• Please use real names.

    Greetings to all who have registered to OPF and those guests taking a look around. Please use real names. Registrations with fictitious names will not be processed. REAL NAMES ONLY will be processed

    Firstname Lastname

    Register

    We are a courteous and supportive community. No need to hide behind an alia. If you have a genuine need for privacy/secrecy then let me know!
  • Welcome to the new site. Here's a thread about the update where you can post your feedback, ask questions or spot those nasty bugs!

CM in browsers with CM-engine

Michael Fontana

pro member
A post in the landscape-board rises the question, how CM savy/aware webbrowser like Safari or IE handle the untagged webjpg.

In my usual workflow, the tiffs were converted to sRGB, prior to be °saved for web°.
In the settings, I did through the profile away, so basically, this produced a untagged sRGB-jpg:

PS_CS_2_prof.jpg


The browsers not beeing ICC-aware (like FireFox) display the image "normally"; this means it looks like the tiff in PS, within its working space.

For the ICC-aware browsers, we have two options:

A) The profile is added in °save for web° and the browser recognises the tag (sRGB), and shows therefore the image properly. Perfect!

B) Nativly this beeing a sRGB, but throwing the profile away, (screenshot above), the ICC-aware browsers doesn't know, what native colorspace it had and display it differently. This is correct on the CM-level, but bad for the viewer. "Nativ" means here not the workspace of the tiff; but the sRGB profile, beeing converted in, prior to be send to °sace for web°

So what colorspace does a ICC-aware browser displays, when it doesn't finds a profile?
Is it the monitor profile? This has been my first guess, but its not!

Any insights?
Maybe I should check the profile-box, to tag the jpg with the sRGB-profile; the °profile-through- away° might come from the days whit lowspeed modems. (??)

By adding the profile, the NON-ICC-aware browser will not recognise any profile, but as sRGB is the webstandart, it will show it properly. Is this correct?


Here the examples, one jpg is °saved to web° with the profile, the other one without.

safari.jpg



Firefox.jpg
 
By adding the profile, the NON-ICC-aware browser will not recognise any profile, but as sRGB is the webstandart, it will show it properly. Is this correct?
NON-ICC-aware browsers send RGB values to monitor without any transform. So the rendering is more or less good depending on monitor color space. For monitor with a gamut similar to sRGB the rendering is good. In any case this is the best you can do as monitor gamut is , generally, not so far from sRGB.

So what colorspace does a ICC-aware browser displays, when it doesn't finds a profile?
As you said ICC-aware browsers have to make an assumption if the image does’nt have an embedded profile.

I don’t know Safari, but Mac world experts claim that Safari assumes the monitor profile. If this is correct the Safari rendering should be a transform from MonitorColorSpace to MonitorColorSpace (same as null transform) and image RGB values should go untouched to monitor.

So you should see the same colors using Firefox or Safari.

Now, you said that colors are different. So :
1- Safari doesn’t assume monitor profile for image without profile
OR
2- Somthing goes wrong in your test
 

Asher Kelman

OPF Owner/Editor-in-Chief
Are you saying that we should see the same colors using Safari=Firefox=CS2/CS3?

That Giovanni is the question!

Asher
 
Asher,
you have a photo without profile and you (or Safari, or any other color-managed application) assign the monitor profile to this image.

In this case:
- Firefox send RGB values to graphic card
- Safari, or any other color-managed application, before sending RGB values to the graphic card transforms them from image profile to monitor profile. But imageprofile=monitorprofile, so RGB values don't change.

If this is the case, yes you see the same colors.
 

Asher Kelman

OPF Owner/Editor-in-Chief
The problem is, I believe when an un-referenced or incorrectly assigned Image profile is remapped to your profiled monitor, the colors will be mapped incorrectly.

When an image is sent to someone from a computer with its own monitor profile, it arrives to you in a foreign state, from a smaller and different gamut. So now when Safari reassigns the colors in order to appear correct for it's own profile, it has to stretch and do its best. So the color is not perfect.

Asher
 
The problem is, I believe when an un-referenced or incorrectly assigned Image profile is remapped to your profiled monitor, the colors will be mapped incorrectly.
Yes, if the assigned color space is different from the image color space resulting colors are wrong.

When an image is sent to someone from a computer with its own monitor profile, it arrives to you in a foreign state, from a smaller and different gamut. So now when Safari reassigns the colors in order to appear correct for it's own profile, it has to stretch and do its best. So the color is not perfect.
I don't think so. unless the profile or the CMM is incorrect
I think Mac CMM is correct. I can't be sure the monitor profile is corret (I saw so many wrong monitor profiles...).
There is no reason to embedd the monitor profile into an image. You have to avoid that. But, if it happens, for a mistake, then color management can trasform the image for the current monitor you are looking. If the image is not converted from a CMM, color may be more or less wrong.
 

Asher Kelman

OPF Owner/Editor-in-Chief
I just downloaded Fed Spencer's post # 34 in the Ballroom Dancing thread, (original by © Ron Hiner).
So what I'm seeing is a file which is dragged on to the desktop

  1. is "RGB", not Adobe RGB.
  2. opens up supersaturated and blotchy in Photoshop (which us set to a color space of Adobe RGB (1998)
  3. "assigned" to Color LCD matches it the the Bone China appearance in Safari, Firefox, Explorer or Preview and the OPF post #34.
  4. Converting to Apple RGB is better, much better than Adobe RGB but still not the same as "Assigning" to Color LCD
  5. Reloading the desktop RGB file into OPF post, produces orginal Bone China look once more, indicating that the grabbing of the image to the desktop does not, itself, alter the file.

So how on earth can one make any adjustments in Photoshop when we are not seeing the image that we see in the browsers? It's against the rules to work with "assign" to color profile!!

Asher
 
Last edited:

Ray West

New member
Hi Asher,

The original image (post no 34) has no assigned colour space. Picassa is not colour aware. srgb is the default colour space for 'web space'.

If I copy the image four times (different names) and load each image, assigning srgb, adobe rgb, applergb, prophoto, it is quite clear that the colours are shifted. However, for the colour range used in this example, there is little difference between srgb and adobe rgb (adobe is a little warmer). Apple rgb, is more washed out, prophoto is more flourescent. srgb in cs2 is virtually the same as the browser shows.

As I mentioned in Mike's earlier thread, the problems come in converting it back/saving for web.

I normally edit in prophoto/16bit within cs2, I can still do that, if the incoming image is assigned to its correct colour space, in this case assign to srgb, then convert to working colour space. The image looks like it does in the browser, but I can push the colours further in pro photo(but may not be able to see some of them on the monitor ;-).

Now, if I save it as a jpg (have to convert to 8bit), because the browser ignores the colour space info (assumes srgb) the colours become drab cf the prophoto coloursin cs2. However, loading that jpg back into cs2 allows me the choice of using the embedded colour space, which makes it the same as before it was saved, or ignoring it, which makes it appear the same as in the browse, or assigning another space. Basically, cs2 assigns a colour space which is saved with the jpeg, but ignored by the browser.

This is why, in the previous thread, I said that I worked in srgb for web - it is much easier.

I'm not sure what rules you are referring to wrt assign being against them.

If I am printing, but just want to post an image to opf, then I use picasso, or Irfan view, on the final image. Either are plenty good enough to present an 600x400 image for web use, and simple to get the colours, more or less the same as in cs2. If I'm not printing, then I find it easier to work in srgb throughout.

As you are aware, it can take time to discover what is happening, but it is consistent, if complicated.

Best wishes,

Ray
 

StuartRae

New member
The original image which we've all been working from was tagged with Nikon's variant of aRGB.

profile.jpg


So without converting to sRGB we're to some extent wasting our time trying to improve it.

I'll shortly make a post in the other forum regarding this.

Stuart
 

Ray West

New member
Ah, well, in for a penny, in for a pound....

Speaking in broad terms, back of an envelope sort of thing, generally speaking, so none of the following is to be taken to be any more accurate than I think it is. (drawing not to scale)

r.jpg



u represents all the wavelengths in all the universes. The line is infinitely long - there is no beginning, nor end
h represents human vision - very short
a represents adobe colour space
v is the colours that a vdu can display
s is the srgb colour space
p is the prophoto colour space
e is an epson printer colour space

very general

we can't see everything.
it may not be possible to contain all of one colour space within another.
there is a range of frequencies common to all 'useful' colour spaces.
some colour spaces have a range outside of human vision, and outside of any vdu display technology currently available.
anyone can make their own colour space, hope it becomes popular, hence become a standard, or set up/partake in a standard committee to promote it.
if the colour space is not known (unassigned), it may show differently than intended if transferred into another colour space (how to handle the extremes - stretch or compress, or just throw away the values or set to the highest, perceptual/actual etc...)
If we are seeing something on a vdu, the most we can see lies within the wavelengths bounded by the vdu and our own eyesight capabilities.


specifics

there are plenty of pretty coloured charts on the web and elsewhere, and some you can overlay, to see the area that each colour space overlaps another. However, they may make you cross eyed, and such things are confusing, because they try and represent 3d in 2d space (since some time ago the colour wheel was invented and since then we are locked into a 2d representation of colour, whereas sometimes it is simpler to just consider wavelengths and numbers.)

reader's exercise

find the wavelengths for the end of each line, draw each line to linear scale (better exclude the universe one). Discover if the known spaces convert the wavelengths consistently (linearly). Does it matter?

Best wishes,

Ray
 

Jim Gilley

New member
Let me see if I understand things correctly. In the past, when we were using CRTs, we expected them to have a gamut roughly equal to that of sRGB (which is what the profile was created for, as I recall). I also believe that sRGB is understood to be the "universal" color space, in that if an image has no profile attached (say from the web, for instance), then we should assume it is sRGB. Furthermore, when someone works up an image in PS using AdobeRGB, then saves it with profile, if we view it in many Windows apps that are not profile aware, then it will look "washed-out" because AdobeRGB has a wider gamut than sRGB. Am I correct so far?

So now, let's suppose that instead of an old CRT that approximates sRGB, we have a new high-end CRT that approximates AdobeRGB, or in other words, has a native gamut that is wider than sRGB. If we were to use "dumb" Windows apps to view an image, presumably the color is going to be bonkers because the image was likely saved with the assumption that people were going to view it as sRGB, but our monitor is really more like AdobeRGB. In this case, wouldn't said image now look super-saturated?

The reason I ask, is because I have a lot of trouble with CM. I have a Dell HC LCD that comes very close to AdobeRGB right out of the box. I use Monaco Optix to calibrate and profile this monitor, and when I check the profile using the puck, it's not too bad. I have all my Windows and PS color settings correct (as best I know, and I have double checked these many times), but I still have a bear of a time with color consistency.

Usually, I do my raw conversion in DPP, which is alleged to be profile-aware, and I save the result with the AdobeRGB profile. Then I open and edit in PS, using AdobeRGB working space. All is well, and everything looks great. Now I convert it to sRGB and save it as a jpeg for use on the web. In PS, I can reopen this jpeg, and it looks fine. But if I drop out of PS and open the jpeg using a browser, all of a sudden the image has gone nuclear on me, it's ridiculously over-saturated. Is this a problem others have experienced, and is my understanding of it anywhere close to being correct?

Also, the other day, I had some fun when I visited http://www.color.org/version4html.xalter and saved the four jpegs locally. The two goofy-colored ones, I opened using a variety of different Windows programs, to see which apps seemed to respect the profiles, and which did not. The results were very surprising. Oddly enough, that little preview pane in Windows Explorer is profile aware (or at least, it shows all four squares in the correct color). Yet the default picture viewers (both Windows and Office) are not profile aware. And so it goes.
 

Andrew Rodney

New member
So now, let's suppose that instead of an old CRT that approximates sRGB, we have a new high-end CRT that approximates AdobeRGB, or in other words, has a native gamut that is wider than sRGB. If we were to use "dumb" Windows apps to view an image, presumably the color is going to be bonkers because the image was likely saved with the assumption that people were going to view it as sRGB, but our monitor is really more like AdobeRGB. In this case, wouldn't said image now look super-saturated?

Actually desatruated. But the point is, without a COLOR MANAGEMENT savvy browser, the color appearance is incorrect. Even for sRGB (although less so).
 
Stuart is correct about embedded profile..but
So without converting to sRGB we're to some extent wasting our time trying to improve it
If you work in a color managed application, you can edit and at the end save in sRGB color space for web. Or, if you like, you can convert to sRGB before editing.

Jim,
Now I convert it to sRGB and save it as a jpeg for use on the web. In PS, I can reopen this jpeg, and it looks fine. But if I drop out of PS and open the jpeg using a browser, all of a sudden the image has gone nuclear on me, it's ridiculously over-saturated
All is ok. Your monitor has a gamut larger than sRGB. So sRGB image viewed on non-color-managed application appares richer. Your situation is inverted respect to people with sRGB-like monitor viewing Adobe 1998 images in non-color-managed applications.

The rule is
image (large gamut) viewed on monitor (small gamut) ==> washed colors
image (small gamut) viewed on monitor (largel gamut) ==> richer colors
 

René Damkot

New member
Actually desatruated.
Nope. When you view an sRGB image in a non CM application, on an AdobeRGB display, the colors will appear over saturated.
If you view an AdobeRGB image in a non CM application, on an sRGB display, the colors will appear desaturated.

I wrote a longish post on CM on POTN: Here. Might be interesting to some.
 

ron_hiner

New member
Ray -- that back-of-the-envelope drawing speaks volumes... thank you for that.

I've seen many of those funky color charts -- not really knowing what to make of them, or how to find them useful. You sketch is infinely more useful in demonstrating the problem.

Now... whats the difference between "sRGB" and "sRGB IEC61966-2.1"? Both appear on my system's drop down lists in varoius applictions, but I don't know the difference. I always choose the latter now when I'm send pics to someone who I know wouldn't have any idea what a color profile is. But I don't know if that's the right choice.

Ron
 

Ray West

New member
Hi Ron,

Thanks for looking - it really was on the back of an envelope, then transcribed into paint (You can see the remnants of my wavy trackball in the universe line ;-)

Concerning your query -

Now... whats the difference between "sRGB" and "sRGB IEC61966-2.1"?

well, strictly speaking "sRGB IEC61966-2.1" is "sRGB IEC61966-2.1" and "sRGB" is not.

But, in the loose language of this stuff on the web, and elsewhere, folk will use sRGB as short hand for the IEC standard. Sure you're not mistaking the "s" for an "a"?

Best wishes,

Ray
 

ron_hiner

New member
well, strictly speaking "sRGB IEC61966-2.1" is "sRGB IEC61966-2.1" and "sRGB" is not.But, in the loose language of this stuff on the web, and elsewhere, folk will use sRGB as short hand for the IEC standard. Sure you're not mistaking the "s" for an "a"?

My system (Mac Pro/Tiger) has several "sRGB" profiles....

"Nikon sRGB 4.0.0.3000"
"sRGB Profile"
"sRGB IEC 61966-2.1"

I have no idea what the difference is.

And there a dozen more so that that have "RGB" in their names.


Ron
 

Ray West

New member
Hi Ron,

Thanks, I see what you were saying, now. I'm not a MAC guy, but I guess its the same for Adobe cs2 in windows.

If I go to print, say, I can choose a printer profile - some came with th software, some with the printers/paper, others I've generated myself. Also, it list the profiles for the cameras, scanners and working colour space. I won't be using many of them, there are dozens there. I am only concerned with the ones I use.

I do not think that you will notice much difference 'twixt "sRGB IEC61966-2.1" and "sRGB". The Nikon profile is probably for a camera. Are you working in a colour managed environment - i.e. have profiled your printer/paper/monitor, etc? In which case, I think you need only bother with the ones you use. If not, then someone using a MAC can probably give a clue as to its native settings.

Best wishes,

Ray
 

nicolas claris

OPF Co-founder/Administrator
My system (Mac Pro/Tiger) has several "sRGB" profiles....

"Nikon sRGB 4.0.0.3000"
"sRGB Profile"
"sRGB IEC 61966-2.1"

I have no idea what the difference is.

And there a dozen more so that that have "RGB" in their names.


Ron

Hi Ron

"sRGB IEC 61966-2.1"

is the official one, forget, avoid all other sRGB if you don't want to be in trouble!

IEC = INTERNATIONAL ELECTROTECHNICAL COMMISSION

Google "sRGB IEC 61966-2-1" and you'll get some nights reading … ;-)
 

Andrew Rodney

New member
I don't have anything but the sRGB ICE on my machines so I can't check the profiles in ColorSync utility (those on a Mac can). I do recall one (I think it was called "simplified sRGB" did NOT have the sRGB TRC, the other's do. The sRGB color space doesn't have a true gamma curve! There's a tweak to the curve in the deep shadows meaning it doesn't follow the gamma formula.
 

nicolas claris

OPF Co-founder/Administrator
I don't have anything but the sRGB ICE on my machines so I can't check the profiles in ColorSync utility (those on a Mac can). I do recall one (I think it was called "simplified sRGB" did NOT have the sRGB TRC, the other's do. The sRGB color space doesn't have a true gamma curve! There's a tweak to the curve in the deep shadows meaning it doesn't follow the gamma formula.

Bonjour Rodney,
I can send you some of these bizarres profiles if you want to dig them… for the fun?
 

Herman Teeuwen

New member
I don't have anything but the sRGB ICE on my machines so I can't check the profiles in ColorSync utility (those on a Mac can). I do recall one (I think it was called "simplified sRGB" did NOT have the sRGB TRC, the other's do. The sRGB color space doesn't have a true gamma curve! There's a tweak to the curve in the deep shadows meaning it doesn't follow the gamma formula.

Those not on a Mac can use the free "ICC Profiler Inspector" that can be downloaded from here:
http://www.color.org/profileinspector.xalter
 
Top