color-managed apps.

The monitor profile can then be read by color-managed software to accurately display colors on the monitor. You HAVE to give up the notion that non-color-managed software is going to do anything accurate for you. It just isn't. With Firefox3 and Safari both having color management capabilities now, you can pretty much give up using or relying on non-color-managed software.

Further, given that many new LCD monitors are capable of displaying a color gamut that is greater than sRGB, if you somehow made the monitor into an sRGB monitor, you'd be throwing away all that new display capability.

So, anyway I got a system that works properly when I just stopped relying on any color display from non-color-managed apps. It will be off. Ironically with today's wide gamut monitors, it's off even more than it used to be. But, more and more apps are color-managed so just stick with the color-managed world and don't try to force your monitor to be like sRGB and you can make it all work fine.

[Anthony]

I'm piping in here only because I went through the same process you are. When I first started calibrating my monitor, I thought that calibrating my monitor meant that it was going to make my monitor work like sRGB and therefore even non-color-managed apps should work OK if given sRGB input. As it turns out, that is just not the case. Let me repeat, that is just not the case and you shouldn't be trying to make it be that with LUT loaders and all that.

[..]


John, another good explanation, which if read *before* and in conjunction with, Andrew's detailed information in this thread, really answers all the points raised.

Anthony.

[skibum4]

yeah i know 0,34,56 in sRGB is not the same as 0,34,56 in AdobeRGB and unless my LCD is exactly sRGB then 0,34,56 RGB on it is not likely to be 0,34,56 in sRGB.

that is not the issue

i'm not expecting non-color managed stuff to ever display adobergb or prophoto or any of that stuff properly

and i'm not expecting my monitor, BEFORE profiling, to display sRGB properly, but shouldn't loading the LUT and/or matrix conversion into my gfx card then make my vid card adjust anything sent to it as sRGB to display properly on my monitor?

and the problem is that color-managed stuff on windows vista appears to convert whatever profile the pic is in to Monitor Profile but when the LUT is loaded into the video card it is already doing an sRGB->Monitor Profile so it is getting the output that has already been converted to monitor profile and then treating that as sRGB and then converting it again, at least this is what it really, really looks like is happening.

ok so everything i had been hinting at was WRONG :)

i had gotten led astray by an almost unique coincidence where the picture I was examining happend by pure chance to have waht the LUT did to it and what the monitor profile each seperately be nearly the same and by the fact that the monitor profile for my monitor seems to not be quite correct in the second stage.

but for those who have run into similar confusions
and since there is tons of variously conflicting stuff posted all over the net much wrong to varying degrees and to make up the mess of wrong stuff I myself posted earlier....

to sum up:
Monitor Profiles have two parts:
part1:
LUT loaders only set things so that gamma and color balance for a certain selected color temperature are displayed properly and DO NOT apply a transformation to the monitor's color space. Monitors vary in how closey they can be adusted by monitors knobs to exact sRGB color space (ones that have a full CMS can often be adjusted exceedinyl close, those without a full CMS usually have the green primary somewhat off and sometimes the primary intensities off as well, usually too saturated and hot)
part 2:
there is a second part to a monitor profile, the first part gets loaded into teh LUT above (and MUST be loaded for the rest of the monitor profile to make sense, when peopel say to remove AdobeGamma and LUT loaders they only mean to remove ones that did not come with your color profiling package you still MUST run the one that your profiling package made or if you use a different once insure that it loads the proper profile), but the second part is the transformation into the monitor color space. It accounts for the actual position of the monitor gamut's primary locations and intensities. Some color profiling programs onyl correct for this and just apply a simple matrix transform others at least appear to do more and perhaps use some sort of long color table to also try to adjust for an non-uniformities within the monitor's gamut (not sure if this is true, but it seems like some more advanced progams such as coloreyes do produce this a more advanced type of monitor profile that goes this extra step).
-----------------------------------------------------------
Programs for viewing:
there appear to be four sorts
1. completely non-color managed
these WILL take advantage of what the LUT loader did so they will at least display pics with the more or less proper gamma setting and color balance temperature (but only if you did make sure to have the LUT loader load the correct profile for your monitor at the start). They will not transform into your monitor's actual color space and pay no attention to teh second part of the color profile mentioned above so they will NOT display pictures entirely correctly on your monitor unless it has a full CMS and was carefull adjusted internally. But sRGB pics should still look at least vaguely and perhaps very, very closely to how they should. Pictures in other formats such as adobergb or prophotorgb will be displayed COMPLETELY wrong since these programs do no do a photo color space to sRGB transform (not a photo color space to monitor color space transform)
example programs appear to be the windows desktop and Internet Explorer
2. partially managed type A
these know how read the ICC profile in a PHOTO so they can adjust for photos using adobeRGB or pro2photo or whatnot, however they convert them to sRGB colorspace and then send to the graphics card they DO NOT know how to read the monitor profile and convert to monitor color space. But at least all pictures should look within some at least semi-reasonable ballpark.
an example is irfanview (IF you turn on color management)
3. partially managed type B
they onyl understand sRGB pics but can transform them to monitor color space
all pics viewed with these that are in sRGB format will look perfect and exactly match CS4 and the like but anything in a different format such as adobergb, etc. will be wrong
4. fully color managed programs
these read the ICC profile in pictures and so can properly work with a picture in any format that it knows how to convert from and as well they also understand monitor profiles so they can convert not just to sRGB for viewing but to the monitor's actual color gamut (note that if the profiling was done with a program that creates a LUT for a gfx card you must first load that in for this to work properly)
basically anything you toss at them should be displayed as correctly as the quality of your monitor and the monitor profile and LUT make possible, everything should look extremely much as it should in all cases
a few examples are CS4, CS3, CS2, Photomechanic
some possible other examples are perhaps firefox if you go in and set all of things that need to be set properly, maybe safari (but apparently there are many ways in which it can fail ro fail if things are not all set correctly and it's not to be trusted, although the comment i read on this was from soem while back and may no longer apply)
----------------------------------------
wrap up:
the dual-stage monitor profiles explain why loading the LUT in is NOT enough for proper colors
the four types of program awareness explain why even if you do have the proper profile loaded into the LUT various programs will displays pictures differently, EVEN pictures just using the plain old simple sRGB that all programs understand.

----------------------------------------------------
final note: on profiles gone bad:
sometimes profiles are not made correctly and if the first stage goes wrong the monitor will not measure to the proper gamma and/or colro temperature and if the second stage goes wrong (as appears to have happened to me a bit, whcih further led to my initial confusion) the monitor gamut will not be corrected for entirely properly and this means anything from you'd have been better off without using it and colro managed programs of type 3 and 4 will actually do worse than type 2 or they will do better but not as much better as they could.
--

[arodney]

there appear to be four sorts


Nope, only two. Color managed or non color managed.

[skibum4]

Nope, only two. Color managed or non color managed.

thanks

well irfanview says that it does look at picture ICC profiles and DOES use them (if you tell it to in preferences) but it merely converts them to sRGB and DOES NOT use the monitor profile. So this does seem to be program of type 2. It may be the only such program. As for type 3, you may be correct, I haven't found any examples yet.

So I think correct would be most programs are not color managed, some such as CS4 and Firefox with the setting set are, and then there is one oddball, irfanview, that can translate various image profiles into sRGB but that is as far as it goes (I haven't tried this yet to see if it is true or not but this is what they claim on the irfanview site).

anyway, yeah, you are all correct, the last three psots there above mine explain it all too, I was very wrong to think simply having the the LUT loader load in the LUT did everything.

But I still have a weird issue and it is next the next message:

[skibum4]

my remaining issue appears to be with how the color-managed apps are handling the monitor profile or some setting I have wrong in vista or how my profiling program is making the second stage of the monitor profile:

So I have my profiling program's LUT loader load the LUT at the start.

If I go and use something like Color HFCR to measure the gamma curve and the gray-scale color temp scale these now look VERY good, far better than without the LUT loaded. So the profiler did a good job in making the LUT and it is being loaded in properly. All is good with that.

What is not good is what happens in my color-managed apps since if I use color hfcr to measure the primaries for say sRGB as displayed in say CS4 they do not line up well on the CIE plot to the sRGB primaries and actually line up much less well than they do if I display them in a non-color managed app. Also the blue primary, even visually, looks rather far towards purple indeed in CS4 (and interestingly is a similar color to the water in the tank with the jellyfish the pic that got me noticing all this to begin with). I know the sRGB blue primary DOES look a bit toward purple compared to what we are used to thinking of as blue, but this looks totally out and out purple.


Anyway viewing colors bars in various programs and measuring them (using ColorEyes probe and colorhcfr program to make these plots) and all of these assume the LUT was loaded (not that it really affects the primaries anyway though):

this is how they measured using a non-monitor profile aware program (not too bad really):
http://skibum4.smugmug.com/photos/453863692_EdbhP-O.jpg

this is how they measured using the color-aware photomechanic with management turned on (worse):
http://skibum4.smugmug.com/photos/453863668_fc8A9-O.jpg

this is how they measured using the color-aware CS4 (even worse still):
http://skibum4.smugmug.com/photos/453863686_uqzEF-O.jpg

CS4 had the workingspace set to sRGB and no-proofing (if I turn on proofing and set target as my monitor then they look like what the non-aware program did, more or less, which I guess makes sense).

and here is one set of color bars i used in reduced size for easy posting (i also used some full screen per color ones froma different source and the results are more or less the same as above):
http://skibum4.smugmug.com/photos/453863630_vYMsc-O.jpg

so what is going on here?
is the profiler making a bad second stage profile?
do i have something set wrong in vista?
should i not be using these as the color bars?

thanks

[marlinspike]

The monitor profile can then be read by color-managed software to accurately display colors on the monitor. You HAVE to give up the notion that non-color-managed software is going to do anything accurate for you. It just isn't.

See, I get that, but the question is why is it actually being accurate and the color-managed stuff being grossly inaccurate, and I think the answer is in this double-proofing stuff, if only I could figure out how to make it stop.

[skibum4]

See, I get that, but the question is why is it actually being accurate and the color-managed stuff being grossly inaccurate, and I think the answer is in this double-proofing stuff, if only I could figure out how to make it stop.

i'm not so sure it is double profiling anymore.

my jellyfish pick made it seem exactly like it was that, but now looking at other pictures it does not seem to be the case (at least for me) since I see that the LUT load alone or the monitor profile in a color aware program alone, for almost all pics, make very different adjustments and the monitor profile adjustments are usually much smaller in degree (that one picture I had was a weird coincidence it seems where both those things appeared to do the same thing to the same degree).

but in the end, I do seem to be getting the color-aware programs actually making the colors worse (and I think not matching my prints as well but i need to check again to be sure now). They sure do seem to give a big purple push on the blue primary for one. I can't figure out why. Also note that on the CIE plots you can even see the purple push on the blue primary.

Hmm I didn't sell of my spyder yet, when I get back to my other place I can do a profile with the spyder software and see what happens. That might help give some more clues.

[jfriend]

See, I get that, but the question is why is it actually being accurate and the color-managed stuff being grossly inaccurate, and I think the answer is in this double-proofing stuff, if only I could figure out how to make it stop.

My guess, just a guess because I know nothing about your software, is that the monitor profile was created assuming no LUT load. Thus, when you do the LUT load and then the color-managed software comes along and attempts to manage for the monitor profile, you get double correction and things look horrible. One could test this out by not doing the LUT load and seeing if the color-managed software looks accurate. There are some situations where you should not do a LUT load as that just restricts what the monitor can produce and can generate banding and other undesirable characteristics. There are cases where you will get the best performance in color-managed apps without LUT load.

If you insist on LUT load, then you should load the LUT and then profile your monitor with the LUT loaded so you create a monitor profile that describes your monitor when the LUT is loaded. Then, the color-managed apps can correct to the right thing.

[skibum4]

My guess, just a guess because I know nothing about your software, is that the monitor profile was created assuming no LUT load. Thus, when you do the LUT load and then the color-managed software comes along and attempts to manage for the monitor profile, you get double correction and things look horrible. One could test this out by not doing the LUT load and seeing if the color-managed software looks accurate. There are some situations where you should not do a LUT load as that just restricts what the monitor can produce and can generate banding and other undesirable characteristics. There are cases where you will get the best performance in color-managed apps without LUT load.

If you insist on LUT load, then you should load the LUT and then profile your monitor with the LUT loaded so you create a monitor profile that describes your monitor when the LUT is loaded. Then, the color-managed apps can correct to the right thing.



i decided this time to do no pre-monitor calibration jsut leave all controls at default and run the profiler, this time the oclor-managed stuff makes things look better than the non-color managed (although now my non-color managed stuff is worse with the primaries and gamut arrggggh). maybe when i adjusted the primaries as much as i could with the controls it made it hard on my profiler for some weird reason?

but now i have another issue, photmechanic and CS4 are both color managed but do not show things the same way, close but ot the same, with PM looking a bit better to my eye.

but worst is the CS4 black-point compensation mess i seem to have.
if I have it turned on even painting 0,0,0 looks a little dark gray and not black and the darkest parts of my images are low in contrast and a little washed out and prints come out a tad crushed.
But if I turn BPC off, while i get better contrast and what looks like as black as my monitor can go for 0,0,0, the darks now look a little darker than photmechanic shows them when it has color management disabled nevermind when it has it on, although it seems a bit better comprise.

It seems to me that somethign weird is going on with how CS4 handles deep shadows, doesn't seem as accurate or nice as with photomechanic and worse when I leave BPC on (which people seem to say to do) my blacks are gone and low level contrast is gone and when I turn it on things seems a touch too dark then.

I'm not even sure which is way it should really be, i'm guessing PM is most accurate with color management on and then CS4 with BPC turned off (since BPC on just seems weird, how can a pitch black part of an image not make my monitor show it's darkest color possible??)

what is going on with CS4!

maybe smething about the monitor profile confused it? maybe it didnt set a blackpoint or set it in some way that confuses CS4??

or CS4 was somehow not expecting the LUT loader to brighten up darks as much as it did??


no, it's not that, they fixed that, my LUT stays loaded.

I did decide to do the profile again, this time with NO pre-calibration of my monitor settings, left them back at default, since I heard this works better with some monitors and some profilers.
this time the color-managed programs have the better gamut than the non-colro managed (however it seems a shame since before I could get a decent match for the non-color managed, argggh, and now with the monitor at default the primaries are farther off for my non-managed stuff now)
anyway, maybe i can hit some ideal combo of monitor adjustments for that stuff and for color-managed at once.
but more immediate concern is that i notice various color-managed programs DO NOT displays thigns quite the same, and actually photomechanic seems to make things measure out better than CS4 by a little bit.
of even greater concern is that when I leave "blackpoint compensation checked" in CS4 which I thoguht you were, then it seems to wash things out in the dark colors, even if i paint with 0,0,0 it doesn't seem as dark as monitor is capable of and things seems slightly lighter and less contrasy and less detailed than in photomechanic and then if I turn of BPC then I seem to get rich black back but now it makes the shadows even darker than photomechanic does, about as dark as photomechanic does when I have color-management turned off in that program.
It seems to me that photomechanic handels monitor profiles, at least the ones I making so far and maybe mine is not setting certain things the way CS4 expects as I have it set now, better. Colors are slightly truer overall and more importantly it seems to get rich blacks with a bit more detail and contrast and no washing out something I can't get CS4 to do.
If I use BPC CS4 makes deeps look lighter than they are then my prints get crushed a bit so I don't think I can use BPC for screen editing using CS4??
If I turn it off it does make thing a touch more crushed than photomechanic though.
Not sure what to trust, but not sure i'm liking or trusting what CS4 does entirely.... what photomechanic does in color managed mode seems much more reasonable to me and matches prints better I think.....
argghh...

[skibum4]

i think there may be a bug in CS4 if you have scene-referred profiles option check (it'sjust below BPC) it seems to make the darks fade out almost like it is showing 16-235 video instead of 0-255?? it did say something abotu make sure to check this to maintain compatability with their video editors.

many sites say to keep this check, i wonder if they are wrong,or if the function is buggedsince it doesnt always seem to ruin my blacks....

this might be my final main issue...

[arodney]

t
well irfanview says that it does look at picture ICC profiles and DOES use them (if you tell it to in preferences) but it merely converts them to sRGB and DOES NOT use the monitor profile. So this does seem to be program of type 2.

No, there is no type 2. That program may say its kind of, sort of, maybe, is sometimes ICC aware, if its NOT using the display profile, its NOT an ICC aware application. Does it match the previews in Photoshop?

[arodney]

i think there may be a bug in CS4 if you have scene-referred profiles option check (it'sjust below BPC) it seems to make the darks fade out almost like it is showing 16-235 video instead of 0-255??

Scene referred profiles option?

[skibum4]

No, there is no type 2. That program may say its kind of, sort of, maybe, is sometimes ICC aware, if its NOT using the display profile, its NOT an ICC aware application. Does it match the previews in Photoshop?

well, if it is aware of ICC profiles in pics but not of ICC monitor profiles it is certainly different from programs that are aware of both or neither, whatever you want to call it. anyway, it's a minor point either way in the end.

[skibum4]

Scene referred profiles option?

http://skibum4.smugmug.com/photos/454339803_CzfJX-X3.jpg

where it says "Compensate for Scene-referred Profiles"
not quite sure what is going on yet, but sometimes when that has been toggled things star acting weird and sometimes that alone fades my blacks or sometimes BPC alone does or only in combo (and all with the same picture loaded the same way).

[skibum4]

Actually it may not be that toggle alone but either BPC or that alone or both seem to be acting in weird ways. Sometimes the same combinations fades my blacks and sometimes it doesn't and when I use the preview toggle sometimes toggling it on and off show the difference and sometimes it doesn't (and this is with same picture loaded in the same way).

No clue what is going, not sure if some bug in CS4 or my video card drivers and how it interacts with CS4, etc.
but it is pretty weird that haveing the same combination of those three toggles doesn't always act the same way as I play around with them.....

PhotoMechanic doesn't jump around it always looks the same in the deep shadows for each of the two settings it allows (color-managed or non-colormanaged).

Has anyone heard anything about BPC bugs in CS4? or in combo with certain video cards or types of monitor profiles??

OK here is the story:
CS4 with BPC turned on DOES NOT match any of my other color-aware programs, it makes shadows brighter, fades out black and shows less contrast in images compared to Firefox 3 with ICC turned on or Photomechanic with color management turned on or Lightroom 2. I must have BPC turned OFF in CS4 to get black showing as black on my monitor, however, when I turn BPC off the shadows look more crushed than they do in my other color-managed programs, about as crushed as they do in my non-color managed programs. Oddly, firefox, photomechanic and LR have slight differences too, with PM showing a trace more crush than the other, but they are still pretty much in the same ballpark and FF3 and LR2 may be exactly the same even.

So is BPC totally bugged in CS4 or what??
When it is on (and most sites seem to claim you should have it on) the pics do not look right at all and when it is off they still don't match other color-managed programs since now the shadows are as dark as in non-color aware programs.

[arodney]

http://skibum4.smugmug.com/photos/454339803_CzfJX-X3.jpg

where it says "Compensate for Scene-referred Profiles"
not quite sure what is going on yet, but sometimes when that has been toggled things star acting weird and sometimes that alone fades my blacks or sometimes BPC alone does or only in combo (and all with the same picture loaded the same way).

That's only for Photoshop to video (After Effects) and back conversions. Stay away from everything in the advanced (hurt me) section of this dialog!

That section ONLY comes into play when you use Mode Change.

BPC is a unique Adobe option in their CMM (ACE). It should always be on.

[skibum4]

That's only for Photoshop to video (After Effects) and back conversions. Stay away from everything in the advanced (hurt me) section of this dialog!

That section ONLY comes into play when you use Mode Change.

BPC is a unique Adobe option in their CMM (ACE). It should always be on.

the problem is when I turn it (BPC) on in CS4 black disappears, I can paint with 0,0,0 and it looks like dark gray, not even close to what black looks liek on my monitor, all the fine details in my pics in the shadows loose contrast and it just turns into a gray mush.

And I don't believe this is how it should look since it DOES NOT look like this in photomechanic, in Firefox3, or in Lightroom.

here is one example (Lightroom 2 on the left and CS4 on the right (sRGB working space, BPC on, Adobe ACE, Relative Colormetric):
http://skibum4.smugmug.com/photos/454548517_etbvz-XL.jpg

THE only way I can make images in CS4 look anything like they do in photomechanic, firefox3 (with CM on), or lightroom2 is to use Microsoft ICM instead of Adobe ACE and to turn BPC off. Th every things that have always been things to NOT do in CS. Thinking there must be a bug somewhere or some bad handshaking between CS4 and something my monitor profiling software says abotu the balckpoint or with my gfx drivers or something??

does anyone else get CS4 giving them different blackpoints and shadow detail and contrast them with other programs such as lightroom??

[marlinspike]

If you insist on LUT load, then you should load the LUT and then profile your monitor with the LUT loaded so you create a monitor profile that describes your monitor when the LUT is loaded. Then, the color-managed apps can correct to the right thing.

I insist on nothing other than that I am confused, lol. How do I not do the LUT load?

[jfriend]

I insist on nothing other than that I am confused, lol. How do I not do the LUT load? I don't know anything about the software you're using so it's hard for me to make any suggestions on how to change it. If I were you, I'd poor over the documentation for your profiling software, plow through all the nooks and crannies of their UI and then hit Google with "LUT load" and something that uniquely identifies the brand of profiling software you're running and see what turns up.

[skibum4]

I insist on nothing other than that I am confused, lol. How do I not do the LUT load?

from what happened to me, i'm more thinking that the second stage of your monitor profile turned out less than ideal, tricked by teh way you had the moinitor knobs adjusted and it happaned to make things look a little worse evenif the primaries technically might measure a little better (when i used the monitor controls on my monitor to get the xy in the CIE color chart for the primaries as close as I could the Y was somewhat off and it seemed my profile tried to correct for the Y to such a degree that is didnt care that xy got pushed into worse values, so it imrpoved saturation (not overblown) and intensity (not overdone) but made the actual looks of the colors worse).

maybe you coudl redo the profile using default monitor settings (and if that is how it was done, then maybe instead try pre-calibration with the monitor knobs)


do you have the problem where CS makes shadows look faded out and prevents black from showing (makes it dull gray) too??



anyway to prevent the LUT from loading go into msconfig and go to the startup section and try to find a program calls itself a gamma or LUT loader or is associated with the name of the profiling software you use and deselect it so it doesn't load at start.

alternately:
if you use nvidia go color controls and mess around the sliders and then reset them back before exiting (i froget which color screen, but doing this on one of them resets the LUT to original state)

if you have ATI (i think you said you did) then just go to the color panel and hit the button to "reactivate ATI color controls" and this will dump out whatever LUT your LUT loader may have loaded.

[skibum4]

SOLUTION TO ALL PROBLEMS FOUND:

(some of this might help you, marlinspike, espeicially the part about trying to first adjust the monitor primaries and stuff, on my monitor leaving ALL controls at factory defauly, aside from swithc color temp from normal to cool1, makes the profile work best and now my color aware programs look subtly BETTER than my non-aware programs not worse)

first yes profile do come in two parts, the LUT fixes gamma and color temp along the grayscale ramp but only color aware programs can use the rest of the minor profile for further fine tuning.

CS4 doesn't work well if you use a relative blackpoint choice during calibration it works fine if you set absolute blackpoint during calibration.

for whatever reason other color aware programs have no problem with this.

it may even be that CS4 was the only one handling it correctly, not sure, either that or CS4 has a bug with this form of monitor profile


any success at last:

tips if you use coloreyes use absolute blackpoint
and if you use a samsung 244T monitor DO NOT TOUCH the primary/secondary hue/saturation controls they ONLY MESS THINGS UP and if you adjust them to make the monitor get better locations it makes the gamut have uneveness within it and messes things up and makes at least some calibration programs produce a worse profile, once that actually makes color-aware programs give worse results than non-aware

anyway i finally have all color-aware programs working great (better than non-color aware) and my non-color aware working ok (once the LUT is loaded)


i highly recommend people download colorhcfr to verify the sRGB primaries within CS4 and see if they measure better than from the program itself (non-color aware) to see if you made an ideal profile or not.

also check that deep shadows and blacks look good in CS4 and the same as LR2 or the like.
--

[arodney]

the problem is when I turn it (BPC) on in CS4 black disappears, I can paint with 0,0,0 and it looks like dark gray, not even close to what black looks liek on my monitor, all the fine details in my pics in the shadows loose contrast and it just turns into a gray mush.

You need to read up what BPC is, when it kicks in and why. Its doing exactly what its supposed to do and will only kick in when doing working space to output color space conversions. Its there to FIX a bug in old ICC profiles and older CMMs. You're moving down a rabbit hole from your original ideas about display calibration and ICC aware applications confusing yourself in the process. If you want to discuss BPC, you should start a new post. There IS no bug in CS4 (or 3, or 2 down to Photoshop 5 when ACE was introduced) in terms of BPC!

first yes profile do come in two parts, the LUT fixes gamma and color temp along the grayscale ramp but only color aware programs can use the rest of the minor profile for further fine tuning.


It doesn't "fix" anything and in fact in ICC aware applications, doesn't really make much of a difference (thanks again to Display Using Monitor Compensation). You can (and many do) calibrate with a LUT that does absolutely nothing! That's what a native gamma and white point setting in some calibration software would produce (they even SHOW you the LUT so you can see its a straight line, NULL).

CS4 doesn't work well if you use a relative blackpoint choice during calibration it works fine if you set absolute blackpoint during calibration.


It works fine and as designed. You just don't like the rendering. And " use a relative blackpoint choice during calibration " what the heck does that mean?

for whatever reason other color aware programs have no problem with this.


Its not a problem, they can't access the ACE CMM thus they can't access BPC.

it may even be that CS4 was the only one handling it correctly, not sure, either that or CS4 has a bug with this form of monitor profile


What profiles are using to make such a statement? You're soft proofing with printer profile or just converting working space to working space?

[arodney]

And I don't believe this is how it should look since it DOES NOT look like this in photomechanic, in Firefox3, or in Lightroom.

NONE (at least Firefox and Lightroom) soft proof to output profiles using BPC (because there's no need).

BPC is for OUTPUT profile conversions. In fact, by using the Customize Soft Proof option, simulate ink black, you're now turning off BPC, turning on an Absolute Colorimetric intent. The preview goes "soft and mushy" as the black point is shown to you as an Absolute convention, without BPC!

•Simulate Paper Color and Simulate Black Ink Off: This produces the
relative colorimetric intent with Black Point Compensation.
•Simulate Black Ink: This produces the relative colorimetric intent
without Black Point Compensation.
•Simulate Paper Color: This produces the absolute colorimetric intent
(no Black Point Compensation).

[marlinspike]

I have an ATI card and my monitor, being a laptop (albeit an S-IPS screen), does not have color knobs.

[marlinspike]

w00t, I figured it out! Ok, so I had originally "use my settings for this device" checked in the windows color management and the "profile associated with this device" was the profile my monitor calibrator made. Then I tried unchecking use my settings and some other profile that camewith the computer went into that associated profile slot.

Now I tried re-checking "use my settings for this device" and deleted the profile from the "profile associated with this device" so that there is NO profile associated in the windows color management. Now PS CS works! :barb

Also, all the various non-color-calibrated things seem to be showing accurate colors as well (IE, windows photo gallery, my desktop)

[skibum4]

w00t, I figured it out! Ok, so I had originally "use my settings for this device" checked in the windows color management and the "profile associated with this device" was the profile my monitor calibrator made. Then I tried unchecking use my settings and some other profile that camewith the computer went into that associated profile slot.

Now I tried re-checking "use my settings for this device" and deleted the profile from the "profile associated with this device" so that there is NO profile associated in the windows color management. Now PS CS works! :barb

Also, all the various non-color-calibrated things seem to be showing accurate colors as well (IE, windows photo gallery, my desktop)

it sounds like you just deleted the monitor profile and made color-aware stuff work the same way as non-color aware.

sounds like maybe your calibrator didn't make a great profile (as happened to me at first).

[skibum4]

NONE (at least Firefox and Lightroom) soft proof to output profiles using BPC (because there's no need).

BPC is for OUTPUT profile conversions. In fact, by using the Customize Soft Proof option, simulate ink black, you're now turning off BPC, turning on an Absolute Colorimetric intent. The preview goes "soft and mushy" as the black point is shown to you as an Absolute convention, without BPC!

•Simulate Paper Color and Simulate Black Ink Off: This produces the
relative colorimetric intent with Black Point Compensation.
•Simulate Black Ink: This produces the relative colorimetric intent
without Black Point Compensation.
•Simulate Paper Color: This produces the absolute colorimetric intent
(no Black Point Compensation).


are you saying BPC should not make a difference for screen display at least if soft proofing is turned off?

it didnt seem to be working that way, since i had proofing off and it was showing up on screen (when I used a profile made with relative blackpoint option checked, it no longer does now that I used teh absolute option instead in my calibration software)

anyway i do need to read up more on BPC

[skibum4]

It doesn't "fix" anything and in fact in ICC aware applications, doesn't really make much of a difference (thanks again to Display Using Monitor Compensation). You can (and many do) calibrate with a LUT that does absolutely nothing! That's what a native gamma and white point setting in some calibration software would produce (they even SHOW you the LUT so you can see its a straight line, NULL).


Using a LUT table fixes a monitor that is not calibrated well. How is fixing that not fixing anything? My monitor can NOT be adjusted to a consistent color temperature up and down the gray scale using the limited controls it has nor can it come close to a nice 2.2 gamma ramp. Using the LUT fixes those issues no matter what I use to display things.

Whether I load the LUT or not makes a HUGE difference in how all of my programs display things color aware or not. And in any case I'd sure rather have both types looking at least reasonably good instead of having all my non-color aware apps look hideous by skipping the LUT step.

Yeah, but I don't want to calibrate to the native points on my monitor, since they aren't that great looking and not all that close the D65 standard for movies or photography and the gamma curve is like 2.4 (at best, maybe more like 2.5) and kind of bendy.

[skibum4]

It works fine and as designed. You just don't like the rendering. And " use a relative blackpoint choice during calibration " what the heck does that mean?



it's a lot worse than my just not liking the look it gave with my old profile, it made editing totally useless. Nothing matched the output from ANY other program and it doesn't match prints either. It's almost like the look you get when you mess up PC 0-255 vs. 16-235 video levels.

but now that i set the absolute option in my profile creation, CS4 makes stuff look like all the other programs (with the other profile if I had BPC on it looked washed out and not right at all, or at least not right for editing photos, maybe for seeing what the difference between my monitor looks compared to an ideal monitor with no backlight where black is or something and iwth it off the shadows looked too dark and like non-color aware programs).

anyway i should read up more about what the two choices in my profiling program do. It defaulted to relative so I thought that must make the most sense to use, but perhaps not. All I know is CS4 does not handle shadows and blacks in a way that is of any use for editing photos if I set relative. Hopefully using absolute does make the most sense overall and is not messing anything up that i have not noticed yet. I thought i had something about how using relative sets the gamma relative to what your (non-deal) monitor can actually produce for black and so is more sensible to use, but maybe i'm misremembering, or they are wrong, or I didn't interprate it properly. Or maybe CS4 should not be doing what it is doing when it converts to output profiles for monitors?

anyway pics of the choices follow (using absolute makes CS4 render images as other color-aware programs too and makes everything look good):
http://skibum4.smugmug.com/photos/455077162_XLdz3-X3.jpg
http://skibum4.smugmug.com/photos/455077183_45jZB-X3.jpg

[skibum4]

What profiles are using to make such a statement? You're soft proofing with printer profile or just converting working space to working space?

no soft proofing or anything

I had CS4 set to sRGB working space and had my monitor profile associated with my monitor in windows color management and just looking at pictures on the screen loaded into CS4, nothing more.

[marlinspike]

it sounds like you just deleted the monitor profile and made color-aware stuff work the same way as non-color aware.

sounds like maybe your calibrator didn't make a great profile (as happened to me at first).

No, I don't think so, because the colors are spot on, not simply very close to spot on like they were before I calibrated. I think what was happening is the Gretag Macbeth software is loading the profile when I boot (when I start the computer I do see the screen colors suddenly change when the desktop loads) and then by having that selected in windows color management color-aware programs were double profiling.

[skibum4]

No, I don't think so, because the colors are spot on, not simply very close to spot on like they were before I calibrated. I think what was happening is the Gretag Macbeth software is loading the profile when I boot (when I start the computer I do see the screen colors suddenly change when the desktop loads) and then by having that selected in windows color management color-aware programs were double profiling.

that's what i thought too, but it turned out that I just hadn't made a good profile, i put my monitor controls back into the default state (inside the monitor menus) and in my case when I ran the profile now it looks better in color-aware programs.

I verified by using colorhcfr to measure sRGB primary locations that the LUT loading step has no affect on this, but in color aware programs IF I left the monitor profile attached to the monitor within windows color management then the primaries would shift (into numerically better position in both cases, but in visually better in the second case after I redid it, in the first case; the first time it improved the Y component but made the xy (sort of the shade) a bit worse)

[arodney]

but now that i set the absolute option in my profile creation, CS4 makes stuff look like all the other programs (with the other profile if I had BPC on it looked washed out and not right at all, or at least not right for editing photos, maybe for seeing what the difference between my monitor looks compared to an ideal monitor with no backlight where black is or something and iwth it off the shadows looked too dark and like non-color aware programs).

Digging your way deeper down a rabbit hole of misunderstanding color management. Now you're resorting to mucking around with the video card software/driver?

Your on your own. You're simply failing to understand a color management architecture that's worked since Photoshop 5 (circa 1998).

anyway i should read up more about what the two choices in my profiling program do. It defaulted to relative so I thought that must make the most sense to use, but perhaps not.

Defaulted to relative means nothing. All ICC profiles have that colorimetric table. NOT all have other tables (display and working space profiles). But whatever. You seem to have kludged together something that you believe works.

[arodney]

Using a LUT table fixes a monitor that is not calibrated well.

Nonsense (and in your mind).

There is NOTHING to calibrate on a CCFL LCD but the backlight intensity. Unless you're using a high bit, high end display that avoids 8-bit LUTs to the graphic card (which ONLY introduces banding into your display), the best starting point are to "calibrate" to a native gamma and white point (don't alter via the 8-bit LUT these options, just fingerprint them into the display profile and let the Display Using Monitor Compensation architecture of the ICC profile work in (Photoshop) using 21 bit precision. Otherwise, we adjust the high bit LUT inside the panel of the few worthy LCD's out there (Eizo, SpectraView etc).

How is fixing that not fixing anything?

I just told you!

My monitor can NOT be adjusted to a consistent color temperature up and down the gray scale using the limited controls it has nor can it come close to a nice 2.2 gamma ramp. Using the LUT fixes those issues no matter what I use to display things.

No, it doesn't! And unless this display of yours is a CRT, where you DID have physical control over the electronics, leaving again, the LUT nearly linear, you're not fixing anything. You have failed to understand the differences between calibration and profiling and the role of the display profile in ICC aware applications.

Whether I load the LUT or not makes a HUGE difference in how all of my programs display things color aware or not.

Far more with non ICC aware applications which don't preview the color correctly in either event.

IF you muck around with the 8-bit LUT and build a profile versus leaving the LUT linear and building a profile, the net results are the same in ICC aware applications with the banding cropping up because the LUTs are far from linear. Outside ICC aware applications, you still have the banding AND the color is wrong (because you're not using an ICC aware application). So you've double screwed yourself. Feel good?

And in any case I'd sure rather have both types looking at least reasonably good instead of having all my non-color aware apps look hideous by skipping the LUT step.

You mean you are aware that non ICC aware applications don't correctly or consistently preview the numbers correctly AND you prefer inferior quality from the ICC aware applications because you built this unnecessary LUT?

Yeah, but I don't want to calibrate to the native points on my monitor, since they aren't that great looking and not all that close the D65 standard for movies or photography and the gamma curve is like 2.4 (at best, maybe more like 2.5) and kind of bendy.

What makes you think D65 is a standard to calibrate to or that there are not applications for movies and photo's that can't handle color management properly? Bendy? Oh man, that rabbit hole you've dug is a deep one <g>.

[skibum4]

Digging your way deeper down a rabbit hole of misunderstanding color management. Now you're resorting to mucking around with the video card software/driver?

Your on your own. You're simply failing to understand a color management architecture that's worked since Photoshop 5 (circa 1998).



Defaulted to relative means nothing. All ICC profiles have that colorimetric table. NOT all have other tables (display and working space profiles). But whatever. You seem to have kludged together something that you believe works.

who is talking baout graphics card driver settings?

those settings were in my ColorEyes pre-calibration settings page. They were a part of my CALIBRATION program and have nothing at all to do with my graphics card drivers.

And if I set COlorEyes to do a Relative profile then it looks fine in everything aside from CS4 where it looks a mess no matter how I set CS4 BPC (although much worse with BPC on). Now this may even been the truly correct interpretation of a Relative coloreyes profile for all i know, but it is useless for editing and may not even be a correct interpretation.

When I set colorseyes to do an absolute blackpoint calibration then everything looks the same in FF3,LR2,CS4, etc.

[arodney]

And if I set COlorEyes to do a Relative profile then it looks fine in everything aside from CS4 where it looks a mess no matter how I set CS4 BPC (although much worse with BPC on). Now this may even been the truly correct interpretation of a Relative coloreyes profile for all i know, but it is useless for editing and may not even be a correct interpretation.

When I set colorseyes to do an absolute blackpoint calibration then everything looks the same in FF3,LR2,CS4, etc.

Take it all up with coloreyes, got nothing to do with Photoshop's BPC. And all this talk of BPC has nothing to do with the original post.

[skibum4]

>There is NOTHING to calibrate on a CCFL LCD but the backlight intensity. >Unless you're using a high bit, high end display that avoids 8-bit LUTs to >the graphic card (which ONLY introduces banding into your display), the >best starting point are to "calibrate" to a native gamma and white point >(don't alter via the 8-bit LUT these options, just fingerprint them into the >display profile and let the Display Using Monitor Compensation architecture >of the ICC profile work in (Photoshop) using 21 bit precision. Otherwise, >we adjust the high bit LUT inside the panel of the few worthy LCD's out >there (Eizo, SpectraView etc).


and yet their are certified HDTV calibrator who go around and calibrate LCD HDTVs and some sets have a full CMS that lets you move the primaries and secondaries around and on some sets the controls work quite well indeed which is a good thing since blu-ray players and cable boxes and so on don't support monitor profiles. So sometimes calibrating the set is the onyl choice or the best choice (some sets, especially HDTVs) have very good internal calibration systems running at 10,12 even 16bits and adjustments made there introduce far less banding than even using monitor profiles (most sets have 8bit interfaces so even if the monitor profile is odne in 16bits in the end it usually has to get passed through an 8bit barrier although some monitors do handle 10bit interfaces these days).

Granted some monitors, like my Samsung 244T, have a pretty hideous partial CMS that only seems to make the final calibration worse if those controls are touched and I can see it creating sudden transitions and other issues.


anyway onto LUTs, i'm not sure how you can say they can't be said to fix up monitor issues. I can go from something either cooler or warmer than D65 (and D65 is what HDTV/DVD/etc. expects the screen to be at) to vrying degrees at different gray levels and a very funky gamma curve to the proper oclor temp at all gray-scales and nice a smooth gamma curve of 2.2 without weird dips and jumps. My profiling program actually makes 16bit LUTs and I believe most recent graphics have full 16bit LUT support (perhaps i'm wrong and they truncate to 8bit but i thought they did 16bit these days even 4 years i recall reading that most had at leat 10bit LUT support).

Now maybe if you ONLY care about color-managed programs it can be in cases better to not bother with the LUT stage and have every last thing done in the other part of the profile, but then you are tossing out your desktop, non-color managed programs, HDTV playback, DVD/blu-ray playback and forcing them to all look like junk. Using a LUT loaded into your graphics card sure does fix up monitor issues and make all of those programs look WAY better. In my case, I'd rather have it, perhaps, make the color-aware program color profiling come out a tad worse and have all my other stuff look pretty good than get the 100% best possible profile for color-aware programs and be stuck with having DVD/blur-ray/TV/desktop etc. look way off. For a dedicated photoshop only setup maybe that's not the best, but unless the machine is setup in a procesing studio or something is seriosuly devoted to doing nothing much more than 2d still imaging, I don't see how skipping the LUT step is such a great thing to do.

>No, it doesn't! And unless this display of yours is a CRT, where you DID >have physical control over the electronics, leaving again, the LUT nearly >linear, you're not fixing anything. You have failed to understand the >differences between calibration and profiling and the role of the display >profile in ICC aware applications.

but you might have control over the internal LCD monitor engine, which in some cases always for less damaging corrections than external profiling, although certainly not always, the color egine in the 244T is pretty weak in some parts and the primary/secondary saturation and location controls appear to do a terrible job and onyl partial at that. In most modern higher end LCD HDTVs they do a pretty solid job and same for some of the better LCD monitors. Anyway yeah calibrating a monitor is separate from doing stuff with LUTs in gfx cards and making IC prifles.

>Far more with non ICC aware applications which don't preview the color >correctly in either event.

all i can say is I sure as heck am not willing to give up using a LUT to correct for my LCDs native. I get terrible black crush on movies and pictures and a rather wrong color balance that isn't even terribly even up and down the grayscale. Using a LUT inmy gfx card makes things look WAY better. And if i could make a monitor profile look fractionaly better by not using the LUT and purely profiling to the native LCD what can I say, i'd rather give up that minor difference for a huge difference in all my other stuff.



>IF you muck around with the 8-bit LUT and build a profile versus leaving >the LUT linear and building a profile, the net results are the same in ICC >aware applications with the banding cropping up because the LUTs are far >from linear. Outside ICC aware applications, you still have the banding AND >the color is wrong (because you're not using an ICC aware application). >So you've double screwed yourself. Feel good?

no i have not doubly screwed myself. the primaries and secondaries are not that far off from sRGB on my monitor so the gamut isn't THAT different so the lack of the ICC monitor awareness in my other software is nothing compared to how far off they are without the LUT. I mean come on you are saying that watching a blu-ray with my monitors up and down gamma curve (averaging to nearly gamma 2.5) and a color temp jumping around like D70-80 or D50-60 is going to look good?? The difference (using the LUT) between my non-color aware stuff and my color aware stuff is awfully minor in comparison between my non-color aware stuff using LUT and not using LUT. And if my gfx card dies have 16bit LUT support then i'm not sure the diference would even be visually distinguishable.

Now the totally dedicated photo editor who does nothing other than edit photos may prefer no LUT, but I sure don't. I supposed I could create LUTless profile and then unload the LUT and swap to that everytime I go to use CS4 or something. Whether that is worth it or not I'm not sure, maybe. First my profiler does make 16bit LUTs and perhaps my gfx card does utilize 16bits for the LUT and even if not even if the profile is 16bit it still has to pass through the 8bit interface to my particular monitor anyway (my HDTV might support 10bit) although granted if the gfx card is using an 8bit LUT perhaps applying the corrections that that might be a little bit less optimal than to the monitor straight through the card, but since the PC-monitor interface is 8bit in many cases you can still get a little banding from even the ICC profile anyway no matter it's bit depth so that makes the difference a little bit less extreme than it would be.


> AND you prefer inferior quality from the ICC aware applications because >you built this unnecessary LUT?

yeah I do because as I said my non-color managed stuff looks horrible without the LUT and i'd gladly trade that for a tiny change in the quality in color aware programs (although i guess i get into hot-swapping toa lutless profile for tiems when I use color-aware programs and back to the LUT for other stuff).


>What makes you think D65 is a standard to calibrate to or that there are >not applications for movies and photo's that can't handle color >management properly?

point me to a color-managed blu-ray player please.
i'd love for them to exist.
if you know of one please point me to it by all means.
if you know of a color-managed ATSC program that will work with my ATSC tuner card please point me to it.
etc.
I did at least start using FF3 instead of IE7 so I do have that down now at least.

ffdshow and all that can be manipulated into some degree of color awareness but i'm not teribly sure they are any hacks out that allow it to work with the latest copy protections and trying to get it to use various filters from other programs for DTS-MA and so on, it's a real mess, and involves copying the many GB files to HD first, etc. and I don't recall having heard of anyone really getting it working properly for hollywood blu-ray titles yet.
<g>

bendy yeah. if i use my probe and colorhxfr to measure the gamma curve on my monitor as is, it is not a nice smooth curve and dips around from like gamma 2.6 to gamma 2, mostly around gamma 2.4-2.5. that is kind of bendy.

anyway i need to check if my gfx card truly supports a 16bit LUT or not.
i'm pretty sure it does though.

</g>

[arodney]

the primaries and secondaries are not that far off from sRGB on my monitor

Unless you're monitor is a circa 1993 CRT with P22 phosphors (which is what sRGB, a theoretical color space defines), you're way off bud.

[skibum4]

Unless you're monitor is a circa 1993 CRT with P22 phosphors (which is what sRGB, a theoretical color space defines), you're way off bud.

Listen I never said they were exact or that I didn't wish they were closer, and if every last part of my computer could be made color-aware and every external blu-ray player and cable box and directv box could load in ICC profiles it would be awesome but they can't, and the fact is if I measure the primaries and secondaries they are not THAT far off, even if beyond the ideal at least within 3dE.

If I flip color-management off and on I can see the differnce, but for most of the shades it really isn't all that striking, although certainly much preferred to have management working.

Anyway the point is using a LUT makes all my non-color aware stuff look way better and the difference between my color-aware and not color-aware stuff is FAR, FAR, FAR less than between my non-color aware using the LUT or not using the LUT. How much the LUT makes a difference depends upon the set and what they used in the panel and how they tuned the internal engine defaults.

And on some TV's like the samsung 750 with a full CMS (i know it has at least a 10bit engine, maybe 12) you can dial them in on the set really close.

here are screen caps of color space of the 244T, not perfect, but not absolutely horrible, and yeah this doesn't show the Y in the plot and those values are a bit hot in many cases, but the dE chart shows that stuff):
http://skibum4.smugmug.com/photos/455563456_ZFJZD-O.jpg



http://skibum4.smugmug.com/photos/455563494_izJf2-O.jpg

and here is the plot measured from within CS4 (better):
http://skibum4.smugmug.com/photos/455563532_QCFAP-O.jpg


http://skibum4.smugmug.com/photos/455563568_hw8Qc-O.jpg

[arodney]

not perfect, but not absolutely horrible, and yeah this doesn't show the Y in the plot and those values are a bit hot in many cases, but the dE chart shows that stuff)

Going even deeper down that hole, you realize that the deltaE values provided are virtually useless; they are generated by the same device that set the calibration and profile. So you're comparing the accuracy with the same device, not a known, higher grade reference device. And the values provided for the delta, you can specify them (and you are?) or its some preset group of colors the software selected? NOT all colors would report the same (try measuring dark browns)! With that, you're using a rubber ruler to measure the unit since the device can't tell us how inaccurate it might. That's good news considering the pretty poor values for deltaE you're getting (and just formula of deltaE is being provided? Makes a difference!).

And seeing the gamut in 3D provides vastly more information about the resulting profile. But not its accuracy!

[skibum4]

Going even deeper down that hole, you realize that the deltaE values provided are virtually useless; they are generated by the same device that set the calibration and profile. So you're comparing the accuracy with the same device, not a known, higher grade reference device. And the values provided for the delta, you can specify them (and you are?) or its some preset group of colors the software selected? NOT all colors would report the same (try measuring dark browns)! With that, you're using a rubber ruler to measure the unit since the device can't tell us how inaccurate it might. That's good news considering the pretty poor values for deltaE you're getting (and just formula of deltaE is being provided? Makes a difference!).

And seeing the gamut in 3D provides vastly more information about the resulting profile. But not its accuracy!

well if my device is that useless than why in the world do you think it will suddenly produce the ideal profile if I skip the LUT step??

I mean so my device is useless to produce any claims that my LCD is at least within some ballpark of sRGB to begin with plus the LUT and yet if I skip the LUT step then suddenly you are willing to treat it as the sensor of all sensors??

sure it's not THE best probe but the Xrite X94 is known to be one of the best sensors out there that is not in the really crazy $$$$ (thousands of dollares) range, but yeah of course it is not perfect. And the whether it is measuring stuff reasonbly well or not it at least shows some general ballparks of the RELATIVE difference between the primary and secondary locations between LUT and LUT+rest of the profile even if both sets of #'s are off and even if it is not even measuring the relative difference between the two exactly at least it is giving some reasonable ballpark showing.

here I was just testing the 6 sRGB primaries and secondaries and nothing more. I do have other plots of the 16 step grayscale and 8 step saturation curves out to each secondary and primary.

Listen, I have the sensor I have, if you want to send me $10,000 for a top of the top sensor please go ahead.

I have the monitor I have and the HDTV I have. For non-color aware stuff what can I do? I can adjust the set as close as I can for the HDTV and that will have to do it for stuff with no color support of any sort and I can load a LUT for stuff run from my computer and I can run color aware stuff with the associated monitor profile. In that case, I have too much stuff that needs the set to be adjusted with the internal controls to bother swithching every internal setting back and forth everytime I do color-aware stuff or not and I don't care if that means my deep browns are 3% comprosmised to the most ideal possible way I could have things running with my color aware programs and my given sensor....

I am not a 100% dedicated photoshop and nothing else user, I don't actually have a dedicated editing studio and I do actually use my computer from everything from games to TV to movies to internet to image processing to 3D rendering to writing research papers and when I use the HDTV I run many things into it that either support no external calibration at all or a LUT only and I don't feel like having to go into the menus and reset all the settings everything I swap back and forth to CS4 just to get a slightly (perhaps) better purely ICC aware programs profile using no LUT and keeping monitor controls at the default (although since my HDTV has something a 12bit full CMS that appears to work pretty well and my devices can only talk to it at 8 bits I'm failing to see, in that case, how leaving the TV settings all linear and soley using the external ICC profile in color aware programs would necessarily work out better) and as for my monitor I will try a no-LUT based method and if it actualy appears to be way better maybe I will sometimes go and unload the LUT, go into Vista swap profiles and then run CS4 and then switch back each time, at times, but if my gfx card really is supporting 16bit LUTs as it claims then I'm not sure this difference will necessarily be worth it.

Maybe the Lutless profile will give me the perfect dark browns or maybe it won't make much of a differnce compared to my color profile on top of the LUT.

[skibum4]

Going even deeper down that hole, you realize that the deltaE values provided are virtually useless; they are generated by the same device that set the calibration and profile. So you're comparing the accuracy with the same device, not a known, higher grade reference device. And the values provided for the delta, you can specify them (and you are?) or its some preset group of colors the software selected? NOT all colors would report the same (try measuring dark browns)! With that, you're using a rubber ruler to measure the unit since the device can't tell us how inaccurate it might. That's good news considering the pretty poor values for deltaE you're getting (and just formula of deltaE is being provided? Makes a difference!).

And seeing the gamut in 3D provides vastly more information about the resulting profile. But not its accuracy!

also the LUT doesn't even change the locations of the primaries and secondaries at all, so in that case for the non-color aware measurement I wasn't even using the same probe to test itself it was a first order measurement, they all measure out exactly the same LUT loaded or not.

at least I don't have a fluroescent green primary like some hidef projectors....

[arodney]

well if my device is that useless than why in the world do you think it will suddenly produce the ideal profile if I skip the LUT step??

I didn't say your device was useless, but I'm suspecting after all this, your understanding of this process might be moving into that direction.

The LUT is useless (unnecessary) in color managed application. And getting the LUT to a particular state can degrade the quality of what you're viewing in and out of color managed application.

I mean so my device is useless to produce any claims that my LCD is at least within some ballpark of sRGB to begin with plus the LUT and yet if I skip the LUT step then suddenly you are willing to treat it as the sensor of all sensors??

Back to some comment about your device being useless. Where did that come from? Its getting progressively less likely this discussion will continue based on both your misunderstandings of color management but worse, your attitude. Go ahead and muck around any way you wish with your so called calibration process but lets at least protect some of the innocents who might be reading this and actually thinking you're on the right track.

sure it's not THE best probe but the Xrite X94 is known to be one of the best sensors out there that is not in the really crazy $$$$ (thousands of dollares) range, but yeah of course it is not perfect.

Where did the probe (colorimeter) and its make or price come into this? Can you stay on topic? UNLESS you're trying to calibrate the white point of a wide gamut display, that colorimeter is just fine and yes, one of the best made (well it ain't made anymore...). That colorimeter has a filter matrix which isn't expecting a wide gamut and could produce some errors in the white point.

What you failed to grasp above is that you can't measure the accuracy of a process using the same device! If you think your foot is exactly 12 inches (and lets say its 11.9), no matter how many times you measure a foundation for my new home, its not the correct measurement (albeit, it might be close enough), you can't say your measurement accuracy is correct until you take a known, reference grade measuring device (your $2 tape measure) and compare it to your foot. You could take a device that's 10000X more accurate than that $2 tape measure and find that its off by 1/000 of an inch. Makes no difference in your foundation. But you can't hand the building inspector a piece of paper that says you measured with your foot twice, or a dozen times and yes, its the exact size you think it is and add a difference matrix (which is what deltaE is). Get it? That report in your software is there to make you feel good and nothing more. Its to be removed from evidence.

I am not a 100% dedicated photoshop and nothing else user....

Got nothing to do with dedication. There are only TWO types of applications as far as handling numeric values that represent color: those that do it correctly and those that do it incorrectly. Computers only understand numbers. Numbers alone do NOT define color appearance! You must associate a color space with the numbers. ICC aware applications do this, non ICC aware applications dont' have a clue about color spaces. Even then, said numbers will not preview correctly until you also define a display profile for the Display Using Monitor Compensation I keep trying to get you to get your head around. You don't need any LUT for this to work. Outside ICC aware applications, LUT or not, all bets are off. Its as simple as that. If you want to further digress, dig deeper down that rabbit hole, you'll be traveling down that delusional hole on your own. I'm staying up here in the bright sunlight, there's no reason to go farther.

[skibum4]

>The LUT is useless (unnecessary) in color managed application. And >getting the LUT to a particular state can degrade the quality of what >you're viewing in and out of color managed application.

but as I have said a thousand times I am not aware of a single ICC profile aware TV,DVD,HD DVD,Blu-ray player for a PC nor am I aware of any external devices such as stand alone players or cable boxes that can load ICC profiles so the best I can do for those ones is to either adjust the monitor controls for external stuff and use a LUT with the non-ICC managed PC stuff. And I also like my desktop wallpaper and so on to look better than it does without the LUT as well.

I do NOT have a special photoshop only dedicated setup where I do nothing other than CS4,Photomechanic and the like and care about nothing else. If I did then yeah maybe in many cases skipping the LUT stage work work out better (sometiems probably barely distinguishably better sometimes maybe somewhat if not radically distinguishably better).



>Back to some comment about your device being useless. Where did that >come from? Its getting progressively less likely this discussion will continue >based on both your misunderstandings of color management but worse, >your attitude. Go ahead and muck around any way you wish with your so >called calibration process but lets at least protect some of the innocents >who might be reading this and actually thinking you're on the right track.

my attitude right....


>Where did the probe (colorimeter) and its make or price come into this? >Can you stay on topic



>UNLESS you're trying to calibrate the white point of a wide gamut display, >that colorimeter is just fine and yes, one of the best made (well it ain't >made anymore...). That colorimeter has a filter matrix which isn't >expecting a wide gamut and could produce some errors in the white point.

good to know, but my monitor isn't wide gamut, but good to know for future reference.

>What you failed to grasp above is that you can't measure the accuracy of >a process using the same device!

what you failed to grasp is that the first plot, measuring my monitor's sRGB primaries and secondaries did NOT rely on anything calibrated by my probe. Yeah I had the LUT loaded, but that DOES NOT change the xyY values of the primaries and secondaries, which is all I measured there. So I was not using a device to test it's own calibration in that plot. That was the bare monitor for all intents and purposes.

Now in the second plot (which wasn't even the main point of original contention, that my monitor was so hopelessly far off from the sRGB gamut that I may as well not even bother with a LUT for non-color aware stuff or whatever you were going on about) yeah I did use the same probe to verify the very profile it created and without some $10,000 probe to verify it, yeah I can't verify it perfectly and yeah no device (but if I use some other $300 device who is to say that it's differnces from ideal will be much more useful, if at all than just using mine again, they may be in such a way to even mix things up more). And yes maybe mine says that 12 inches is 11 inches and that 3 inches are 4 inches but there is still some utility in using the same device to test a monitor that you used to create the ICC profile for, since the calibration does NOT build a giant table of all 16 million possible shades each one with a direct mapping and since the systematic error in a probe is hardly ever uniform across it's measurement range there is still some utility in it more than you are implying even if there are issues with it to be sure and let's not forget that the systematic and random errors in the probe are often far, far less than the monitor's difference in its original state to the target than to the color-ware corrected state and the target and as I said not uniform and vary from shade to shade that it measures.

It's not quite the same scenario as using the ruler which is a uniform percentage off for each distance measured and where you have specifically measured every single thing that needs to be measured and where to then go back and measure them again after alteration, not that there are not certain issues with it.


> If you think your foot is exactly 12 inches (and lets say its 11.9), no >matter how many times you measure a foundation for my new home, its >not the correct measurement (albeit, it might be close enough), you can't >say your measurement accuracy is correct until you take a known, >reference grade measuring device (your $2 tape measure) and compare it

yes, i am well aware of this


> ICC aware applications do this, non ICC aware applications dont' have a >clue about color spaces.

yes, and as I said, I was not correct at first. Originally I though the LUT loader stage loaded everything needed for correction into the graphics card including the color space transformations. I thought that they got sent sRGB colorspace and then they knew how to do the transform to the monitor colorspace and not just the gamma ramps or if you had a wide gamut monitor that the driver for that in combination with aware programs would allow it to work ok.

Anwyay, yes that was wrong, only color aware apps know anything about the color space and only they apply that transformation.

but few programs are color aware and some combination of internal monitor adjustment and LUT use can greatly improve the look of non-colro aware stuff and as I showed even my 244T's colorspace while certainly not ideal sRGB by any means is not WILDLY different and toggling back and forth between colro aware and non doesn't always even look all that different, not that I don't prefer the color aware. So it sure as heck is useful to use the probe and software to build a correction LUT to at least get it balanced evenly to D65 up and down the scale and to a gamma 2.2 curve for movie viewing. Again, the difference between an ideal reference display setup and my monitor as is is far greater than my monitor plus the LUT loaded into the gfx card.


> Even then, said numbers will not preview correctly until you also define a >display profile for the Display Using Monitor Compensation I keep trying to >get you to get your head around.
>You don't need any LUT for this to work. Outside ICC aware applications, >LUT or not, all bets are off. Its as simple as that.

listen i have said a thousand times that i was wrong at first when I though the LUT loader loaded everything needed for the full correction into teh gfx card.

and I said that you don't need a LUT to make the ICC profile work perfectly for color aware programs if you profile it without using the LUT.

but I also said a 10000000 times that most of my stuff IS NOT COLOR AWARE! and that I much like the look a LUT corrected desktop to just tossing it to the monitor as is.

I also said that when it comes to my HDTV where i need to hook up stuff with no correction at all I need to adjust the sets internal controls as closely to sRGB color space and to the D65/gamma 2.2 most movies and TV expect to be viewed at as encoded for broadcast or on disc. And I do not want to have to go into the menus and switch them back to defaults and unload my LUT everytime I swap between CS4 work and not on that set.
And with an excellent color engine in my TV set (if not my monitor) adjusting things there as much as possible before profiling can lead to less banding since the profile will have less to correct and the interface to my set from my computer is 8bit.

It may or may not be worth it for when I use my computer monitor to create a profile not using any LUT and unload and swap each time I go to CS4 or not, but I'm not sure the difference will be great enough to be worth the bother especially with a 16bit LUT.

> If you want to further digress, dig deeper down that rabbit hole, you'll be >traveling down that delusional hole on your own. I'm staying up here in the >bright sunlight, there's no reason to go farther.

all i can say is i'm happy in my delusions and watching some blu-rays USING a LUT rather than watching them at gamma avg 2.4 varying from at least 2.1 to 2.6 and color temp bouncing all over up and down the grayscale either too hot or too cold by far. Kind of funnt how watching a nice gamma curve at 2.2 is delusional while watchin a movie at D80 and a bendy gamma curve avg 2.5 is sane....
and again where does one get a color aware blu-ray player?????
show us....

[skibum4]

listen, thanks for being one of the ones who got me onto the difference between the LUT loaded into the gfx card step and the rest of the profile, a very crucial point. I had thought the profile loaders loaded the entire set of corrections, all the stuff the LUT does as well as the color space transform, but they do not. Only color-aware apps can do the transform to monitor space correction. But I already got this part like a week ago and readily admitted whever I had posted abotu this that I had been incorrect. And if I ever get a wide gamut monitor it is good to know about the X94 issues, etc.

but i'm not as dumb as you seem to think nor are my mathematical and scientific skills quite as weak with all your talk about deep rabbit holes and delusions, i could say more, but I will just leave it at that....

and not everyone only uses color-aware apps and nothing else and using a LUT darn well can be helpful for non-color aware stuff depending upon your monitor (my hdtv can be adjusted with controls to the point that the LUT doesnt do a whole lot but the color-ware apps do).

what i've found affects the color-ware apps ultimate performance the most is the controls you set on the monitor, sometimes what works best for making the monitor/tv itself be overall the best is NOT the best way to set it to get the most out of color-ware apps, this difference can be relatively large, seems to be much more so than trying LUT vs LUTless profiling. Major thing to look out for is moving primaries and secondaries since some sets have such a poor CMS system all it does is make a warped mess than the profilers dont deal with well an din other cases setting saturations for overall best fit for monitor alone calibration don't let the primaries get as close to the ideal as they could be so sometimes leaving that set on defaut (likely overdone by itself) allws the profiling to work better since it can adjust for any problems and this lets it reach out far enough to perhaps hit the xy and Y locations of the pirmaries.

[marlinspike]

So...

I'm back on XP. Still can't get this working.

0R 0G 255B is purple in Photoshop.

The towel in this photo http://davidson.smugmug.com/photos/195301775_oTn9s-XL.jpg

is blue when I look at it in the smugmug gallery (the correct color), but it's purple when I look at it in a forum where I have embedded it. It's also purple if I use the link I posted to view it. But go through the smugmug gallery clicks and it shows blue. Any ideas?

For some reason there is one profile I can't disassociate from my monitor in the win xp color settings. I'm thinking this is causing the Eye1 D2 to double profile.