Ad
Ad
Ad
Pages: « 1 [2] 3 4 5 »   Bottom of Page
Print
Author Topic: Costco Prints WAY too dark in Shadows  (Read 34202 times)
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #20 on: January 27, 2007, 05:17:52 PM »
ReplyReply

Quote
With a high contrast image, your mapping algorithm described above would give a flat low contrast image with poor midtones. In these cases one can use the highlight-shadow control of Photoshop or use HDR and possibly some judgment. No photographic artist would use a profile from Costco for such purposes.

You're confusing several things here. First, gamut and contrast are hardly the same thing. When viewing an image on a high-contrast monitor display, a print of the image is going to exhibit less contrast than the monitor image, no question. But recommending that print contrast be further reduced by setting the black point to 25,25,25 or some other value to avoid shadow blocking is rarely a good idea. Doing so to trade dynamic range for wider shadow gamut may make sense occasionally when there are a lot of heavily saturated colors in the deep shadows, but offering it as blanket advice to avoid shadow blocking makes about as much sense as recommending nose picking to solve world hunger. This is especially true when printing B&W images, or any image that fits within the printer gamut. In such cases, if you find it necessary to raise the image's black point from 0,0,0 to avoid shadow blocking, you can bank on your printer profile being bad.

Second, HDR addresses a completely different cause of shadow blocking than what was being discussed. If the image file has a lot of clipped highlights or shadows, HDR can help by capturing additional data to keep detail throughout the entire tonal range of the image. But when the image does not have clipped highlights or shadows, or areas of the image that do have 10-15 levels of detail still print as featureless solid colors, HDR will not help. HDR is only useful until no highlights or shadows are clipped in the image data. After that, you can accomplish the exact same result (bringing the brightest white and darkest black closer together tonally) with a simple level adjustment. But in any event, if there are no out-of-gamut colors in the image, and areas of the image containing detail are printed as a solid color, the profile is bad.

Third, I'm not recommending using profile conversion as the preferred method to deal with out-of-gamut colors. Using relative colorimetric with or without BP compensation can clip colors, which can cause detailed ares to be printed as solid color anywhere in the tonal range of the image. Using perceptual can cause significant color casts and hue shifts. The best way is to manually prep the file to bring all colors within the gamut of the output device.

Fourth, your comment about no "photographic artist" using Costco as a print source is a bit snobbish. Not all Costcos maintain their minilabs to a high degree of color consistency, but many do. I used the one in Vacaville, California for several years with excellent results. The staff there actually knew something about color calibration and profiles, and as a result, most of the pro photographers in the area used Costco for their high-volume print work up to 12x18 inches. If a vendor can meet high standards of quality at a low price with a fast turnaround time, there's no reason to use a vendor who charges more and takes longer to deliver the same result. The only reason I quit using Costco was because I bought a 7600 to be able to make my own prints latger than 12x18, an EyeOne spectrophotometer kit to make my own printer profiles, and QImage to easily and conveniently gang numerous small prints on a roll print job. Using the Epson and my own profiles wasn't that much of a quality advantage over the Dry Creek profile + Costco; what I mostly gained was the ability to print up to 24x36, and to do small print jobs without having to copy the image files to a memory card, drive down to Costco to drop off the files, and then drive back there again to pick up the prints.
« Last Edit: January 27, 2007, 05:20:55 PM by Jonathan Wienke » Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #21 on: January 27, 2007, 10:49:16 PM »
ReplyReply

Quote
I've calibrated my monitor with a Gretag-Macbeth Eye-One Pro spectrophotometer, which has given me excellent matching between all of my monitors (2 LCD, 1 CRT, and a laptop), 4 printers, and several outside print vendors. So I'm pretty confident that it's working properly. Looking at your image, I see plenty of detail in the flatter areas of the water, and there is some detail in the front of the waves, although it is kind of dark and noisy and motion blurred. If that matches what you're seeing, then your monitor calibration isn't too far off, and your local Costco is the problem.
[a href=\"index.php?act=findpost&pid=97757\"][{POST_SNAPBACK}][/a]


That is what I am seeing. Plenty of detail--brightnesss--in the forground water and the relfection on the beach area. I put  alot of time reading and calibrating etc., so I was at an end as to what to do about it.
Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #22 on: January 27, 2007, 10:55:50 PM »
ReplyReply

Quote
I'm by no means an expert in this arena, but I haven't seen this touched on, yet, and its been useful for me in printing to various vendors, etc...

When soft proofing w/ black point compensation turned on, it can be tricky to judge what's going on in the shadows - its taken me a while to get used to seeing those things, and judging against what's going to come out in the print. I've generally found (on three different RA-4 based printers in different shops) that the printer has less dynamic range than my picture, and that the shadows block up a lot more in the print than they appear to on the monitor.

In order to "pack" the print into the dynamic range of the printer and retain shadow detail, I have to reset what I tell the printer is "black" - I do this via the Levels command in Photoshop, and I adjust the OUTPUT shadow slider (that's the bar at the very bottom - not the one directly under the histogram) upwards some until the softproof again shows the details that its missing. Many times, this works out to moving the slider 10-15 units.

The prints come back with black blacks, and look like the monitor - but its a bit unnerving to see it like that on the screen...

Just my experience... I'm sure someone will tell me I'm an idiot, now, but... oh well..
[a href=\"index.php?act=findpost&pid=97762\"][{POST_SNAPBACK}][/a]

That sounds like a logical conclusion to the problem. I've tried moving the sliders too, and leaving the "shadows" set to zero in ACR. I never tried the lower slider. I'll read up on its functin and why it's preferable to the upper sliders for prints. The problem is like Jonathan said though: When it looks good on yuor screen, adjusting to try to match what you will get in the printed version is like trying to nail jelly to a tree.
Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #23 on: January 27, 2007, 10:56:58 PM »
ReplyReply

Quote
The dynamic range of the best glossy prints is no better than 100:1, whereas a monitor can display 1000:1. Compression of the dynamic range is needed in order to print the image and usually an S-curve is applied and the deep shadows are clipped.

Here is an image of a Stouffer step wedge which has decrements of 0.1 OD (1/3 f-stop). The image was rendered with Adobe ACR with the shadow set to zero. Look at the image on your monitor and then print it. I can guarantee you that the monitor will show deeper shadows.

[attachment=1662:attachment]

Bill
[a href=\"index.php?act=findpost&pid=97767\"][{POST_SNAPBACK}][/a]


Bill, that is true and I remeber that contrast information when I was doing research on that topic. The problem is, how to adjust for it consistently?
Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #24 on: January 27, 2007, 11:01:09 PM »
ReplyReply

Quote
i hae a softproofing question. which stage in the process are you using it?

is it the last step or can you put it on right away and so all edits will affect the soft proof?
[a href=\"index.php?act=findpost&pid=97779\"][{POST_SNAPBACK}][/a]

Last step, and then adjust to meet your criteria, proof again, repeat as much as necessary and then "convert" and save the file for printing--  but please post this as a different thread. It's called "thread hijacking."  
Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #25 on: January 27, 2007, 11:13:14 PM »
ReplyReply

Quote
You're confusing several things here. First, gamut and contrast are hardly the same thing. When viewing an image on a high-contrast monitor display, a print of the image is going to exhibit less contrast than the monitor image, no question. But recommending that print contrast be furthe CLIP. . ..
[a href=\"index.php?act=findpost&pid=97846\"][{POST_SNAPBACK}][/a]

Sounds like some good information there. My problem is how I can get good output from my Costco?

I'll tell you what Jonathan. I can upload my RAW file of the image we've been looking at,  and you can do the adjustments using my Costco's profile, and then upload the jpg. I'll down load it and print it and then send the actual print to you in mail. Then we can REALLY see if Costco is the problem, or something I'm doing.

Would you be willing to do that?

If so, if a print size of 8x12 would be big enough, you can upload an 8x12 print in jpg--after your processing and conversion--and I'll get it printed and then send you the print in email. This would definitively ascertain where the problem is becsaue then we would be looking at the exact same print. I'll have two printed. One for me and one for you.
Logged

bjanes
Sr. Member
****
Online Online

Posts: 2838



« Reply #26 on: January 28, 2007, 09:45:33 AM »
ReplyReply

Quote
You're confusing several things here. First, gamut and contrast are hardly the same thing.
[{POST_SNAPBACK}][/a]
You are the one who introduced these terms into the discussion with irrelevant references to "out of gamut colors". You are thinking of gamut as a two dimensional CIE diagram, but forget that there is a third dimension involving luminosity as explained [a href=\"http://en.wikipedia.org/wiki/Gamut]here.[/url]

Gernot Hoffman also has some good 3D representations of CIE Lab. As you may note, there is nothing the eye perceives as yellow at low luminosities.

Quote
When viewing an image on a high-contrast monitor display, a print of the image is going to exhibit less contrast than the monitor image, no question. But recommending that print contrast be further reduced by setting the black point to 25,25,25 or some other value to avoid shadow blocking is rarely a good idea.
[a href=\"index.php?act=findpost&pid=97846\"][{POST_SNAPBACK}][/a]

I hardly recommended setting the black point at 25, 25, 25. You want maximum black in the print to correspond to the maximum density that the printer can use, but if you want a decent rendering, you have to use non-linear methods in some difficult situations.

Quote
Third, I'm not recommending using profile conversion as the preferred method to deal with out-of-gamut colors. Using relative colorimetric with or without BP compensation can clip colors, which can cause detailed ares to be printed as solid color anywhere in the tonal range of the image. Using perceptual can cause significant color casts and hue shifts. The best way is to manually prep the file to bring all colors within the gamut of the output device.
[a href=\"index.php?act=findpost&pid=97846\"][{POST_SNAPBACK}][/a]

Again, you are missing the point. We are not talking about out of gamut colors. The eye does not perceive color at the black point and the OP did not complain about color shifts. Relative colorimetric is basically irrelevant to the discussion. Just as you should manually prep the file to handle out of gamut colors, you should also prep the file to handle luminosities that can not be represented in the print, rather than depending on some profile to automate the job for you.

Quote
Fourth, your comment about no "photographic artist" using Costco as a print source is a bit snobbish. Not all Costcos maintain their minilabs to a high degree of color consistency, but many do.
[a href=\"index.php?act=findpost&pid=97846\"][{POST_SNAPBACK}][/a]

Really,? I am the one who said he got good results from Costco, whereas you originally suggested that the OP go to another lab. Then you suggested that his monitor was mis calibrated. IMHO, he needs to do some nonlinear editing to brighten the quarter tones and not change the black point. A good profile will match up the black points. If you do your homework, you can get excellent results from Costco. However, recent photo ink jet printers exceed the gamut of the Noritsu printers with Crystal Archive paper and this is another reason to do your own printing (in addition to size as you mention).

Bill
« Last Edit: January 28, 2007, 10:09:24 AM by bjanes » Logged
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #27 on: January 28, 2007, 03:49:39 PM »
ReplyReply

Quote
You are the one who introduced these terms into the discussion with irrelevant references to "out of gamut colors". You are thinking of gamut as a two dimensional CIE diagram, but forget that there is a third dimension involving luminosity as explained here.

Oh, please. You're carrying coal to Newcastle here. I'm well aware of the three-dimensional character of color spaces. You're the one defending the notion of raising the black point in an image file to something other than 0,0,0 as something that should be done on a consistent basis, rather than as a strategy to deal with saturated colors in the shadows or as a kludge to deal with a poor-quality profile with shadow problems.

The reason I brought up saturated shadow colors is because it one of the few instances where raising the black point of an image above 0,0,0 makes some sense. Raising the black point of an image above 0,0,0 can take advantage if the fact that gamut increases as luminance rises above the printer's black point. If shadow blocking is happening because the shadows of an image contain some very dark saturated greens the printer can't print, (common when there is foliage in deep shadow) raising the black point a little may in certain instances bring those greens into the printer's gamut. You trade a little DR for the wider gamut you get a little higher up the luminance axis.

Another strategy that can be useful is making a luminance mask, using it to select the shadows, and doing some selective shadow desaturation. This is usually a better option, as it allows you to keep 100% of the DR the printer has to offer, while avoiding blocking from out-of-gamut shadows. If done properly, the visual impact of the desaturation is minimal, but setting up the transparency curve for the selection mask can be a bit tricky to get right.

Quote
Again, you are missing the point. We are not talking about out of gamut colors. The eye does not perceive color at the black point and the OP did not complain about color shifts.

I'm not missing any point. The OP is complaining about blocked shadows. One cause of blocked shadows is saturated, out-of-gamut colors in the shadows that turn detailed portions of the file into dark, solid blocks of mostly-black. The posted image has a lot of saturated colors in the shadows, so it's one factor that needs to be considered.

Quote
Really,? I am the one who said he got good results from Costco, whereas you originally suggested that the OP go to another lab. Then you suggested that his monitor was mis calibrated.

I was trying to suggest the most likely cause of the problem, and that changed when I was given additional information. In the maintenance and medical fields, when there are multiple possible causes for a given set of symptoms, you start with the most likely one first. Sometimes you have to work all the way down the list of possibilities before solving the problem, but examining the most common causes first usually sames time and effort in the long run.

Quote
IMHO, he needs to do some nonlinear editing to brighten the quarter tones and not change the black point. A good profile will match up the black points. If you do your homework, you can get excellent results from Costco. However, recent photo ink jet printers exceed the gamut of the Noritsu printers with Crystal Archive paper and this is another reason to do your own printing (in addition to size as you mention).

We're in agreement here. That was part of the reason I got the 7600.

Quote
I'll tell you what Jonathan. I can upload my RAW file of the image we've been looking at, and you can do the adjustments using my Costco's profile, and then upload the jpg. I'll down load it and print it and then send the actual print to you in mail. Then we can REALLY see if Costco is the problem, or something I'm doing.

Would you be willing to do that?

Yes. Post a link to the RAW, and I'll do a workup on it with the posted profile as a guide, and post an 8x12 JPEG with a description of what I did. It may take a week or two, as I just shot a wedding and processing the shots has been occupying my limited spare time.
Logged

bjanes
Sr. Member
****
Online Online

Posts: 2838



« Reply #28 on: January 28, 2007, 09:44:10 PM »
ReplyReply

Quote
The reason I brought up saturated shadow colors is because it one of the few instances where raising the black point of an image above 0,0,0 makes some sense. Raising the black point of an image above 0,0,0 can take advantage if the fact that gamut increases as luminance rises above the printer's black point. If shadow blocking is happening because the shadows of an image contain some very dark saturated greens the printer can't print, (common when there is foliage in deep shadow) raising the black point a little may in certain instances bring those greens into the printer's gamut. You trade a little DR for the wider gamut you get a little higher up the luminance axis.
[{POST_SNAPBACK}][/a]


One thing that has to be considered is whether the problem related to saturation or luminance. Of course, saturation is relatively meaningless in black, where there is no color. A good printer driver would lay down black and not CMY for blue, which is the saturated color in this photo. In order to investigate this problem, I downloaded the original image (whose profile was the Costco luster) and converted it to AdobeRGB in Photoshop and performed some tests in GamutVision and made some comparisons to what might be obtained from the Epson 2200.

Here is the Delta E (color difference that takes luminance into account) going from the image in Adobe RGB to the Costco profile with relative colorimetric. As is evident, the Delta E is large in the shadows, where the OP is having problems.

[attachment=1685:attachment]

And here is a 3D gamut plot showing the differences between input and output. The difference is in luminance, not saturation. The Costco luster paper simply can't produce a high density black and the output goes to a higher luminosity. There is little change in saturation.

[attachment=1687:attachment]

Here are the same outputs for the Epson Premium Luster paper:

[attachment=1688:attachment]

[attachment=1689:attachment]

I'm just starting to use GamutVision and any pointers would be appreciated. However, from this analysis, the Epson should do quite a bit better than the Cosco printer.

Quote
Another strategy that can be useful is making a luminance mask, using it to select the shadows, and doing some selective shadow desaturation. This is usually a better option, as it allows you to keep 100% of the DR the printer has to offer, while avoiding blocking from out-of-gamut shadows. If done properly, the visual impact of the desaturation is minimal, but setting up the transparency curve for the selection mask can be a bit tricky to get right.
[a href=\"index.php?act=findpost&pid=97971\"][{POST_SNAPBACK}][/a]

Now you talking about non-linear editing and I agree with that. The mask could be used to adjust either lumonsity or saturation, but I think the former would be most helpful. The OP might try [a href=\"http://dustylens.com/luminosity_mask.htm]Steve Bingham's [/url] method if he was shooting in raw.

Bill
Logged
bjanes
Sr. Member
****
Online Online

Posts: 2838



« Reply #29 on: February 02, 2007, 05:36:03 PM »
ReplyReply

Quote
One thing that has to be considered is whether the problem related to saturation or luminance.
[{POST_SNAPBACK}][/a]

For further Gamutvison analysis see this post: [a href=\"http://luminous-landscape.com/forum/index.php?showtopic=14526&view=findpost&p=98941]Click here[/url]

Bill
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8914


« Reply #30 on: February 02, 2007, 08:04:46 PM »
ReplyReply

Quote
No one uses absolute colorimetric for rendering of photographs .....[a href=\"index.php?act=findpost&pid=97823\"][{POST_SNAPBACK}][/a]

I do, sometimes. If the image looks right on the monitor without proof colors on, and absolute col does not show serious out-of-gamut areas with proof colors on, I'll sometimes use abs col. The alternative is to increase saturation and contrast in rel col proof colors.
Logged
bjanes
Sr. Member
****
Online Online

Posts: 2838



« Reply #31 on: February 02, 2007, 09:36:53 PM »
ReplyReply

Quote
I do, sometimes. If the image looks right on the monitor without proof colors on, and absolute col does not show serious out-of-gamut areas with proof colors on, I'll sometimes use abs col. The alternative is to increase saturation and contrast in rel col proof colors.
[a href=\"index.php?act=findpost&pid=98956\"][{POST_SNAPBACK}][/a]

Ray,

I stand corrected then. I should have said hardly anyone uses absolute colorimetric.  

Bill
Logged
dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #32 on: February 03, 2007, 04:06:13 AM »
ReplyReply

Sorry I haven't posted. I somehow didn't get any emails for the topic and just checked. All of this technical information is extremely interesting and I'm glad you all understand it in the death you do. I'm sure we will have some resolution for me soon, which will REALLY take the stress off of me. I have shows coming up and I need to get some reliable prints.

Jonathan, I'm not sure what is the most logical way to go about this test we have in mind. I think maybe the best way to go about it would be for you to process the image and convert to Adobe RGB for print. Print it at your Costco. Given that it looks close enough to your monitor, I'll take exact same file and print it on my Costco's printer. If it is greatly darker, then we know the printer is at fault.

The reason I say use Adobe RGB is that then both the files are identical, although the printers are not. I've tested that before running the same print through through the Costco Profile conversion and Adobe RGB, and the prints are both very close in color and all else.

The reason we need this to be  a controlled experiment is that I will then take those prints and mail them with our explanation to the CEO of Costco. I've given them far to much time to fix this problem, so I'll get the CEO involved now.

If you have a better way to go about it, let me know.
« Last Edit: February 03, 2007, 04:06:35 AM by dwdallam » Logged

Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #33 on: February 03, 2007, 06:28:43 AM »
ReplyReply

Quote
Sorry I haven't posted. I somehow didn't get any emails for the topic and just checked. All of this technical information is extremely interesting and I'm glad you all understand it in the death you do. I'm sure we will have some resolution for me soon, which will REALLY take the stress off of me. I have shows coming up and I need to get some reliable prints.

Jonathan, I'm not sure what is the most logical way to go about this test we have in mind. I think maybe the best way to go about it would be for you to process the image and convert to Adobe RGB for print. Print it at your Costco. Given that it looks close enough to your monitor, I'll take exact same file and print it on my Costco's printer. If it is greatly darker, then we know the printer is at fault.

Since I'm in Germany, the local Costco option (for me anyway) is out. Upload your RAW to your web site with the JPEG and the profile, and post the link in this thread so that I and other interested parties can DL it and  process it to fit into the profile. Adobe RGB would be a bad idea, as the Noritsu+Crystal Archive gamut is significantly smaller than Adobe RGB in many places, and there would definitely be gamut issues. By default, the Noritsu ignores the profile tag, and assumes sRGB, so converting to Adobe RGB means you'll get less saturation and some color shifting in the more saturated colors.

A better idea would be to convert the tweaked-for-Costco file to the Costco profile, and try to get them to print that file with NO COLOR ADJUSTMENTS. You'll need to capitalize that in the printing instructions when you submit the file to be printing to make sure it actually happens. That's the correct way to ensure that you get the results you expect from the profile. The printer will not attempt any auto color adjustments or sharpening; it will simply send the RGB values through with no alteration, and the quality of the results will be directly related to how well the printer matches the profile. Mail me a copy of the print, and I'll post a critique of how well it matches my monitor. As a cross-check, I'll print a copy of it with my Epson R1800, and mail it to you so you can compare my print to your monitor. Send me a PM with your mailing address once you have the RAW posted.
Logged

bjanes
Sr. Member
****
Online Online

Posts: 2838



« Reply #34 on: February 03, 2007, 08:58:23 AM »
ReplyReply

Quote
Since I'm in Germany, the local Costco option (for me anyway) is out. Upload your RAW to your web site with the JPEG and the profile, and post the link in this thread so that I and other interested parties can DL it and  process it to fit into the profile. Adobe RGB would be a bad idea, as the Noritsu+Crystal Archive gamut is significantly smaller than Adobe RGB in many places, and there would definitely be gamut issues. By default, the Noritsu ignores the profile tag, and assumes sRGB, so converting to Adobe RGB means you'll get less saturation and some color shifting in the more saturated colors.

[{POST_SNAPBACK}][/a]

I agree with Jonathan that it would be very helpful for forum members to have access to the raw file. You would not have the copyright watermark, but you probably would get a log of constructive feedback.

As to the color space that should be used for rendering of the raw file, you can not go wrong with 16 bit ProPhotoRGB if you use soft proofing and have a working knowledge of color management. Another more selective approach is to use the smallest color space that will contain the range of the colors captured by the camera as shown by the lack of clipping in ACR or whatever converter you are using.

It is true that the gamut of the Noritsu-Crystal archive combination is easily encompassed by sRGB, but you have no control of how out of gamut colors are handled once the camera output is clipped to sRGB. These matters are discussed at length by Bruce Fraser, Thomas Knoll, and other experts in the Lightroom podcast Number 8 (see [a href=\"http://photoshopnews.com/2006/07/07/lightroom-podcast-episode-8-posted/]Jeff Schewe's[/url] post for details).

Bill
Logged
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #35 on: February 03, 2007, 01:19:34 PM »
ReplyReply

Quote
As to the color space that should be used for rendering of the raw file, you can not go wrong with 16 bit ProPhotoRGB if you use soft proofing and have a working knowledge of color management.

I use 16-bit ProPhoto as my main editing space, make a copy and tweak it as necessary for the output device, then convert to the device profile after I've dealt with any out-of-gamut colors. But I always keep a ProPhoto master copy so that I'm not starting with a device-limited file if I want to print to something with a wider gamut later.
Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #36 on: February 05, 2007, 12:02:20 AM »
ReplyReply

Yes, I've tested ARGB side by side with teh profile, but it is very close to the profile colors. Anyway, point taken.


NEVER, NEVER have I forgotten to do that. I usually upload the photos using thier web space, and then check off NO ADJUSTMENTS. I also look on the back of the print for the code. The code, as you probably know, for NO ADJUSTMENT is N-N-N-N-N.

Jonathan. Before we do anything, let's do this. Since our monitors are both hardware calibrated, let me send you a file processed for your printer. You can see then if I am doing anything wrong by simply looking at or printing the file on your Epson. We can see if I'm blocking up the shadows due to my monitor being out of calibration somehow, or some other error I am making. Then we can move on to see if Costco is the culprit.

If that sounds like a logical starting point, then I'll need your printer profile.

Quote
Since I'm in Germany, the local Costco option (for me anyway) is out. Upload your RAW to your web site with the JPEG and the profile, and post the link in this thread so that I and other interested parties can DL it and  process it to fit into the profile. Adobe RGB would be a bad idea, as the Noritsu+Crystal Archive gamut is significantly smaller than Adobe RGB in many places, and there would definitely be gamut issues. By default, the No. . . .SNIP
[a href=\"index.php?act=findpost&pid=98999\"][{POST_SNAPBACK}][/a]
« Last Edit: February 05, 2007, 12:10:02 AM by dwdallam » Logged

bjanes
Sr. Member
****
Online Online

Posts: 2838



« Reply #37 on: February 05, 2007, 08:07:51 AM »
ReplyReply

Quote
Jonathan. Before we do anything, let's do this. Since our monitors are both hardware calibrated, let me send you a file processed for your printer. You can see then if I am doing anything wrong by simply looking at or printing the file on your Epson. We can see if I'm blocking up the shadows due to my monitor being out of calibration somehow, or some other error I am making. Then we can move on to see if Costco is the culprit.

If that sounds like a logical starting point, then I'll need your printer profile.
[a href=\"index.php?act=findpost&pid=99229\"][{POST_SNAPBACK}][/a]

It sounds as though you guys are making progress, but there are a couple of points that I would like to mention and hope will be helpful. Firstly, the default black point setting of most raw converters clips the darkest shadows considerably as shown by this composite graph for the Nikon D200. With the default shadow setting, ACR rolls off the shadows starting at about log Exp of -2. This doesn't make much difference for most prints, which have a DMax of 2 or less. The Kodak Q14 reflective target has a DMax of 1.95 and this is the deepest black that can be produced on most reflection prints, and the default ACR curve for the Q14 is also shown.

[attachment=1759:attachment]

I have little doubt that the deepest shadows in your image can not be printed but can be seen on a good monitor with a higher contrast ratio than 100:1. As shown by my previous GamutVision plot, the Costco profile does remap the darkest tones in the print and soft proofing should be reasonably accurate. However, I think the best print will require special editing.

Since the final image will most require non-linear editing of luminance or saturation for the best printed results, it would not be a good idea to apply the Epson profile to the file you send Jonathan. I would send the raw file or at least a 16 bit tiff converted with a linear tone curve and shadow level of zero.

Since the DMax of Epson photo printers is greater than that of Costco with Crystal Archive luster, you can expect better results from the ink jet printer. Glossy would give the highest DMax, but the Epson Premium Luster also has a good DMax.

Bill
Logged
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #38 on: February 05, 2007, 02:45:58 PM »
ReplyReply

Quote
It sounds as though you guys are making progress, but there are a couple of points that I would like to mention and hope will be helpful. Firstly, the default black point setting of most raw converters clips the darkest shadows considerably as shown by this composite graph for the Nikon D200. With the default shadow setting, ACR rolls off the shadows starting at about log Exp of -2. This doesn't make much difference for most prints, which have a DMax of 2 or less. The Kodak Q14 reflective target has a DMax of 1.95 and this is the deepest black that can be produced on most reflection prints, and the default ACR curve for the Q14 is also shown.

You're making a lot of assumptions that are only valid for absolute colorimetric rendering, which is only rarely used, and never by Costco. With relative colorimetric rendering and blackpoint compensation, the blackest black of the image source space is mapped to the blackest black the printer can print. The whitest white of the source space is mapped to the printer's paper white, and all intermediate values are scaled evenly between the printer's black and white points. No color value will fall out-of-gamut solely due to its luminance value, the hue/saturation must be out-of-gamut in order to cause a problem. If you send a grayscale image to the printer that has no clipped pixels, no pixels will be clipped during the color space conversion. A matte-finish print will exhibit less DR than a glossy print, but that is because the entire DR of the source image is being compressed into a smaller range with the matte print than with the glossy, not because any of the image's DR is being clipped away. So the discussion of step wedges and reflective targets isn't really relevant. Pixels may be clipped to the edges of the printer's gamut if they are too saturated, but any unclipped image can be printed without any pixel being clipped to the printer's white or black point.

Here's an example. The image is one I shot in Yellowstone a few years ago; the water going over the falls is in direct sun, and the rocks on the right are in deep shadow. Let's say the scene has a 10-stop DR altogether.

[attachment=1761:attachment]

If we print this to glossy paper, the remapping of the white and black points means the print will look something like this:

[attachment=1762:attachment]

If we print on matte, the narrower distance between the white and black points means the print will look something like this:

[attachment=1763:attachment]

But in neither case are the highlights or shadows actually clipped, at least not any worse than the original image was. All that has happened is the original image's DR was remapped into a narrower range, more so with the matte print, which looks correspondingly "flatter". Note that the same image file was sent to the printer in both cases; the change in black point was done by converting from the source space to the printer space, not by altering the black point of the image with levels or curves.

My default ACR conversion settings are shadows set to 0, and linear tone curve, or a custom curve to boost global contrast a bit without clipping any pixels. This keeps as much of the subject's DR in the image file as possible.
Logged

bjanes
Sr. Member
****
Online Online

Posts: 2838



« Reply #39 on: February 05, 2007, 07:57:16 PM »
ReplyReply

Quote
You're making a lot of assumptions that are only valid for absolute colorimetric rendering, which is only rarely used, and never by Costco. With relative colorimetric rendering and blackpoint compensation, the blackest black of the image source space is mapped to the blackest black the printer can print. The whitest white of the source space is mapped to the printer's paper white, and all intermediate values are scaled evenly between the printer's black and white points.
[{POST_SNAPBACK}][/a]

You are the one making invalid assumptions. As I said before, no one uses absolute colorimetric for ordinary photography. The Costco profiles are look-up table based and have tables for relative colorimetric and perceptual colorimetric. Since perceptual   colorimetric distributes the tones in a visually uniform manner, BPC shouldn't be necessary with this rendering intent, but can be applied in the case of a bad profile. BPC does not apply to absolute colorimetric.

Your assumptions about linear mapping from black to white with relative colorimetric and BPC are incorrect. BPC is described in some detail by Adobe here: [a href=\"http://www.color.org/AdobeBPC.pdf]http://www.color.org/AdobeBPC.pdf[/url].

Black and white points are mapped more or less as you describe, but the mapping of intermediate values is not linear as you assume and this was clearly shown by my GamutVision plot shown in another related thread and reproduced below. For details of the mapping process, refer to section 7.3 of the reference. In step 3 of the algorithm,   a decode function is used to map from L*a*b*, and different nonlinear equations are used to convert from L*a*b* to XYZ for L ≤ 8.0 and L > 8. XYZ output values are then calculated from the linear equations shown in the reference and the results are then handed off to the profile for further adjustment.

[attachment=1764:attachment]

Here is where the step wedges come in: in a short scale subject a linear mapping can be done as you describe. For a high dynamic range image, highlight and shadow tone compression can be performed according to the lookup table provided with the profile, and the values used are for average subjects. These profiles are not "smart" and do not take subject content into account. To maintain midtone contrast, an S curve may be applied to the shadows and highlights as determined by the lookup tables in the profile. Rendering from scene to output luminance in high contrast situations is complex and basically nonlinear. We are not talking about clipping, but tone mapping and the process may require artistic intervention. You appear to place too much confidence in the profile to make these decisions for you. In black and white work, this process was known as dodging and burning. With digital one can use tone curves or localized adjustments.

Bill
« Last Edit: February 05, 2007, 09:23:20 PM by bjanes » Logged
Pages: « 1 [2] 3 4 5 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad