Untitled Document

Hi all--As a nature photographer I'm used to coaxing color out of subdued scenes, so subjects like pumpkins are a welcome change. However, I've had a really hard time printing a fairly "simple" image of a stack of pumpkins. The prints (& soft proofs) rob the Orange of most of its "Orang-ness" and they wind up looking too blue. This is true on my Epson 4000 with and without a RIP and even when I soft-proof for a LightJet (haven't sent one off to print on the LJ yet). Rendering intents affect the outcome but none really improve the result (my profile is built with PrintFix PRO, the RIP uses ImagePrint's custom profile).

If I push too hard to get the brightest oranges to print I wind up with a loss of definition in those areas. I'm using InkJetArt Micro Ceramic Luster, which usually serves me pretty well as paper.

I've posted the image and some more thoughts at:

http: //nikondigital.org/forum/ubbthreads.php/ubb/showflat/Number/43846/Main/7475/#Post43846

I'm very interested in what techniques other folks might have used to print this type of Orange, or for this specific image. In the meantime I'll be fiddling with various adjustments to see what I can come up with.

Thanks!--David Cardinal

Cardinal Photo / Pro Shooters LLC

___________________________________________________________________________

David,

Even though you're starting in sRGB, you have a large proportion of colors that are OOG for that paper profile / printer. I took your image in sRGB, extracted the unique colors in it (in LAB) into a list using ColorThink, then plotted them and compared them to your profile. You may find a better paper/profile, but it will be very hard to get these OOG colors to look good from your Epson 4000. Looks like a great test image for papers/printers, though! ;-)

http: //www.wildnaturephotos.com/Private/Davids-Pumpkins-are-OOG.jpg

Those interested can get the color list here:

http: //www.wildnaturephotos.com/Private/DavesColorList.txt

--Rich Wagner

___________________________________________________________________________

David,

These are VERY intense colours, so you are confronting an out-of-gamut situation - too much saturation for the Epson 4000 to handle. I owned a 4000 before the 4800 and I've seen this problem often enough. This is what showed when I soft-proofed and activated gamut warning on a downloaded vesion of the JPEG you posted at the link you gave. I have used three different approaches for this situation, depending on what works best.

One deals with colour space - if you are using a wide one, this is one of those situations where a narrow colour space such as sRGB will reduce the saturation problem substantially by clipping OOG colours. However, an Epson 4000 can reproduce some tones outside the sRGB space, and you have no control over what the colour space converison actually does, so this may not be the optimal choice, but it does work to tame the saturation and uncover detail.

A second in Photoshop is to use a Selective Color Adjustment Layer viewed under soft-proof as you work, so you tame the orange back into gamut by adjusting the extent of magenta, yellow and black that gives you workable hues with less saturation.

The third is that if this image started life as a raw file, the best approach may be to work in ACR 4.1. Here you can adopt one or both of two strategies: reduce the contrast and reduce the saturation. For contrast reduction you can reduce Exposure, increase Fill, then go to the Parametric curve and bring back just enough luminosity and contrast to make an acceptable image without re-introducing excessive saturation. As for saturation, you can reduce Vibrance, then if you need more or more selective saturation reduction, in the HSL tab reduce the saturation of the orange group, and perhaps the yellows and magentas as well. Various combinations of all these tools can help tame the image to make it both pleasing and printable.

Finally, you may have read Chapter 15 of PP5E, where on page 361 Dan uses somewhat flattened "a" and "b" curves in Lab space to reduce colour contrast and saturation, but brings the luminance back with a well-targeted steepening of the L curve, then brings the image into CMYK for final tweaks of those curves. This looks to work well. Dan then continues with another approach using a "False Profile" (pages 363 to 368). I have no personal experience using it, but it too looks worthwhile trying.

The overall objective is to resuce detail and improve the quality of the print by reducing saturation in a manner that allows you to control the resulting hue appearance. All that said, the print will simply not look as vibrant as do these particular colours on a display.

Mark Segal

___________________________________________________________________________

Images dominated by large areas of vivid color, like this one, are an important category, so it's worth discussing the principles.

Here, we want the orange to appear as intense as possible, and we know that the printer is not going to cooperate, because it is incapable of producing the color we actually want. There are many ways to get to the desired result, but there should be no dispute as to what that result is: in the most colorful area of the pumpkins we have to have absolutely the most orange color that the printer can produce. In CMYK, that means 0c?m100y0k for these pumpkins, with the magenta unknown because it determines how orange they are. But definitely the maximum yellow, and zero cyan or black. When sending an RGB file to the printer, we must check that it's converting to this formula, with a loupe if needed.

Outside of these brilliant areas, we need a rapid falloff in intensity. Right now the viewer is just confronted with a ton of orange. When everything is orange then nothing is orange. Accelerating the move towards darkness/neutrality emphasizes the orangeness of the brighter areas.

This can be done in several ways but the objective is always greater detail in the red and green channels. Here's how I would attack in RGB, understanding that on all these layers the opacity can be adjusted to taste.

1) Save a copy of the green as an alpha channel. This will be used to add contrast to the red, which often starts out disastrously shapeless in such images.

2) Apply a curve to the alpha channel, moving the highlight setting to the right until the brightest areas of the pumpkins blow out completely. The result should be a sharply detailed channel that is blank in the brightest areas.

3) On a duplicate layer, apply the alpha channel to the red channel, Darken mode, opacity 50% (much more than enough) and adjust layer opacity to taste, thus adding detail to the pumpkins.

4) Add a curves adjustment layer. On the green curve, locate the range of the pumpkins and steepen it drastically with an S-shaped curve.

5) Change mode to Luminosity, and adjust layer opacity to taste.

6) Duplicate the curves adjustment layer, but this time change mode from Luminosity to Color. This will create more variation between yellow and red. Although the two adjustment layers use identical curves, it's likely that you would choose different opacities for the two.

7) Add a Selective Color adjustment layer. Choose Reds or Yellows, and add yellow and magenta. It's almost always better to heavy up the inks in images like this, getting a richer look.

I've shot a layered copy out to David.

Dan Margulis

___________________________________________________________________________

David,

Maybe the only practical solution would be to switch to a printer that has orange in the inkset?

Mike Sellers

___________________________________________________________________________

Mark & Rich -- Thanks for the work & the thoughts. Some good suggestions on what I can fiddle with. In particular on the idea of a selective color layer a channel mixer layer reducing the blue by -red seems to create an easier to print orange, although it doesn't keep all the color contrast (drama). I'm going to experiment with your thoughts some more & will post my results.

Mark--You mentioned the Epson 4000 possibly being a culprit. Does the 4800 do a better job with these colors. Or for anyone, suggestions on what printer/paper combo can print these intense colors would also be welcome.

Thanks again! -- David Cardinal

http://www.cardinalphoto.com

___________________________________________________________________________

The 4800 would be slightly better on yellow-orange-reds but less so on deep reds-magentas-purples than a 4000. A couple of years ago I was able to do a spot color comparison on a 7600 vs. 7800 (same ink sets 7600=4000 / 7800=4800) and the 7800 did a quite a bit better on PANTONE Orange 021 and did comparatively worse on certain blues. The Ultrachrome K3 magenta had nearly the same chroma as the previous Ultrachrome inks but the hue angle of the magenta was more red/yellow than the colder hue of the Ultrachrome magenta.

You'd think I could actually test that out since I have BOTH a 4000 and 4800 not four feet behind me with at least two RIPs at my disposal (GMG and ColorBurst) but I've got no time to play right now even if wanted to. :-(

Regards,

Terry Wyse

_____________________________

WyseConsul

Color Management Consulting

G7 Certified Expert

704.843.0858

http://www.wyseconsul.com

http://www.colormanagementgroup.com

___________________________________________________________________________

David,

There's VERY little difference between the 4000 and the 4800 for colour printing. Neither of them are a "culprit" - these still rank very high in the quality fine art printer line-up. As Rich Wagner ably pointed out - those colours just hit the gamut constraints of today's inksets and media. Perhaps the new 4880 with the vivid magenta ink would help with this problem, but you'd sacrifice media switching flexibility. The great advantage you have with the 4000 over the 4800 is that you can switch papers painlessly - no black inks to swap at high cost. You'll get the widest gamut and least clipping using a paper such as Epson Premium Luster, Innova F Type Gloss, or Pictorio Gloss - something along those lines. Another point I forgot to mention is that you may also wish to experiment with the Rendering Intent. This may be a case where Perceptual does a better job than Relative Colorimetric. Also, if you have Lightroom but no raw file, you can return the rendered TIFF, JPG or PSD to Lightroom where it reverts to linear gamma and take advantage of some of the very useful controls it offers to adjust such images.

Mark Segal

___________________________________________________________________________

On Fri, October 12, 2007 7:42 pm, Mark Segal wrote:

I agree - and I would also recommend converting from your original's color space (Adobe RGB) directly to your printer profile, rather than going to the intermediary of sRGB. If you look at the LAB plot of the colors I extracted from the sRGB JPG, many of the colors are "against the wall" of the sRGB gamut boundary if you were to superimpose the sRGB profile, showing that those colors were clipped (as expected for a relative colorimetric conversion). As clipping often results in a loss of detail, a perceptual intent conversion directly to your printer space may preserve more of the original detail.

--Rich Wagner

___________________________________________________________________________

Dan's explanation involving variation in saturation is spot-on. Curve the image in HSB space, and you'll see an instant improvement in the detail, and the roundness of the pumpkins.

http://mike.russell-home.net/tmp/DavidPumpkin.png

Mike Russell - www.curvemeister.com

___________________________________________________________________________

Dan,

I am not sure I am doing step 3 correctly. I am duplicating the background layer in the layers palette and then, with that layer selected, I am using apply image to apply the curved green alpha channel to the red channel in darken mode per your instruction.

When I do, the color washes out and it is no longer as orange. It is losing saturation and going more yellow. If I put this layer in luminosity mode, then it appears to be OK and it does gain in detail. But you did not include that in your instructions, so I assume I am making an error. Can you please try and clarify this point?

Thanks,

John Arnold

___________________________________________________________________________

Mark Segal writes:

The problem here is surely that the image is massively saturated and bright, and no combination of reflective dyes can possibly match it. The brightest orange in that image is R=252, B=3, G=80. This corresponds to L=59, a=66, b=70. I don't think any pumpkin is this colour, but I confess I haven't seen every pumpkin!

It certainly seems to do so with my custom profiles for the 7600.

I think the real cause of this problem is earlier in the workflow, at the time the picture was taken or perhaps at the time the raw conversion was done. It's all well and good to use a little saturation and brightness boost when printing, but IMO it's been done too early in this case.

If I had a pumpkin to hand I'd measure its real colour with a spectrophotometer...

Andrew Haley.

___________________________________________________________________________

Andrew,

You are right about that. Off-list I asked David if he would supply me with the original image so that I could see the raw material from the start point and analyse what happened to it, before recommending anything. Problems are usually best anticipated and headed-off at the outset. It is a copyright image so I cannot share it around; let me just say his raw file is good, however it is a low-contrast, intense-colour image which tolerates only gentle pumping-up in Camera Raw and Photoshop. I worked on the raw file under soft-proof last night and recommended to David a small number of basic, gentle adjustments to make it - we hope - print successfully, preferably on non-matte media which has better D-max and more vibrant portrayal of intense bright colours. I can't print it on non-matte myself (the Epson 4800 ink swapping issue), but he can without penalty on his Epson 4000, so although the soft-proof looks encouraging on my (colour-managed) display, this is a work in

progress.

Mark Segal

___________________________________________________________________________

I decided to try Dan's technique (exactly as stated) on a problem image I have worked on from time to time in the past. I was more than pleasantly surprised, I was amazed at how much difference it made. I am going to post both versions; I hope the amount of detail revealed is as obvious here as it is on my monitor.

Oh, I should add I did adjust the brightness of the modified image slightly (curves layer) after I completed Dan's instructions. Neither image was resharpened.

If anyone would like to try this on this image, let me know how to make that available--not sure how that works.

Gloria Wilson

___________________________________________________________________________

David, in spite of serious clipping of the colors, especially the red, the magenta, and the yellow, an excess of these same colors makes your pumpkins too much of a flat, orange-red. Reducing the reds and yellows improves not only the color but the contrast as well. Further enhancing the contrast in your full resolution image should give you much more believable pumpkins.

An Adobe RGB print on one of my Epsons (the 7500), using a pure cotton, white matte paper was about as close to the monitor image as one might expect. Both images were too orange-red. On the bright side the detail appears to be very good, though it's hard to tell with a low resolution image. Great composition, but then you already know that.

I hope you will post your final version so we can all see what you have in mind for a final print.

Howard Smith

___________________________________________________________________________

Sorry, David. My original answer would have worked for me, using my computer, paper, and printer, but it didn't really address your problem.

At a point like this you might want to print out similarly colored images exactly as you did the pumpkins. If necessary, you can always download some from other sites. Either your pumpkins print and the test images will have the same problem, or they won't. Either way you have a better guide as to a direction in which to proceed.

I would also suggest the old tried-and-true method of going back to the basics. Print the pumpkins image on paper with known characteristics, without the use of the RIP, with only the most basic of corrections, without any profiles except those that came with your printer's driver, etc. Then add variables, either one at a time or using an experimental design that combines several variables and gives you results that can be analyzed to pinpoint the source of trouble. Apparently your image is not receiving enough magenta and yellow instead of too much, hence the tendency toward blue. No doubt you realize that this could be the paper, clogged ink nozzles, a bug in the software, an isolated problem with your printer or printer driver, a defective ink cartridge, etc. I once experienced a sudden increase in the intensity of the reds produced in one print after another. Finally, when it became increasingly worse, I found that it was caused by a faulty magenta cartridge. The fact that this probably never happened to anyone else made it that much harder to recognize.

If you are interested in burrowing into this problem to find the culprit, I will be happy to send (offline) a simple experimental design, then analyze the results for you. No cost of course, and no obligation. Had it not been for the cooperative spirit exhibited by so many of the Forum members, I would still be poring over my old Photoshop books and trying to make sense of it all.

Hope you will let us know what works. Probably something simple, but that makes it even harder.

Howard Smith

___________________________________________________________________________

Hello Howard,

My guess on this image is that David being a professional photographer, his original is surely photographically very good, he probably "enhanced" the image for a more striking print, blowing out the red channel in the process.

So this image doesn't need more than a desaturation and adding some details to the blown out red channel. That can be done in 4 or 5 steps using only Photoshop's most basic adjustments features, and it should print correctly.

I have posted my version of it in "Photos"

André Dumas

___________________________________________________________________________

Andre--Yes, that certainly summarizes things. I might not have been clear enough in my original post that the problem wasn't getting _any_ print of the orange pumpkin photo, but getting one that showed the autumnal colors which were more or less in my brain as I photographed the pumpkins in a warm sunrise. If I'd been doing a shot for an ad for the pumpkin farm, that would have been a different story of course.

I really like your result. I've gotten about 5 different versions from folks ranging from a more or less "fall back to the original" to several versions of maintaining the relative color contrast but toning things down to make the image print. Yours fits pretty nicely as a combination of some of the strengths of both.

If you're up for posting (on or off list) your specific steps I'd be happy to include them in a round up of the various options that have been contributed. All in all this has been a good exercise for me and I'd love to share some of what I've learned. Thanks again!--David Cardinal