Lawrence,
Your "assign" and "convert" are working just fine on the sRGB and AdobeRGB colorspaces. "Assign" does not change any color numbers, only "re-interprets" them according to the gamut of the destination colorspace. "Convert" does change the RGB numbers to the "equivalent" numbers in the new space, but should not change the appearance ( except perhaps for out of gamut color).
I think your converting to a printer profile, though, involves a conversion to CMYK ( a very different critter from RGB) and what you see is the result of the difference in gamuts of the two spaces plus the effects of the RGB to CMYK conversion ( or vice versa ! ). At least that’s my current theory <G> .
🙂
Brent
The printer conversion is to an RGB printer, usually on Fuji Chrystal Archive. Mostly I use that to see what the rendering intent should be, or is, for that printer profile.
What bothers me is that in either the Assign or the Convert, the Document Color space reflects the change to whatever you have changed to, but for Assign, you see a change, for Convert, you don’t. So the question is Who (or what) is the Convert to profile intended if it is not for the person in front of the computer? I can test it to see, but it appears that if I send a document to the printer after Assign Profile, I will see a different result, close to the monitor. If I send a document to the printer after Convert to Profile will I see the same change? If not, what the hell is the conversion for? If so, why am I not seeing it before I print?
I figured I would hear from you Ian! How was NY?
I read the link, as well as yours on the subject. I must say yours is more succinct. He confused matters by digressing in his description, to PS 5 and what it did. I couldn’t care less!
So here’s what I did and the results:
I generated a color patch; R-G-B values, Adobe 98.
100,
25,
100
L-a-b values
27,
47,
-24
Document Profile Adobe 98
I printed that my usual way.
Next I Assigned Profile to sRGB. Numbers are now:
R-G-B
100,
25,
100
L-a-b,
25
47
-27
Document Profile sRGB…..
Now both the screen and the print color changed.
Next Convert to Profile sRGB: (from original document)Numbers are now:
R-G-B:
116
18
102
L-a-b:
27
47
-24
Neither the print nor the screen changed BUT:
Document Profile:
sRGB
Conclusion? Monitor and printer obeying Lab numbers (Duh!)
Document Readout, screwed! (Very scientific word, screwed!) So is operator.
So, what is the real world meaning of a profile change that seems to be in name only? Especially when it is masquerading as something else?
Paraphrasing Saturday Nite:
Is this the color to which I am looking?
Lawrence,
I may be a fool for interjecting here, but so be it. I think that you need to take a different perspective here. When you "assigned" sRGB to your created image, you told Photoshop to interpret the RGB numbers according to the gamut of that sRGB colorspace ( and tag the image sRGB ). You might also have done the same thing with the ColormatchRGB colorspace and come up with a new image ( different L-A-B numbers to be sure, but the same RGB numbers ).
All the images that were "Assigned" a different space appear differently ( and have different L-A-B numbers ) , but have the same RGB numbers.
That’s what "assign" does. I use "assign" occasionally when my digital photos ( Coolpix 995 ) warrant the attention. Some of them ( depending upong the character of the light, I suspect ) are more pleasing when assigned ColormatchRGB, rather than my usual BruceRGB ( or sometimes, AdobeRGB ) . The "assignment" gives me a starting point that’s closer to my finished editting point, so hopefully I lose less in the translation.
"Convert" is intended to "translate" an image from one colorspace to another, while not affecting the appearance of the image ( consistent L-A-B numbers, if you will ) .
"So, what is the real world meaning of a profile change that seems to be in name only? Especially when it is masquerading as something else?" The profile change effected by the "assign" command is not necessarily a valid change… It’s only because YOU are the boss, and YOU say that the image is sRGB ( or whatever ).
"Convert" is real . "Assign" is a useful tool ( and I , in my amateur toying around, use it much more than "convert" ) but it certainly has nothing to do with the integrity of your image .
Hope this is not too far off base. Maybe I’d better go hide out in the Elements forum again.
🙂
Brent
Close, but no cigar. We are naming two different outcomes the same.
That’s crazy making, IMHO.
Lawrentz – Brent’s got it.
Convert means the numbers change, while trying to keep the appearance the same.
Assign means the numbers stay the same, but the interpretation of them changes, so the appearance changes.
It’s still crazy making, Chris. I have read that descripion many times and I understand the difference,but the problem is the name’s the same and the result is different. Just how do you keep track of what you did say 3 months later? You see a profile change, but which one? I can show you two versions with the same name for the profile but they look different. No adjustments, nothing except how I chose to make the change and they both are called the same.
Doesn’t that give you pause? It certainly gives people in my part of the engineering profession raised eyebrows.
Look, the same problem showed up years ago when Tektronix introduced the concept of Delayed Sweep. It was solved by calling one version Delayed, and the other Delaying. We had to draw the distinction, even as the difference between terms were not much. It was not enough to even call on A and the other B, which they were.
I actually know the differences. I want to name the differences so I don’t mess up later.
And actually, numbers change in either profile. In one, RGB changes, in the other Lab changes. I believe what you are saying is that in Convert, RGB changes. In Assign, they don’t. I would be equally correct in taking the other tack, ie in Convert the numbers don’t change therefore the appearance doesn’t, which tracks experience, and in Assign the numbers change but the appearance doesn’t.
People who do that for a living are called magicians. 😉
Perhaps it’s nit picking, but I tell you from only about three years with Photoshop, I do not have another issue that causes so much consternation when I use ii as Assign/Convert.
Maybe I am the Lone Ranger here……:-(
No, the names are different, and the results match the names.
They aren’t very similar, that is why they have VERY different names.
And no, you are not equally correct there. Convert implies that the numbers are changing. Assign implies that you’re just reinterpreting the existing numbers.
Ok, I’m speaking of the label. You have named it: Assign and Convert, and I do understand the difference, even better after running the test. The finished transformation however, is labelled sRGB with either method (in the case described), which is true, but insufficient. That’s the rub.
Perhaps a color code might be applicable. Convert and the label is black. Assign and the label is red. Or visa versa.
I think I’ll go watch a Woddy Allen flick! My head hurts! 😉
Thanks for listening.
..
Think of it like languages:
If I label a text as French, it is French.
If I translate something to French, it is French.
In both cases, the end result is French.
But in one case I changed the contents, and in the other case I did not.
The situation is ambiguous at best and possesses the qualities of a conundrum at worse. Obviously, if one tags a file with a profile, that profile must be named. And, Converting or Assigning will tag the file, correct?
The mystery to me is why, if I am changing the color space from a wide gamut to a narrowed one, with a file that contains a large color range, don’t I see the results of such a change in either mode? If I am putting that image on the web, I want to see what it’s going to look like, and Convert will not give it to me, even though the same process, converting to a print space, will show me the change.
Or does it?
I’m going to check again.
PS: Woody was no help!
One more and I’m off to bed.
I opened my fall shot and made two copies. One stayed in Adobe 98, the other Convert to sRGB. Both still look the same. Now I went to Save to Web and ran both.
What was to outcome you would expect to see?
What I got was this:
The Adobe RGB file dulled down considerably not unlike Assign sRGB. The sRGB file, however, stayed as bright as the original. I mean the original Adobe RGB.
So if I Convert first, then save to web, I retain 99% of the original Adobe98 file. If I Save to Web with the Adobe 98 file, I lose color.
Most unexpected!
What was to outcome you would expect to see?
What you saw!
What I got was this:
The Adobe RGB file dulled down considerably not unlike Assign sRGB. The sRGB file, however, stayed as bright as the original. I mean the original Adobe RGB.
So if I Convert first, then save to web, I retain 99% of the original Adobe98 file. If I Save to Web with the Adobe 98 file, I lose color.
Most unexpected!
The Save for the Web preview in its default configuration is not colour managed so what you are seeing for the Adobe RGB (1998) image is as expected – an Adobe RGB image being sent straight to the monitor without any form of compensation will look desaturated at best. Given that you’ve proably calibrated and profiled your display to G2.2 and WP 6500K you’ll find that the differences between monitor_RGB and sRGB are suffiecenlty small that your sRGB image will not appear to change when previewed in Save for Web. To overcome the preview descrepency with Adobe RGB images you’ll find on the top right hand corner of the Save for the Web preview window a little triangle – select it and and choose "Use document Profile". Your images will now be displayed correctly (they’ll be colour managed) inside Save for the Web. However, the Adobe RGB image will still look like crap outside of Photoshop because most of your other apps including IE, Netscape etc aren’t colour managed. Images that you intend to show on the web REALLY should be CONVERTED to sRGB before saving.
The Adobe RGB file dulled down considerably not unlike Assign sRGB. The sRGB file, however, stayed as bright as the original. I mean the original Adobe RGB. So if I Convert first, then save to web, I retain 99% of the original Adobe98 file. If I Save to Web with the Adobe 98 file, I lose color. Most unexpected!
Not at all unexpected!
This is exactly the way it should be.
Convert to (sRGB) profile = Preserve appearance (in web Browser)
Leave in Adobe RGB = Assign sRGB profile (in web Browser) = change appearance.
Widows Browsers pretty much assume your files to be sRGB
feed them what they expect.
Thanks. I’m glad to know that it’s normal.
Since I am beginning to develop a web site, perhaps stumbling on all this at this time is timely. I’ve posted images to other sites in the past, and sometimes I would look at the final result in the edit mode of the site and say no way! It’s too dull.
So, what is happening is the Assign profile is the way the file is changed for the web, not convert. The next question to ask (and I will run it) is if I Assign instead of Convert before Save for Web, will it still look dull?
My guess is yes, and that should not be. However, I can learn to live with it.
Sleeping bags are "assigned" stuff sacks and a large "gamut" becomes (converted to?) a small one. I see the term stuff sack and I know what to expect. What would you think if the same bag supposedly in the same stuff sack looked like it was not in it’s sack, but everyone else said it was and labelled it such? 😉
So, what is happening is the Assign profile is the way the file is changed for the web, not convert. The next question to ask (and I will run it) is if I Assign instead of Convert before Save for Web, will it still look dull?
Your NOT listening!
You need to your CONVERT your images from Adobe RGB (1998) to sRGB for the web. Conversion changes the numbers but it also does it in a way that ATTEMPTS to maintain the VISUAL appearance of the original. Converting from Adobe RGB (1998) to sRGB or even your monitor profile will present no problems in maintaining the original appearance.
Larry,
Listen to Ian!
Forget Assign Profile (for now) except in these situations:
File does not have embedded profile (untagged) or you know Embedded profile to be incorrect.
You dont just (re) assign different profile to change (correctly tagged) files color space. You need to CONVERT to another profile (color space).
Ian, Andrew, I am listening! I also am analyzing. Simply to say listen to me is proof by pulling rank. No way!
Skepticism has driven science/technology a long way, and many a time it is the skeptic who says "show me" that drives it to a new level.
Forget Assign profile? If I had, I still would be Saving for Web erroneously, and innocently. I would have accepted the loss as part of the game.
If textbooks were enough, we would have no need for labs. If theoretical physics were enough, we would not have Fermilab, or Lawrence Livermore.
For what it’s worth, I always reconcile disparate elements, or at least explain why they can’t. I certainly don’t hide from them. Otherwise, teaching the subject to others will eventually catch up with one, usually with egg on their face. 😉
wrote:
Widows Browsers pretty much assume your files to be sRGBâ¦feed them what they expect.
I tend to prefer using Soft Proof > Monitor RGB inside Photoshop to predict how colors would look like in my browser, because images in sRGB inside Photoshop doesn’t always look the same in my browser (Internet Explorer 6). They’re often very different.
This in an sRGB image viewed inside Photoshop and in the browser: <
http://home18.inet.tele.dk/madsen/photoshop/color/srgb.jpg>. I can’t understand why many people say that most browsers "speaks" sRGB. My browser sure doesn’t.
This is the same sRGB image but soft proofed to Monitor RGB: <
http://home18.inet.tele.dk/madsen/photoshop/color/monitor.jp g>.
I haven’t got a change to know how the image would look like in another users non-color managed browser anyway, no matter what profile I convert to, but at least I know how the image would look like in my own browser, if I use the Monitor RGB soft proof.
It may be true that sRGB comes closer to the browsers gamut than Adobe RGB does, but Windows browsers don’t "speak" sRGB, as far as I can see. They haven’t got a clue what a color profile is really. They just send the color numbers directly to the monitor, like Monitor RGB inside Photoshop does.
Am I wrong?
—
Regards
Madsen.
Switching back and forth, I could see very little difference, Thomas.
I can see that, now I am undertaking a website for my work, that I really need to pay attention, if I expect clients to pay money!
There are some dynamite color websites, so I know it can be done.
wrote:
Switching back and forth, I could see very little difference, Thomas.
Okay. I often see a huge shift in especially reds and blues if I have an sRGB image open in Photoshop and Internet Explorer side by side. Maybe it’s only happening here on my monitor but I hope not.
There are some dynamite color websites, so I know it can be done.
Sure. I’m just wondering where the statement that "browsers sees sRGB" comes from, because I just don’t get it. From my point of view, browsers sees Monitor RGB and Monitor RGB is often very different from one monitor to another. You can calibrate them of course, but the gamut from one monitor to another is still different. You can buy CRT monitors with a gamut very close to Adobe RGB(1998). Does the statement "browsers sees sRGB" mean that even if you have a monitor like that, a browser like Internet Explorer will cut of colors that are out of the sRGB gamut?
—
Regards
Madsen.
Those are good questions, Thomas.
I’d like to see them addressed as well. Perhaps a Google search might help?
Lawrence
Madsen,
Browsers sees Monitor RGB and Monitor RGB is often very different from one monitor to another.
Bingo! Yet color gamut for majority of monitors (high end included) is NOT much bigger (if at all) than sRGB, which is why sRGB is widely accepted as standard color space for Web images.
I didnt mean to imply (in my earlier response to Larry) that Widows Web Browsers are color managed or somehow limited to sRGB. They are not.
There is nothing special about sRGB either. It is simply a known standard and good compromise (or middle ground) between my monitor RGB, your monitor RGB, Larrys monitor RGB etc.
wrote:
Yet color gamut for majority of monitors (high end included) is NOT much bigger (if at all) than sRGB, which is why sRGB is widely accepted as standard color space for Web images.
You can buy monitors with a bigger gamut than sRGB. Mitsubishi’s Diamondtron RDF225WG for instance.
<
http://www.dpreview.com/news/0310/03100701nec22adobemon.asp>. The gamut of my Sony GDM-FW900 is also bigger than sRGB, if I compare the profile made for my Sony to the sRGB profile, but you’re probably right that the majority of monitors is lower or very close to sRGB.
I didnât mean to imply (in my earlier response to Larry) that Widows Web Browsers are color managed or somehow limited to sRGB. They are not.
Okay. I was just wondering if that was what you meant, because I have heard others say that they are indeed limit to sRGB somehow. (Something about ICM 2.0 but they couldn’t tell me more than that).
There is nothing special about sRGB either. It is simply a known standard and good compromise (or middle ground) between my monitor RGB, your monitor RGB, Larryâs monitor RGB etc.
I agree.
—
Regards
Madsen.
Madsen,
The gamut of my Sony GDM-FW900 is also bigger than sRGB, if I compare the profile made for my Sony to the sRGB profile
You do realize that you need 3-D plot to REALLY evaluate color gamut dont you?
I have Sony GDM-F 520, which probably uses the same tube technology as yours.
When I looked at 2-D graph of my monitor profile it looked more impressive than sRGB. When I had 3-D plot made based on the same profile I got little reality check.
Is my monitor gamut different than sRGB? Yes. Is it bigger? Not by much.
I think statements made in the past by some of the industry big guns were bit overblown.
Andrew
wrote:
You do realize that you need 3-D plot to REALLY evaluate color gamut donât you?
Yes, I’m aware of that.
I have used iccProfile Viewer by Chromaticity. It can give me a 3-D plot of different profiles. In that program I’ve loaded the profile made for my FW900 (made with OptiCAL and a monitor spyder from Colorvision) and the standard sRGB-profile. I’m not saying that FW900 is much bigger, but it’s bigger than sRGB.
On the other hand, if I compare the profile made for my TFT monitor (Viewsonic VP201s) to sRGB, it’s smaller than sRGB.
—
Regards
Madsen.
Sorry about the strange signs in my posts. I don’t know what’s happening, but sometimes the charset goes nuts.
Regards
Madsen
I thought it was my glasses !! <G>
