AIM color space - Gamma 1.00

[Ron]

Color Space - Are you seeing the real picture?

If you visit some of my entries you will find I usually begin by opening my images into "AIM RGB = Trinitron D65 G1.00" color space. If you're curious about what this means I'll try to explain in general terms what this is all about. Bear in mind that this subject is controversial and hotly debated among digital professionals. (Some believe this system, for making digital manipulations more accurate, to be overkill and not required).

First: A brief introduction to the terms color space, gamut and gamma.

As I understand it, color space, in the digital world, is a term used to describe the transference of color information from the real world onto a computer monitor or television screen. Computer monitors are incapable of displaying all the true colors found in nature so various formulas have been developed to adjust the monitor spectrum to get the best representation possible. These adjustments control how the Red, Green, Blue, color guns affect the color properties of the screen. Two of these mathematical formulas are best know to us as "Adobe RGB 1998" and "sRGB IE C61966 2.1". When a photograph is scanned into a computer and opened in i.e. Photoshop the color information contained in the original scan is mathematically altered so it looks like the original print (as much as is possible).
Another aspect of color space is the term wide gamut. For example "Adobe RGB 1998", "sRGB IE C61966 2.1" and "AIM RGB PRO = Widegamut D65 G1.00" are all wide gamut color spaces. This means that they can handle color that falls outside of the capabilities of a computer monitor. A normal color photograph can contain color information that a computer monitor cannot display (similar to the way some high and low pitched sounds are inaudible to the human ear). If you are going to create prints from your work then you should always use a wide gamut color space. If, on the other hand, all your work is only going to be displayed on a monitor (or the WEB) then you can use a "narrow gamut" color space. All my RetouchPro entries are done with "AIM RGB = Trinitron D65 G1.00" which is a narrow gamut color space.

"Color space" dovetails with Gamma space. Gamma space refers to the settings on your monitor that contain the adjustments for contrast, brightness and full black (Black Point). For the uncalibrated Mac system the gamma setting is 1.72, for the native PC the setting is 2.5. These setting are industry standards that govern manufacture of monitors for the Mac or PC. The "sRGB" specification for a Gamma of 2.2 is an attempt to have a digital image display somewhat the same on both the Mac and a PC. This is particularly important when it concerns the “World Wide Web” because most information on the WEB is not platform dependent. A Mac or PC computer might view the same digital images.

If I haven't lost you, I'll try to explain the perceived advantages of using Gamma 1.00

Both "AIM RGB = Trinitron D65 G1.00" and "AIM RGB PRO = Widegamut D65 G1.00" use a gamma space of 1.00 (G1.00). (Remember that normally, the Mac is set to 1.72 and a PC to 2.5).

When a computer system is set to Gamma 1.00 it’s said to be linearly calibrated and properly balanced. Setting your system to gamma 1.00 is considered important because all corrective filters we use every day are created as pure mathematical manipulations done in linear space. In other words they only work correctly, as intended, when used in a linear color space. When we use a Blur or Unsharp filter, for example, in a color space that is NOT set to gamma 1.00 (i.e. Adobe RGB 1998) it's like applying the filter through a distortion lens. When using a non-linear color space (i.e. Adobe RGB 1998), more filtering is required to do the same job and the results are dirtier with a higher degree of distortion to the image.
You should also be aware that all scanners scan in linear color space (i.e. gamma 1.00) then software is used to adjust the color space to match the monitor. Some poor quality scanners have this color adjustment built in and it cannot be changed. You should be able to reset your scanner back to gamma 1.00.

Using Linear color space - the ins and outs:

When I open an image from RetouchPro into Photoshop it is usually from an "Adobe RGB 1998" or "sRGB IE C61966 2.1" color space. With my default color space set to "AIM RGB = Trinitron D65 G1.00" a profile mismatch screen appears asking if I want to convert the image. The first thing I notice when I click OK is that the open image is a lot lighter in color than normal so my first adjustment is Image\Adjustments\Levels where I reset the image to a better "brightness" level. At the same time I usually adjust the colors to normalize the whites of the image - adjusting the colors until an area that I believe should be white, actually becomes white (or a shade of white).
After making the Levels adjustment I go to work on the image. When I have completed all the step required to repair, retouch or enhance the image I reset it back to Adobe color space before sending on to RetouchPro. I do this by using Image\Mode\Convert to Profile set to "Adobe RGB 1998". If this step is missed the image (in "AIM RGB = Trinitron D65 G1.00" color space) will be much too dark when someone else looks at it on the WEB.

In conclusion, I personally believe there is a noticeable advantage to using a linear color space. This is not a complicated process but your computer must be properly calibrated and the conversion steps need to be completed correctly.

As this is just a brief introduction to the topic, further information (212 pages, be prepared to spend some time here) on system Calibration, Editing Techniques, examples and free tools can be found at the AIM (Accurate Image Manipulation) Color Calibration Web Site at http://aim-dtp.net/aim/index.htm . Although the information is of a technical nature, the writing is generally quite good and easily understood. They also have a complete section on Adobe Photoshop Calibration and Issues.

Ron

[ghoffman]

Ron,

Thanks for the technical details. I look forward to learning lots more at this level. At this point, I'm in way over my head, so my question may be way off base.

In your Using Linear color space - the ins and outs section, you stated that after converting from "Adobe RGB 1998" or "sRGB IE C61966 2.1" color space to "AIM RGB = Trinitron D65 G1.00" color space you darken the image and normalize the whites. But when you are done working on the image, all you mentioned was changing the color space back to "Adobe RGB 1998" color space.

Without first undoing your normalized whites and lightening the image won't it still be too dark for others? i.e. why are the extra steps required when converting one way, but not the other?

Thx!

Gene

[Doug Nelson]

Ron is being kind when he says this is viewed by some as overkill. In fact, it's the subject of some of the hottest flamewars on the Photoshop usenet group. While I have no idea why so many care so passionately about colorspace, it is a fact that they do.

That said, I feel the need to mention that Adobe does not endorse the linear colorspace theory. Their own recommendation is (surprise) AdobeRGB.

While I hope this thread lives a long and informative life, I will be watching it closely for any signs of excessive heat.

[Ron]

Ghoffman,

That’s a good question. When the initial image opens in Photoshop with a profile mismatch (my working color space = "AIM RGB = Trinitron D65 G1.00" vs "Adobe RGB 1998") the query screen actually ask if you want to "assign" the working space. Later on when I've finished correcting the image I used "Image\Mode\Convert to Profile" to change it back to the Adobe color space. The difference is between "assigning" and "converting". With converting - the image appears exactly the same as before converting (but the color space has changed).

This question got me to thinking if "converting" doesn't physically alter the appearance of the image, could I opened the image in "Adobe RGB 1998" working space and then converted the image to "AIM RGB = Trinitron D65 G1.00" before working on it? Would that spare me from having to use Levels to correct the brightness?

Yes, and Yes

Your question made me put on my thinking cap; as a result I've found a better way to initiate the process. From now on I'll opens the image in its embedded working space i.e. Adobe RGB 1998 and use Image\Mode\Convert to Profile to change it to "AIM RGB = Trinitron D65 G1.00" color space (and use the same process to change it back again). Thanks Gene..

Doug; you may be right about the volatility concerning this subject but personally, I've never seen any conflict.
I discovered the AIM web site by accident while looking for better corrective filters. With some time on my hands I looked over the information. To me, it clearly showed compelling reasons for using gamma 1.00 color space. So I do.

Ron

[john_opitz]

That's Mr.Timos' site. Even though I have nothing against Mr. Timo. The Expert's don't agree with Mr.Timo, with the 1.0 gamma monitor. He was once on an e-group(I think it was Adobe) and he was thrown off of it. I don't know why(maybe about misleading information) I don't know for sure. All I know, Bruce Fraser does not agree with him. Plus "many" others don't. Andrew Rodney.
I have been using A.P.S. since 5.0. And no one, agrees with it.
Now you can use,what they call a false or imposter profile. Like, using Adobe rgb and changing the gamma for that profile to 1.0 and keeping the x,y primaries the same. This is done for the purpose of lightning ones file.If the file itself is dark. Mr. Dan Margulis and Mr. Bruce Fraser goes into this in their articles and books. But this is not the same as a 1.0 gamma monitor.

[john_opitz]

Here this is what I was telling you about.

adobe forums

You have to sign in at adobe to view the thread.

Like I said. I have nothing against Mr. Timo.
Just backing up the statement I said.

[DannyRaphael]

Wow... 107 replies to the thread John posted at adobe.com. I see what you mean about it being a passionate subject, Doug.

~~~~~~~~~~~

"Less Filling!"

"Tastes Great!"

[john_opitz]

<<While I hope this thread lives a long and informative life, I will be watching it closely for any signs of excessive heat.>>

Hey, Man ! This is going to be better than watching WWF Smackdown. Let's get ready to Rumble !

[Stephen M]

Timo has some good stuff at his site - but I do not agree with his linear gamma crusade for the overwhelming majority of images...which must mean that there is a rare case for this low gamma editing space, rather than a higher number as in more common spaces.

As John mentioned, there are two issues which should not be confused when it comes to gamma tricks -

i) Assigning or presuming a profile as a 'trick' for altering the definition of a file without editing the files numbers. Similar result as editing, but no numbers have changed - only their interpretation. When this assigned or presumed gamma is converted to another mode - it then takes affect and the numbers in the file will reflect this change.

http://ep.pennnet.com/Articles/Artic...alse%20profile (free rego reqd).

ii) Working in a RGB editing space that has a 1.00 gamma.

In 8 bit data, there are 256 discreet levels to play with - and different gamma spaces have a 'sweet spot' in different parts of the gradation range. As an extreme, 1.00 gamma concentrates the data in the extreme highlights to quartertones - providing lots of gradations but destroying the midtones to shadows with posterization and other errors. Even high bit edits do not stop this from happening. At the other end of the extreme is a 3.00 gamma space, where all the data is concentrated in the shadows.

1.8 (slightly darker) and 2.2 (slightly lighter) spaces are more common. 1.8 is a lighter gamma, but the effects on data produce a slightly darker image, while 2.2 produces a slightly lighter image. These gamma spaces expend their bits in the general midtones - which makes them ideal for 99% of edits.

If you have a image of a polar bear in a snowstorm or a egg on a white wall - then a gamma lower than 1.8 might just give you an edge when it comes to preserving tonal detail. Sadly Timo does not qualify his findings with this important fact.

It is very easy to see what is going on - you should not accept _ANY_ of these opinions at face value (mine or others)...quantify and qualify things for yourself.

Get a high/low and 'average' key RGB image for a quick test. Change the test files to LAB mode so that you have an 'unbiased' master with no gamma (Kodak PCD are also good for this too).

Then do some common edits and sharpening etc in the 1.00 gamma space (convert a dupe from LAB to the profile in question) . Perhaps record an action or save the edit settings files so that the same edits can be applied in the other colour spaces. Save/print etc.

Next do the same edits in your regular gamma (1.8 or 2.2) and then perhaps try a 3.00 gamma space.

Then compare all three images from all three gamma spaces and note what is going on!

There are other tests which demonstrate what gamma and edits do, but this is a real world test - rather than the math which I can't remember.

Regards,

Stephen Marsh.

[john_opitz]

Hello Mr. Marsh,

Been wanting to contact you about those av. curves you use for the images from P.P.S. 6.
It's been a few years since I've been at Mr. Timos' site. The only time I ever went their(other than today) was when 5.0 first came out. And yes their is some good stuff at his site. Have you ever tried the space filters or something like that, that he has? Btw. You also have some good information(and links) at your site.
With the real world and the math part. I can understand and relate to both. Being an electronic engineering major in college, (now a photographer) I can relate to those, that have that outlook. And. Again. I can relate to the real world outlook........... Does it look good.

I don't think that the 1.00 gamma monitor thread is as heated as the 8/16 bit threads though. The 1.00 gamma thread was a one time deal(I think). But the 8/16 bit threads go on forever. We will see,I guess. When Mr. Margulis' book comes out. But I don't think that will change the camera makers mind though. To process images in 16 bit(optional).

[Stephen M]

Greetings Mr Opitz, when I joined this list I knew we would be conversing one day - since we are both ACT members (it is easy to tell from your deeper posts to this list that you belong to the ACT list or have read various articles or books from Dan Margulis).

Alain does not have PP6 - so she does not have the CD. I offered to see a few low res files of hers that she was having problems with, since she has PP5 which did not have a CD. If you have PP6 and the CD then the offer is open to you too, just let me know the images that you would like to see how I write the curves for, and I will send you the ACV file (up to three images).

I have kicked various things around at Timo's site, but I do not think that I have made any real use of them for production - only to try see where he is coming from.

Thanks for the compliment about my humble site - it is in need of an update but this will take more time to get to than I would like.

Unlike the regular and high bit debate, the 1.00 gamma editing space can be quickly proven to be a very poor choice for the vast majority of images and users - so it probably died a quick death on that list...while the bit depth issue is a lot deeper and harder to show any visible real world results - it is a lot easier to keep the discussion alive. Theory can also show the problem issues with a 1.00 gamma space, while on the other hand theory seems to support the high bit debate (with little real world results to back up the theory).

Regards,

Stephen Marsh.

[Guest]

I went and looked at the site and it appears that one could benifit a lot from doing this aim thing, but I'm lazy and it hurts my eyes to read. What does one do? Buy a program or what.

greg

[Guest]

I guess this thread is dead, since no one will answer my Qs.

[Ron]

Seeing is believing….

All the rhetoric doesn't change one simple thing. Images turn out better using linear space.

The original file is a 16-bit/c AdobeRGB(1998) PSD image (in a zip archive) " http://www.aim-dtp.net/aim/evaluatio...0-adobergb.zip " at the AIM web space dealing with linear space induced errors (see the information) = http://www.aim-dtp.net/aim/evaluatio...rror/index.htm =

The experiment "Download the image file and try it yourself".

The original PSD image is opened into Photoshop with out changing the Color Space "AdobeRGB 1998"

Using Image\Duplicate the Image is copied twice. Now we have three copies and the first one is left as the original for comparison.

Copy 1 will demonstrate the Unsharpen Filter using Linear space AIMRGBpro Gamma 1.0 Color Space.
Copy 2 will demonstrate the Unsharpen Filter using AdobeRGB 1998 Gamma 2.2 Color Space.

COPY 1

First we convert Copy 1 to the AIMRGBpro Color Space Using Image\Mode\Convert to Profile....
Next we convert to 8-bit Color Depth (most filters only work in 8 bit space) with Image\Mode\8bit/Channel
Finally we apply the Unsharp filter using Filter\Sharpen\Unsharp Mask (Amount 500% Radius 0.5 Threshold 0)

Copy 2

First we convert to 8-bit Color Depth (most filters only work in 8 bit space) with Image\Mode\8bit/Channel
Finally we apply the Unsharp filter using Filter\Sharpen\Unsharp Mask (Amount 500% Radius 0.5 Threshold 0)

As you can see, other than converting Copy 1 to Linear Color Space both images had the same effects applied.

btw = The results look similar at 16-bit Color Depth

Now look at the final comparison. Which image do you think looks better? Try it yourself if you don't believe your eyes.

[Ron]

fugitive;

You don't have to buy anything to take advantage of the AIM information. The site distributes both of the Linear Color Space profiles "AIM RGB = Trinitron D65 G1.00. ICM" and "AIM RGB Pro = WideGamut D65 G1.00 ICM" FREE and they can be downloaded at http://www.aim-dtp.net/aim/photoshop/index.htm as a zip file at http://www.aim-dtp.net/aim/download/aim_profiles.zip

You should also check out their information on Editing Techniques at http://www.aim-dtp.net/aim/techniques/index.htm

Ron