PDA

View Full Version : Joseph Holmes Ekta Space P5



Kirk Gittings
9-Jan-2006, 09:50
Is anyone using any of Joseph Holmes' spaces (like Ekta Space P5) for their scanning profile? Tell me about what you think the real advantages are if you would.
Thanks.

tim atherton
9-Jan-2006, 10:11
Kirk,

I've used it because a drum scan shop I use have picked it as their color space (as a result I also did some of my own scans using it)

I read all the blurb about it and it's potential benefits.

In theory it is supposed to offer some advantages.

In practice, with my printing workflow - I couldn't say I've ever noticed any major difference. Perhaps if I ran some rigorous side by side tests? Or on a particualr image where it's benefits especially applied to that one image. But I've given up using it on a regular basis, apart from converting the drum scans across to Adobe RGB

More based on instinct than empirical testing.... :-)

Doug Dolde
9-Jan-2006, 12:47
I use it because I also have Joe's Chroma Variant profiles (both Chrome 100 and Ekta Space) and use them instead of normal saturation methods. The variants are assigned profiles (rather than converted to) and you can change them anytime in the workflow without and file degradation (since the don't affect the file itself).

To quote Joe Holmes:

"The full set contains 17 profiles, including the master profile, Ektachrome Space, J. Holmes. When assigned (not converted into the various variants, the image chroma is increased or decreased by an amount equal to the lable of the variant (from minus 50% up to plus 50% in increments of about six percent).

The effect can be seen by converting an image into Lab in Photoshop, and then using the Hue/Sat tool to increase "saturation". Compare that effect to the same adjustment with the Hue/Sat tool in RGB mode. The latter is the problem and the variants are the solution. Image colors retain lightness much better as their colorfulness is being adjusted. And yet the system allows for the color to be adjusted in this way at any time in your image editing process, non-destructively, unlike conversion into and out of Lab, which both means the adjustment you made in Lab becomes permanent and not undoable, but involves a conversion penalty unless the file is a 48-bit file.

I use them for every image and they are a revelation for many photographers who use them."

steve_782
9-Jan-2006, 13:28
Yes, I have the profile and use it on my Imacon. The colorspace is huge. Easily 2x larger than Adobe 1998. However, that's also part of the problem. You can't currently print out that large of a colorspace. You can edit in PS in the larger colorspace, but, if you compare the Ektaspace colorspace to what you can reproduce on paper - in the end it doesn't matter, you end up having to make the Ektaspace colorspace fit whatever paper you're printing on.

Another problem is that the profiling targets are all made in Adobe 1998. This means that if you want to try profiling a paper using a larger colorspace, you have to create the targets yourself for paper profiling to use Ektaspace as the embedded profile in the target.

While this can be done, I haven't convinced myself that the extra work is worthwhile for the following reason. I haven't found a paper that can accomodate the Adobe 1998 colorspace much less a colorspace that's nearly 2x larger.

His comments on editing the image in Ektaspace and making adjustments are true, but the question is can you translate that into a print? Or, do the colors merely get reassigned through either the perceptural or relative colorimetric print settings anyway?

This can all be easily seen through color analysis software. I have both Monaco GamutWorks and X-Rite ColorShop X; either program will produce 2D and 3D plots of colorspaces and plots of images within a colorspace or profile. Once you see these rendered in graph form, you can easily see the problem with translating from Ektaspace to a paper profile.

The reason I use the colorspace is for some of the editing advantages, and the idea that maybe in some future print process I'll have a larger colorspace to work with.

bglick
9-Jan-2006, 16:31
Steve, nice post... I came to the same conclusions several years back when contemplating using Joes Color Space. So what is the motivation, since Joe prints on Epson printer anyway? He must have had something in mind for this color space?

Jack Flesher
9-Jan-2006, 20:05
The advantage to being Joe's space to begin with is then being able to apply his saturation profiles. By doing so, you alter saturation without changing contrast or losing color data, both of which can (and often do) happen when using the saturation, contrast, levels and/or curves adjustment sliders.

bglick
9-Jan-2006, 22:44
Jack, so i guess this falls under the " more ceiling space for edits" category, just like 16 bit vs. 8 bit issues? Ok, so this is could be a potential benefit, is there any drawbacks?

Kirk Gittings
9-Jan-2006, 22:58
How complicated does it make the workflow and how expensive are his programs?

John Hennessy
9-Jan-2006, 23:02
Kirk,

I use Ekta Space P5 as a working color space but not as a scanning profile. For scanning, I use a profile made with the profile-creating capabilities of Silverfast which uses a target specific to the film I use.

Am I missing something in your question? Or in my workflow?

Jack Flesher
10-Jan-2006, 08:38
To be clear, a color-space is a color-space, regardless of whether you are trying to represent it with 8-bits or 16-bits of data per channel. But as for headroom, your comment is on the mark.

The problem really has been that conventional "standardized" color spaces have not been matched to the photographic medium. For example, Adobe RGB is not big enough to contain all of the colors that can be output by the current line of professional digital printers, while the Profoto space is so huge that it contains colors the human eye cannot even see! sRGB is tiny and matches only what a basic computer monitor from 15 years ago could render. Joe's space OTOH, originally matched the color gamut of color transparency films and by odd coincidence (or not), this gamut is very close to what current digital printers can match. (Actually he has a new one, Chromespace 100, that is a bit more tailored to the digital workflow. As for cost, I believe the entire set is $100, Mac or PC.)

So by using Joe's space, we have a more optimal color palette to begin with, which allows us to maximize efficiency; not too many extra colors we can't use, and not too many lacking. With the saturation profiles, we have the ability to alter color saturation while at the same time keeping all the colors at the same relative brightness so contrast remains stable. We also stay inside our working space and don't "lose" any color data. By contrast when you start pushing pixels around digitally with a program like PhotoShop -- for saturation, contrast, brightness or hue changes -- you can easily push colors outside the edge of the usable portion of color space (called "clipping"). When this happens, that data is simply lost -- it is like a pipe backing up and all the subtleties of color that fall outside your space take on the last hue available before they landed outside the space. How those border colors ultimately get represented gets us into discussing rendering intents, which another topic all on its own. Suffice it to say, this become a less critical issue when you keep all of your colors within a tuned space like Joe's to begin with.

As to why you don't see much difference much of the time: Either all or enough of your colors remained inside the range of your output space to begin with (indicating a good capture), or those that fell outside were limited enough that the rendering intent chosen did its job and successfully camouflaged the changes. When you hear people start to complain they can't get their sky the proper blue, or the clouds look funny, or my greens are off, you can bet they are dealing with clipped color issues.

Hope this helps,

Jack Flesher
10-Jan-2006, 10:06
PS: If you scan into a large space like Profoto, then when you do a convert to Joe's space you will not be giving up much since the bulk of the original capture is still inside Joe's space. However if you scan or capture inside say Adobe RGB, you may in fact clip some of the color information during capture and this can never be regenerated, so it's still lost when you convert to Joe's space.

Silverfast (and most quality raw converters) allow you to define the output colorspace -- so selecting Joe's space at this stage and setting Joe's space as your PhotoShop working space makes sense.

bglick
10-Jan-2006, 10:07
Jack, wonderful post..... I am curious of your comments....

You mention that Joes color space can match that of most chrome films, which of course, makes perfect sense for its existence. However, then you mention, it also matches the color of digital print output. How can these two be true? What digital print output can match the gamut of film? I am very very curious of your response on this.

Also, since so much of digital processing was a result of scanned film, how is it possible that Adobe did not create such color spaces to accomodate this for so many years now? Seems odd they continue to make new color spaces throughout the years, yet never, created one for color chrome film? Why didn't makers of the digital darkroom printers create such? Without this, it could continually represent the bottlneck to the final output....and when you have a $250k printer sale that is dependent on output quality, well, thats a good incentive!

How does digital capture compare with scanned film, as it relates to the size of the color space required? Is proRGB bigger then Joes space? If the two are radically different, does it make sense to keep your digital capture files in the color space which the camera maker provided and your scanned chrome film in Joes space? TYIA

Kirk Gittings
10-Jan-2006, 11:48
Interesting points Jack. I sometimes work with a truely great printer here sometimes when I need a large print. I found however that what he was printing large, I could not get smaller on my 4000. The reason was a worflow similar to what you describe above with scanning in ProPhoto and then converting to Joe's space for printing on his 9600. His prints had a glow to them that I could not match with my workflow.

steve_782
10-Jan-2006, 11:49
Pro RGB is so big, it won't even fit on any of the color analyzing software programs that I have. It's literally outside of the entire L*a*b* space - which is why the comment that it can reproduce colors you can't see is absolutely true.

The comment on the Ektaspace capturing all of the range of a the film is correct - which is why I use it for my scanner colorspace. I'm not sure what color printer Joe is using (I'd like to hear more about that), but with an inkjet system, I haven't been able to reproduce all of the Adobe RBG colorspace. You run out of gamut from the base material's (paper) reflectivity as well as the gamut range of the ink itself.

If there is a way to extend an inkjet printer's color reproduction capability - I'd certainly like to hear how to do this.

Doug Dolde
10-Jan-2006, 15:02
There are actually two sets of profiles. Best if I quote Joe's email for all the details.

""Ekta Space PS 5, J. Holmes" (aka Ekta Space) has not been changed.
I always had two master profiles. The one I changed was Ektachrome
Space, J. Holmes, which became Chrome Space 100, J. Holmes.

I have new chroma variants for each of them. The two master spaces
have different tone curves, but are otherwise functionally identical.
The chroma variants share the tone curve of the master profile for
which they were made. That's why if you have a file in Ekta Space,
and you assign Chrome Space 100, the image's tone curve changes,
opening up the deep shadows.

This property of Chrome Space 100 (and Ektachrome Space before it) is
a useful one for people who have gotten scans that were saved into
Ekta Space PS 5, and who wanted more open values near 0,0,0. If the
scan already has enough shadow detail, then you're better off using
my new set of chroma variants for Ekta Space PS 5. I used to
occasionally sell a set of 6 variants for Ekta Space PS5 for $50, but
have built a much larger and also revamped (better) set of 23 that I
sell for $80. Since you already have the other one, I'd sell you
that one for $50, because the combined price of both sets has been
$150 where I've sold both together."

Kirk Gittings
10-Jan-2006, 15:39
If allot of this is new to you as it is me, here are a couple of articles by Bruce Frazier that relate to the issues above:

http://www.creativepro.com/story/feature/6541.html

http://www.creativepro.com/story/feature/8582.html

Jack Flesher
10-Jan-2006, 15:58
My appologies -- I always get into trouble when trying to keep a complex issue simple ;)

The Epson Ultrachrome and K3 inks have a surprisingly large gamut. For example, they are both larger than AdobeRGB in most colors, though it does fall inside Adobe along the yellow corner -- regardless, both of these inks are larger than what current monitors can display. If you look at a map of Joe's space and of Epson's K3 space, you will find they are very close, though Joe's space is of course larger, containing even some non-visible colors.

My main point was the fact Joe's space is not as overly huge as Profoto, but larger enough than Adobe so it at least contains what current printers can print. This in turn means it is probably a practical color space to be working in, at least given present day technology -- and assuming you won't want to re-scan 10 years from now when things improve even more ;)

As to why Adobe's space does not meet current printers, I really don't know, though I suspect they standardized it to match output capabilities at a time when processing images in 16-bit was still a futuristic concept and well prior to Epson's breaking the printer gamut barrier with Ultrachrome ink in the 2200/7600/9600 printers.

Lastly to Kirk's point, that "glow" was most likely the fact that using Joe's space instead of Adobe allowed the 9600's UC ink to display the additional color it was capable of rendering; before that it couldn't print what it wasn't given.

Again, sorry for any confusion,

Kirk Gittings
10-Jan-2006, 16:03
Well said Jack.

"Lastly to Kirk's point, that "glow" was most likely the fact that using Joe's space instead of Adobe allowed the 9600's UC ink to display the additional color it was capable of rendering; before that it couldn't print what it wasn't given."

What I believe exactly.

And Bruce space? Any thoughs? Smaller gamut than Joe's space I believe.

Jack Flesher
10-Jan-2006, 16:54
Re Bruce space...

I have never tried it...

In general, when I'm using something that aint broke, I usually don't try to fix it ;)

Kirk Gittings
11-Jan-2006, 00:46
Perhaps a dumb question, but I assume you would have to develope custom printer profiles to take advantage of the wider gamut of a larger color space like Joe's.

Jack Flesher
11-Jan-2006, 20:36
Not a dumb question at all... But the answer is no ;)

A printer profile (or more accurately a paper profile for a given printer) is nothing more than a way to standardize that printer to print the exact colors it actually thinks it is printing on that specific paper. It is really more a function of the paper's base color, the actual ink hues and the way a given printer lays those inks down than anything else. Hence the working space has no impact on it -- in fact many test prints are printed from LAB to insure a specific, repeatable color value between all imaging systems. (In actuality, it is only RGB or CMYK values that are meaningless without a specified profile.)

Surprisingly, different printers of the same model from the same manufacturer can each differ slightly. Hence while the paper or printer manufactuer's "canned" profiles may be close, it is unlikely they will be perfect. Which in turn is why those of us that print critically all build (or pay for) our own custom-generated paper profiles.

Kirk Gittings
11-Jan-2006, 21:02
Thanks again Jack. Why do you have so much knowledge in this area? Are you just real demanding for your own work or do you have a service bureau?

Jack Flesher
12-Jan-2006, 08:36
I was once a dedicated LF shooter. Then three years ago began dabbling in digital. Two years ago I went head-first and sold all my LF gear to fund a high-end DSLR, tilt-shift lenses (had to keep some movements!) and a computer and printer capable of procesing and printing large prints. Then in the normal pursuit of getting the best results with that gear, I ended up learning all this...

However it is now a few years later, and having missed the enjoyment of making my captures while viewing them upside-down and backwards, I have recently re-entered the LF fraternity.

Cheers,

allen mckinney
12-Jan-2006, 10:11
Jack:
To use these new large color spaces to maximum advantage how would you recommend scanning
should be done for them. Assuming high end scanners.
Allen McKinney

Jack Flesher
12-Jan-2006, 10:44
Make sure you are using a colorspace large enough to capture the entire gamut of the film you are scanning, and scan at the highest bit-depth possible. (Highest is almost always stated to be 16-bits per channel, though it is rare that you actually get a full 16-bits. However, even 10-bits per channel is a big advantage over 8 -- 2^30 compared to 2^24 -- and most newer scanners will probably give you at least 12, and some even 14-bit output. )

Next, do not compress the shadows or highlights during the scan -- rather leave a skosh of floor below the darkest pixels and a skosh of headroom above the brightest pixels. In this fashion you are not "clipping" (throwing away) any of your raw image data. You can easily process them out in PhotoShop after the scan to improve the file's appearance in print, but assuming printers continue to improve you will then still be able to replicate these tones in the future if you so desire.

Lastly, I think too many people think they should make the scan look as good as possible and thus limit the amount of post-processing later. IMO this is counter to good workflow. In other words, the ideal scan may look pretty darn "flat" relative to the ideal print you'd make from it -- but this allows you the most flexibility in how you process the file for all future images you might want to make from it. As long as all the data remains available, you can decide what to do with it later. If it isn't there to begin with, there is no going back -- unless you want to re-scan everything.

Kirk Gittings
12-Jan-2006, 11:07
I agree with you Jack except in one m.o. If your intention is to make rich contast b&w prints that require allot of tone manipulation. It is best to do your major moves in the scanning stage. For instance It is better to apply a steep curve in the sampling stage rather than PS otherwise you get gaps in the information even in 16 bit later on. I have tested this extensively, because my aesthetic requires allot of tone and contrast manipulation (http://www.gittingsphoto.com/). I get in allot of trouble down the workflow road. if I don't scan with an approximate curve rather than flat.

By the way, I assume that in b&w there is no adavntage that I can see in scanning using a large color space. Many of us do scan in RGB and then discard all but the sharpest less noisy channel. Any thoughts?

allen mckinney
12-Jan-2006, 13:35
Jack:
What would be the best solution if I need to make master files (16bit rgb) and
also needed to produce cmyk separations for publication? I have, in the past,
noticed that if I scan rgb then convert to cmyk in ps, I end up with sever lose
of "shape" (colors) in the reds, whereas if I scan directly into cmyk this doesn't
occur. Are there tools available to allow this conversion without the severe
clipping done in photoshop?

Kirk:
Do you have an idea as to what the reason is for why b&w needs the steep curve?
Other than your testing.

Kirk Gittings
12-Jan-2006, 14:07
Allen, it is not b&w in general. It is my b&w. My aesthetic. When I was developing to print traditionally, I always preferd to print on grade three paper. But rather than always doing a +1 developement and increasing grain, I prefered to do it with paper grades. Now that I am printing digitally I get the contrast by applying a steep curve to the file. Look at my site. I think you will see what I mean. www.gittingsphoto.com.

allen mckinney
12-Jan-2006, 14:39
Kirk:
I can certainly see the need for a vertical gamma. I guess my question has more to do
with why at the scanning rather than the retouch stage. If the tones are fully separated
and both ends are there I'm trying to understand the why of it.

P.S. Your work is really outstanding!

Jack Flesher
12-Jan-2006, 22:01
Kirk: Are you scanning your B&W in 16-bit RGB? I have a hard time believing you run into poterization issues even with a huge S curve applied in PS if you are. However, if you are scanning them in 16-bit grayscale, I can see where you would have that issue due to insufficient data.

Alan: Without knowing specifics, I would think that if you were scanning into a large space like Joe's or Profoto, then you would not have issues when converting to CMYK. However, it is my understanding that Adobe RGB does not contain all of CMYK clips some significant colors -- this may be your issue.

Kirk Gittings
12-Jan-2006, 22:42
Jack,
Of course I am scanning in 16 bit RGB. It is not so much a question of posterization as increased noise in the transition areas between stretched tones that appears as very granular areas in prints. I am not the only one who has faced this issue. Everyone bw printer I know who doesn't print flat runs into this. For over a year I carried on a dialogue with some of the top bw printers in the business over this problem and we all came to the conclusion that to avoid it you have to make your major curve adjustments in the scanning stage.

Thanks Allen, "the tones are fully separated" and PLACED (Zone System term) if my curve is approximately applied in the sampling stage. This of course requires that you know how you want to print something before you scan it. I do.

Jack Flesher
13-Jan-2006, 09:26
Kirk: I just wanted to be sure I understood your workflow completely. My next question would be how are you applying the curve in PS? Are you applying it via an adjustment layer in the "normal" blend mode? If so, I would suggest you try changing the blend mode to "luminosity" and see if that helps ;)

The alternative is to try scrubbing off the blue channel and curving just the R and G, since blue is where most noise resides -- in 16-bit you still have enough data to push.

But most importantly, if your current workflow is working for you, then IMO there is really no reason to change it.

Kirk Gittings
13-Jan-2006, 09:51
"The alternative is to try scrubbing off the blue channel and curving just the R and G, since blue is where most noise resides -- in 16-bit you still have enough data to push."

Completely depends on the scanner-with some the green is sharper and cleaner (1800f) and with some the blue is (4990). You can't generalize. I have tested this with nearly 20 scanners (3200, 4870, 4990, 1800f, 9950f, 8000).

I have not tried the luminosity trick, but I don't need to by simply by doing my major curve adjustments in the scanner. A number of people agree with me on this including Bruce Frazier and George DeWolf. I did not come to this conclusion alone.

Rich14
9-Jul-2006, 15:10
Kirk and Jack,

I don't know if this thread is still active or if new posts alert your email. I have recently joined the forum and would like to add something here.

Kirk has found that if he applies his required tone compression/expansion curve in his scanner settings vs the same moves in PS, he avoids holes in the histogram. The question why the difference between the two methods was not answered.

I believe the answer is clear, at least if the scanning is being done on a drum scanner. I'm going to assume that the same applies to a high-end flatbed, though I'm not sure, and I don't know what kind Kirk is using.

In Photoshop, even with a 16 bit file (12 bits of image depth), it's possible to apply a severe-enough tone compression/expansion (adjustment in Curves) move to drop enough bits to see "gaps." This happens with very little adjustment when dealing with an 8 bit file, and while it's less likely with a 16 bit file, (1/16 as likely if there are 12 bits of image data - 4096/256), it certainly can happen.

On a drum scanner, any adjustment of gamma, highlight, shadow, curves (in any or all of the RGB channels), changes the behavior of the color (log) amplifiers to bring the 16 bit scan in AT THOSE ADJUSTED SETTINGS. So, one has the full 12 bit range in the 16 bit file at the desired characteristics. Such a file SHOULD have a much smoother range of tones on printing than a low-contrast scan adjusted in Photoshop to the same characteristics.

Of course, if the scanner has been profiled for its default condition, then doing a scan with any settings other than default invalidates the profile. A profile should be made at the new settings for use with the adjusted scan, or others made the same or similar adjustments.

Rich

Hugh Sakols
9-Jul-2006, 20:09
I scan using the profile that came with my scanner and then I assign the Ekta Space Profile. I have found that Ekta Space fits my images more times than not than ProPhoto. However, in photoshop you can experiment with what working space looks best.

"Perhaps a dumb question, but I assume you would have to develope custom printer profiles to take advantage of the wider gamut of a larger color space like Joe's."

I just use profiles developed by Bill Atkinson (the color management guru) that are available free of charge on his site. In fact I recently purchased an Epson 4800 printer because his 9800 profiles work beautifully with it.

Kirk Gittings
9-Jul-2006, 21:17
I do not think that assigning an expanded color space like the Joseph Holmes in PS to an image that is scanned in Adobe 98 or sRGB has the same charteristics as a file that is originally created in the expanded color space. My tests show very different files.

Rich14
10-Jul-2006, 06:49
Hugh,

To get your scan into Ekta Space, you need to CONVERT from your scanner profile space to Ekta Space. Just assigning the Ekta Space profile won't do it. I'm not sure how your scanner handles the profiles, but if the scan file is tagged (embedded) with the scanner profile, then the conversion can be done on opening the scan file in Photoshop, or once it is in Photoshop.

I use Ekta Space PS for the most part. But there are reasons not to use it, and there are many photographers who use Adobe RGB and sRGB (yes!) as a working space who produce dramatic prints that are very pleasing to them and their clients, who would complain that images held in Ekta Space lack saturation and and "punch" in their workflow.

As far as seeing "what working space looks best" in Photoshop, that's not readily done. Changing working color space between large gamut profiles or even comparing a large gamut space to a smaller space such as Adobe RGB, the changes that happen are in colors that lie far outside the monitor's gamut. So nothing seems to change on the monitor.

It is possible to test, in Photoshop, if saturated colors are being clipped by a particular color space, but nothing about the screen appearance will alert you.

A soft proof can be done, comparing several working spaces to the output profile for the printing paper (or other output condition) you use. But the monitor screen is difficult to adequately evaluate as the differences can be subtle or confusing for a number of reasons.

Many people can't see differences even in printed output, leading them to conclude that photographers who labor long over issues such as the right large working color space to use also spend time counting numbers of angels dancing on pinheads.

Rich

Hugh Sakols
10-Jul-2006, 14:12
Rich,

Thanks Rich for making the clarification between assigning vs converting profiles. Yes my scanner profiles are tagged thus I convert to Ekta space.

baked gamut
27-Jul-2006, 22:44
Hello Everyone!

First time poster here. I must admit to not having read _all_ of the posts in this thread as well as I no doubt should have, but, I have some real world experience that might apply and or help those trying to figure this out.

I'm involved in a fairly high volume fine-art reproduction operation. I live and die by color-space and gamut issues. Firstly i'll say that Ginormous color spaces are abstractly very cool and do have a place in the real world. I'll use Ekta-Space for a customer with the caveat that it really is for "archival" use only. By Archival I mean that nothing for the forseeable future will ever be able to touch them (Joe/Pro) in terms of real world reproduction. This includes your monitor. It's difficult for me to judge without the help of mathematics and wire-frame models just how big a color space is even on the _best_ of CRT monitors when the monitor can't even display these colors, let alone _any_ printer print them. Your monitor, not to stick a firecracker in the hornets nest here, is really only, and roughly at that equivalent in gamut to sRGB. It's true. I don't care how much you paid for it (well, o.k. there is the Eizo that may or may not be really, fully, able to display 98 at the thrifty price of $3000.00)

Right now i'm working with a printer that REALLY can go beyond 98. I'd show you my profiles, but, they're Hexachrome, so... sorry, you'll just have to trust me here until Adobe decides that adding support for Hex is worthwhile. But, suffice it to say, there are colors that I cannot see on my monitor (LaCie E-blue profiled with Monaco) that will reproduce perfectly at the printer (D'vinci system, using ErgoSoft StudioPrint and color GPS) that I cannot display. It's a bit strange to get used to at first. You really have to learn to trust your input profiles.

So... long story short. If your reproductions are clipped in color or density or gamma or all of the preceeding and yet, there they are on the monitor... something else is wrong by my estimation. Could be many MANY things. The most likely culprit being the gamut limitations of your printer, the profile you've built for the media your printing on or a combination of the two. It also could be solar flare activity ;)

If your dealing with Mega-Turbo-Big color spaces, forget about what your eyes are telling you, your portal into that world is only as rough as sRGB. I'd also say, forget about scanning and working in 8-bit also, these big boy spaces are just too large a hammer for subtle changes. Here's what I do.... I'll take a capture from a BetterLight Super 6K-2, for example, completely raw *in 8 bit* Having built a profile from a Macbeth color chart at the same settings used for capture time after time, bring it into PS and assign the input profile that i've built for the Betterlight. Then before doing ANYTHING else convert it into my working space (98). Cool. Good to go. Now... After judging the quality of the scan pre-edit i'll take it directly into L*a*b and do basic contrast adjustment in the *L* using the time honored levels method. If any sharpening needs to happen it happens here, in the *L*. Back to 98 for overall color correction, at this point _no_ gamma or density corrections are made. If you need them at this point, start over, no big deal. Proof. Then based on what you see on the paper, fine tune the color. If density or gamma need adjusting, move the file back into Lab make the changes in curves, then back to 98. Proof. Works 9/10 times for me. But then, the printer i'm using can make nearly all of the colors i'm actually able to see, and bear in mind that these as many times as not are oil paintings. Pretty harsh stuff in the world of color reproduction.

So, at this point i'll conclude my mini-rant. As unfinished as it may be for another day. For those that have waded through it, I welcome your comments and experiences.

-b

Rich14
28-Jul-2006, 08:08
b,

I'll agree with much of what you've added.

However, if I understand your workflow to involve bringing your scan into PS as 8 bit and taking it into Lab and back to RGB even once, I'll have to object. Any color moves at all (Curves. Levels, etc) in 8 bit throw away data, lots of it. Taking the file into Lab suffers serious loss due to quantization (rounding) errors. The trip into Lab is bad enough. Coming back into RGB really tears the image apart. And it seems you may make the move more than once?

These moves must be done on high bit data (REAL data). That means 12 or 14 bits of scan information in a 16 bit format during all edits until the image is ready for output.

You can accomplish the same thing in RGB that you are doing in the L* channel edits. Make the "gamma changes" in RGB, then do a Fade to Luminosity. The file never goes through a RGB-Lab-RGB transform, but the end result is the same as though you had been working in Lab mode.

Rich

baked gamut
28-Jul-2006, 12:50
Rich,

You know, I just tried the Fade to luminosity (and by that I mean making a dup layer, setting it to luminosity making the contrast changes in curves and then fading the edit) method and hmmmmm.. not too bad. It does work, though it's a bit more touchy-feely than I like.

Your point about the trip from Lab to RGB is well taken and understood. However, the proof being in the pudding of the final print, I have yet to see degredation or even really the slightest shifting of the image even after making the journey as many as 5 times. You can see the decay in the histogram, for sure..... but the prints remain unaffected. Customers happy (and I have some VERY picky customers) and my head doesn't hurt so much for trying to manage the invisible. I'm going to try to adopt this new luminosity thing into my workflow for a week or so and see how I feel later.

Thanks for the input and suggestion, i'll try to keep you posted on the real world results.

-b

Tim Lookingbill
28-Jul-2006, 13:58
Add saturation without affecting luminance and a trip to LAB space by applying saturation and then "Fade To" set to Saturation or Color. This works really well with brightly lit scenes with colorful objects like flowers or folds in clothing where you don't want the highlites to get brighter while adding saturation to mids.

chris jordan
28-Jul-2006, 16:43
There's another reason to use Ektspace: It converts seamlessly back and forth between RGB and Lab (Joe designed it that way on purpose). There are some things that can be done to control color in Lab that can't be done in RGB, so I jump back and forth between the two a lot. With colorspaces other than Ektaspace, there can be some image degradation when making the switch, but with Ektaspace it's flawless.

~cj

www.chrisjordan.com

QT Luong
28-Jul-2006, 17:25
What do you give up exactly when you use Ektaspace for digital capture instead of Prophoto ? In naturally occuring scenes (both nature and man-made) are there any colors actually captured digitally that are in Prophoto but not Ektaspace ? Currently I am using Ektaspace for scans (I have drum scans made in 8 bit, so Prophoto isn't a reasonable option) and Prophoto for digital, but using a single wide-gamut colorspace is certainly tempting, plus Ektaspace looks like it has some nice features.

Jack Flesher
28-Jul-2006, 20:04
What do you give up exactly when you use Ektaspace for digital capture instead of Prophoto ?

IMO you give up nothing. My point early on in this thread -- and I'm simplifying for brevity -- was that Joe-space contains all the colors we *need*. Profoto contains them too, but it also contains a lot more. Since both are limnited by the same bit-depth, the main advantage to Joe-space is that you'll end up with more color-points per bit-depth than Profoto and hence smoother color transitions and tonality in the colors that matter.

Cheers,

Rich14
28-Jul-2006, 20:35
Chris,

Conversion between Ektaspace and Lab experiences the same errors that happen with any RGB space conversion to and from Lab in Photoshop. And the more times the trip is made, the more the errors accumulate. Joe has always advocated working in 16 bit files (containing at least 12 to 14 bits of real image data) and Ektaspace has become almost synonymous with a 16 bit workflow. Staying in 16 bits reduces the final errors involved in any colorspace transforms whether in and out of Ektaspace or sRGB. But it's not flawless.

The advantage of Ekataspace is that: 1. no clipping of the typical E-6 film(s) gamut occurs when the scanner space is brought into the working space. Clipping can happen (doesn't always) when a smaller working space is used. And 2. the space is not so large (as Prophoto) that the data points are "stretched" in color space, increasing the possibility of histogram "gaps" in the final down-conversion to 8 bit color for output.

Ektaspace is "just the right size," and no larger (than the E-6 gamut).

QT,

Using Ektaspace for digital capture can clip yellows and blues of high-quality DSLRs and digital medium format backs. Right now, ProPhoto is better for digital capture.

Jack,

Ektaspace theoretically will give smoother color transitions than Prophoto when the file is output to 8-bit devices (every output device on the planet, right now), but we need Prophoto for digital capture. Ideally, we need a space a skosh larger than Ektaspace for digital, but still much smaller than Prophoto.

You'd think one of the color geeks would have whipped up such a thing by now.

Rich

Jack Flesher
29-Jul-2006, 17:42
Jack,

Ektaspace theoretically will give smoother color transitions than Prophoto when the file is output to 8-bit devices (every output device on the planet, right now), but we need Prophoto for digital capture. Ideally, we need a space a skosh larger than Ektaspace for digital, but still much smaller than Prophoto.
Rich

I agree for the most part, but the original post had to do with scannig film into Joe-space ;)

baked gamut
3-Aug-2006, 23:54
Ok, Rich....

So I promised to keep you posted on my progress with the "fade to luminosity" method. So, here it is. It's good. But not good enough, at least in my world. I've told my editors to "stay the heck away from Lab, and here's how" So, real world results in a production enviroment aren't bad.... but with input profiles that are geared to capture all shadow and highlight detail in a scan or capture, assigned to that space and then converted to our defacto working space (A98), contrast adjustment must be made, lest the broth be flat and tasteless. That one conversion into Lab for contrast and sharpening to us is without equal. We've found the luminosity method to be a bit.... flat. Sure there must be methods for improving this, but time is of the essence, and time equalling money, it's just not practical. I've been wincing every time I see an image go from Lab back to 98... I stop and say "pull a histogram on that" Hmmmm... Yeah... hmmmm... Now let's see your final on the screen..... Hey! looks great.... Ok... lets soft proof it. Yow! Looking good! print it and then lets do a monitor comparison... Geeeze... That's damn near perfect. The histo is still smooth although a bit nudged to the bright, the ZBE profile dosen't even come close to those values anyway... The customer is happy, having purchased the best print possible in this town, and goes away and then comes back.

So, in conclusion... Big spaces=good for archival purposes, though don't edit within them. Scan into them, curve them at the scanner if you must, then leave them alone for another time when all of this means something on a real and printable level. Heck, I never thought i'd see a day where I was working with a 12 color ink-set but here it is. Keep those 16 bit masterpieces in a safe place. There probably will come a time when it will be possible to do good things with all of that extra data.

Lots of luck to you all, may you find the print of your dreams. Dreams are often crafted in 8 bits per channel. YMMV. Geeking out is good, lord knows I do it, but if your trying to find a great print..... a little less abstraction and a little more profiling/printing. Really, practically speaking, great prints can be had in Srgb. The Fuji Frontier? It's an Srgb device. Check the profiles made for it.... Damn close. Crystal Archive paper? Close to Adobe 1998 but not quite, if printed on an profiled ZBE Chromira. Reds of course suffer at the utmost Chroma but hey... Red's a pain unless you have printer loaded with Red ink ( yellow and magenta will never be able to make 255,0,0.)

Once again... Typing to hear myself type. I say your printing problems reside not in the color space you're scanning/working in, but in other things. First of all, stop looking at Gamut mappings for inks/printers/paper. In the real world it dosen't mean all that much.Be skeptical. Invest in good quality spectos/profiling software. If your RIP has it's own profiling functions, use those first. Don't believe the hype. Get it out to the edge, and then realize where that edge is.


*off soap box*
thank you for your patience.
-b

chris jordan
4-Aug-2006, 10:09
Rich, there are color adjustments that can be made in LAB that cannot be made in RGB, so I make at least one jump from RGB to LAB and back in every image I work on; sometimes several back-and-forths. In Ektaspace the histogram stays flawless; Joe Homes designed Ektaspace specifically with this in mind. Just for kicks I once tried converting back and forth 50 times using a Photoshop action, and the the before-and-after images were virtually identical when viewed at 800%. The histogram changed slightly, but not in a way that affected the image at all.

Rich14
8-Aug-2006, 05:57
baked gamut,

Editing in RGB Curves followed by Fade to Luminosity is for all practical purposes the same as a Luminosity edit in Lab mode and has all the previously-described advantages of staying in RGB. You are simply more comfortable with your old workflow.

Better yet, try the edits in RGB mode using a Curves Adjustment Layer with blending mode set to Luminosity. Using the non-destructive method of Adjustment Layers is a better way to work anyway, and you'll see the image change only in luminosity as you edit the RGB Curve. You'll be able to adjust image contrast to your liking in one step.

Chris,

"there are color adjustments that can be made in LAB that cannot be made in RGB"

Of course!

When it comes to COLOR changes, there is no more powerful control than that afforded by Lab. Use it when you need it!

But the point was about luminosity changes. And that can be accomplished without ever leaving RGB. The differences between luminosity edits in RGB vs Lab can probably be shown to have some tiny mathematical consequence, but nothing that can ever be seen visually.

Rich

Rich14
8-Aug-2006, 06:23
And then there's this. Looks very interesting:

http://www.xs4all.nl/~tindeman/raw/curve_tools.html

"CurveTools: LuminanceCurve & LightnessCurve

This is unlike other common tools, such as Photoshop's standard curves or a lightness curve in Lab mode, both of which have side-effects on saturation and sometimes hue."

Rich

QT Luong
17-Aug-2006, 18:08
About two weeks ago, I emailed Joe to ask him if there were colors actually captured that are in Prophoto but not Ektaspace. He was graceful enough to give me a very quick and detailed answer. Since he was still working on the new spaces, he asked me to hold on reposting that answer. Now that he is done (http://www.josephholmes.com/profiles.html) here it is.

----------------------------------------------------------------------------------

From Joe Holmes

Thanks for inquiring. As it happens, I have recently completed a new series of five RGB working spaces for use with digital captures, in a wide progression of gamut volumes: DCam 1, DCam 2, DCam 3, DCam 4 and DCam 5. The first is the smallest and the last is the largest. 2, 3 and 4 are general-purpose spaces. 4 is very similar to ProPhoto in gamut (very very large). 5 is bigger still and just barely includes all visible colors but is overly large for general use. 2 is 25 percent larger in Lab gamut volume than Adobe RGB, but much better shaped, so it can hold about 50 percent more useful color range than Adobe RGB. DCam 3 is in between, and somewhat larger than Ekta Space, plus differently shaped.

The DCam spaces have more on the red to greeen (yellows) side, mainly, and their shape and color range was very, very carefully considered only after a great deal of research into which colors in the world around us wind up where after various common RAW processing methods.

And I have new and much improved chroma variant sets for all seven of my spaces. Plus I have chroma variant sets for ProPhoto and Adobe RGB. So whatever suits you, I've pretty much got it now. I will be announcing all this on my web site as soon as I finish the automation necessary to make online selling of the sets feasible. Also, once anyone has purchased one set at the normal price, they will be entitled to purchase any of the other sets for their own use at a $35 discount. In most cases the normal price is $95, but for the Adobe RGB and Pro Photo sets it will be $85.

It is unfortunate that Camera RAW 3 and Canon's DPP v.2 both are closed to profiles for both the source (camera) and output (working space), and only let us use the built-in profiles. Many RAW converters are open, however. At this point, my favorite converter is RAW Developer by Brian Griffith, at http://www.iridientdigital.com, version 1.5.1 or later. It's exceedingly clever and has lots of great features, as well as being completely open.

All of my working spaces share the same superior tone curve except for Ekta Space PS 5, which uses a gamma 2.2 tone curve, like Adobe RGB. If you have files in Ekta Space (as indeed you do), you can open up the deep shadows nicely by <assigning> Chrome Space 100, an updated version of my original Ektachrome Space profile, which is otherwise identical to Ekta Space (just a different tone curve). So in your position it's often ideal to have both the Ekta Space PS 5 and the Chrome Space 100 profile sets, each w. 29 chroma variants.

That way, on an image-by-image basis you can decide to open up the shadows quite a bit, and you can adjust the chroma in either condition (w. either tone curve in place).

And you can decide if you want to jump to one of my better designed digital camera working spaces or stick with one of the two main industry standards, and there is a variant set for any of those. Pro Photo's design is a good one, except for the tone curve, though it will only make a very small difference in the quality of your printed work to use my DCam 4 with its more perceptually linear tone curve (more printable levels when converting into an 8-bit per channel printer of decent linearity in its raw state, i.e. as profiled).


Anyway, once I get my site pages up, I won't have to try in vain to explain all of these details in emails. It's a long story, too long to tell completely this way. But you can get the idea. For most people, I recommend DCam 3 as the best general-purpose digital capture RGB working space. It's big enough to hold, for example, neon reds. And just about 100% of all flower or brilliant artificial fabric colors. A few flower highlights will still be clipped that will clip less with one of the larger two DCam spaces (4 or 5), but those spaces have more room by far than any printer or display for those lighter colors, so it's not necessarily beneficial to have that extra room. It's sort of a clip now or clip later deal, when you are converting from the RAW file.

It's time for bed, past actually, so let me know if I can help with more details tomorrow, Saturday.

DCam 2 is intended to be a super-efficient, conservatively sized space that nevertheless holds just about all common colors. DCam 1 it a super small space, somewhat smaller than sRGB, but intended for use only with pictures that have fairly neutral highlights and no very saturated colors, in order to minimize quantization error with them to a great extent (a subtle, quality-protecting method), and not for general use.


Thanks again,

Joe Holmes

Kensington, California

http://www.josephholmes.com