processing and printing image files

Photoshop CS5 Performance on OS X

Mon, 23 Aug 2010 11:56:57 -0600

I just took delivery of a new 12 core Mac Pro. Something about “cores” is techie right now, so I suckered for the whole enchilada ... 2.93ghz dual hex-core. 12 cores running at that speed .. now that’s gotta be fast.

I spent most of the weekend setting it up. I found using migration assistant would have taken about 3 days, so I opted for a rather convoluted process. First I cloned the old mac 2x1tb raid 0 to a single 2TB drive via eSata ( which only took about 9 hours). I then used it as my startup drive for my new MacPro so I could keep working, and used the old tower to setup the drives for the new MacPro. In the old MacPro I installed 4 2Tb Seagate XT Barracuda drives, each drive was partitioned with 4 segments of 50, 700, 1000 and 250 gigabytes in that order. (This was a tip from Jack Fletcher over at GetDPI.com). This meant the 50gb partition was on the outer rim of the platters (the fastest part of the drive), each partition relatively slower. I then set up the first 3 partitions of all 4 into raid 0 drives, leaving me with a 200gb raid of the fastest part of each drive for the Photoshop scratch disk, a 2.8tb and 4tb raid 0 drives for system and data, and 4 individual 250gb backup partitions. I then cloned over my system from the single 2TB drive, and even using FW800 it only took about 4 hours. Finally after all that was done, I moved the 4 drives over into my new MacPro and fired it up.

I launched PS and ran some tests and everything was crawling. I ran the same tests on the old MacPro (about a 3 year old machine) and it did everything in 1/2 the time ... yes my new MacPro was twice as slow as my old one. In activity monitor it was obvious the CPU’s weren’t even being used as it showed the CPU as about 95% idle. It turns out all system 10.6.4’s aren’t equal, and so I took the time to re-install OS X from the DVD’s (no archive and install option anymore which I would have preferred). Obviously something in the newer 10.6.4 enabled the new processors correctly. I should have known better.

So the new 12core MacPro was finally faster than the previous one, but only by about 20%. That’s not much of a gain for such a large investment and 3 year newer technology. I started messing around with the settings in the performance preference pane, and some new options there had me puzzled.

“tall and thin”? “Big and Fat”? I’ll be honest, I’ve never heard of a Photoshop document described this way. Maybe this is standard jargon for Photoshop guru’s but it was meaningless to me without the little tooltip when hovering over the button. I decided to ignore those and just try a different setting. I upped the cache to 6, and changed the cache tile size to 1024k. After restarting Photoshop, I re-ran the tests which resulted in a dramatic speed improvement. A test that took about 55 seconds on my old MacPro, and about 43 seconds on my new one now only took about 22 seconds on the new one. I went back to the old MacPro to compare, but it didn’t seem to affect it that much ... actually I found had installed a plugin which did the same things as choosing the larger tile size.

Turns out this setting replaces an old tip about using a photoshop plugin called biggertiles. In CS4 you had to find this plugin on Adobe’s site, and after installing it could see a nice performance gain in Photoshop when working with large image files. I moved it over when installing CS5, so my old Mac was already taking advantage of this without changing the preference pane, which explains why changing the preference didn’t provide different results. With CS5, you have control over this in the Performance preference pane. Interestingly enough when I click Big and Flat on my new MacPro (24gb or ram) it changes the Cache Tile Size to 1024 (the number I tested), but when I select it on my MacBook Pro (8gb of ram), it stays at 128k.

As an aside, for Photoshop and probably Lightroom, I’m now suspecting the 3.33ghz 6 core version may actually be faster. I’m not sure Photoshop and other programs really leverage all of those cores, so a faster processor speed with 6 cores might be a better choice. (there are 12 actual cores and 12 “virtual” cores, meaning when you view you CPU graph in Activity monitor, you see 24 cores of information ). However, this assumes the only thing you do is run one of those. As soon as you start doing multiple things at once, the cores help a lot.

I did some tests using an Apple Developer preference panel that let me disable cores and virtual cores just to compare. Running a Photoshop test took about 23 seconds as compared to 55 seconds on my old dual QuadCore Mac Pro. However, turning off 8 of the cores only took a little longer ... 27 seconds. It was obvious by monitoring the cores I was only get about 30% utilization. Then I fired up Lightroom, started a task building 1:1 previews from PhaseOne files and had Handbrake ripping a DVD at the same time. The Photoshop task took longer but with all cores running the impact wasn’t nearly significant ... 34 seconds vs 55 seconds. All three programs churned along just fine, as though nothing much had happened. I do this quite a bit (trying to digitize my DVD library so I can stream more of my movies to AppleTV’s I have throughout my house).

And of course there is the distinct possibility that Adobe is tweaking code as we speak to improve performance by utilizing more cores.

The xrite colormunki ... how can this possibly be any good?

Wed, 27 Jan 2010 14:56:17 -0700

When I first heard about the xRite ColorMunki, I pretty much dismissed it as little use to those interested in serious color management. I’ve seen it mentioned many times on forums, and it seems to get pretty good reviews. I was skeptical ... how can any device create a decent profile with only 100 or so colors being sampled? It couldn’t imagine how it would create a decent profile.

Recently I’ve been hanging out at Pixel’s Foto and Frame quite a bit ... being retired is great as long as you have some fun things to do, and messing around with gear is always fun. The other day I spotted a ColorMunki on the shelf and felt maybe it was time to review it. It’s nothing new, and there are plenty of reviews out there, so really I just wanted an excuse to check it out for myself.

The ColorMunki consists of a software package and a hardware device. The entire concept of the device is to take the complexity out calibrating and profiling your display, and profiling your output devices. The software is simple and almost foolproof ... do this then click next.

Profiling a display is pretty straight forward. It gives you a choice of a target luminance (brightness) or will read the ambient light and then provide a target. Personally I believe the logic behind the ambient light step is misguided and I see little point in using it. It recommended a setting of only 80 cd/m2, which was way too dim, resulting in light prints. The prints looked fine next to my display ... which is pretty dimly lit. It’s almost like the ambient light reading is trying to determine how bright your print viewing conditions are, not your ambient light settings ... they should not be the same. I still recommend screen brightness be adjusted by opening a blank white document in photoshop and try to adjust the white to match a blank piece of paper in your viewing station. If the two aren’t close enough to be viewed together, then100-120 cd/m2 is a much better starting target.

Another slight issue I have with monitor profiles is you can’t set a custom white point with the ColorMunki. I find most of the time a default 6500k white point is too cool, and most of my profiles are built around a white point of 6100k. That being said I was surprised when profiling my Apple 24” LED display with the ColorMunki the end result was nearly identical to the profile built with the i1 Pro and i1 Match software. The profile built by the ColorMunki was perfectly acceptable, and certainly far better than not have a calibrated and profiled display.

Now it was time to build a printer profile. I used the software and device to build a new profile for my Epson 7900 printer and Epson Ultra Premium Luster paper. My first try resulted in a terrible profile ... obviously problematic. I concluded this was due to the current Leopard/Snow Leopard issue when printing targets. The second time, I changed the setting in the ColorMatch tab to use Epson Color Controls rather than ColorSync, but left color management disabled in the printer settings dialog, which resulted in an accurate target. Here are some screen shots to show the settings... (windows 7 may also have some issues, be sure you have the latest software from xRite, but the targets from windows should be correct).

The profile creation process surprised me, and after creating a profile I understand a little more about the logic and ability to use such a small number of patches to create a profile. The process to me is quite ingenious ... you print out a specific target of colors, which you read in. The software then analyzes the results, and produces a second target based on the data from the first target to further refine the profile. The end result was a profile that was slightly smaller in gamut size than Epson’s profile for the same paper, but not by much. It appears to have some limits in refining the highest saturation points (edges), but the overall profile seems to be quite accurate.

Comparing Epson’s profile (wireframe) with ColorMunki profile

A simple comparison like this in ColorThink doesn’t really indicate how good the profile is. Additionally in ColorThink I can compare a curve graph of the new profile with Epson’s profile, and visually you can see they are very close.

In the bottom right corner is a small bullseye target. As you drag through the graph a blue linemoves and a corresponding red dot in the bullseye will show a Chroma value for that area of the profile. The closer this tracks to center (0) the better. While the colormunki profile isn’t as good as the Epson one, it still is very good. Here is a little movie, you can drag the scrubber and watch the value and the dot move to compare the two (the ColorMunki profile is the one on the left, Epson’s is on the right). I’ve always felt anything that stays within the smallest circle or exceeds it slightly is a pretty good indicator of a decent profile.

Once I’m pretty confident there are no extreme anomalies in a profile, the next test is a standard print. I printed out Bill Atkinson’s test page using both the factory profile and my new ColorMunki profile and the results were extremely close. I did find an additional problem in that by default the ColorMunki generates a version 4 ICC profile which has the odd issue in Snow Leopard of printing a gray color wherever there is no image data. (so the borders past the image are grey). This is easily resolved by a setting in the preferences to generate a version 2 profile instead.

An additional feature is the ability to refine a profile. This is especially helpful if you have images that tend to have a lot of similar tones, such as flesh tones. Here you actually load an image file into the software and it will produce a target based on the colors of the image. You read this target in, and a new profile is created. I did this step using Bill Atkinson’s test page, and the resulting profile in ColorThink does show some increased gamut in a few key colors.

The device itself is easy to use for most things. I found the revolving switch a little awkward to use and it seemed far too easy to accidentally click the button (although normally that didn’t cause any problem or errors).

For those wanting better color management tools this seems to be a very capable device for a very reasonable price. If you have no need of making printer profiles then perhaps a display only device would be less expensive, but for a little more you have the ability to make a profile if you want ... not a bad thing.

TROUBLESHOOTING EPSON PRINTERS AND SNOW LEOPARD

Tue, 8 Dec 2009 15:46:01 -0700

Since the release of Snow Leopard I have seen many threads from photographers having problems printing. Epson printers appear the most affected, but this might be due to the fact there are just more Epson pro grade printers.

At this point, Snow Leopard is getting blamed for every Epson printing problem, although incorrect installations or improper workflows are a continual issue, so some of these probably aren’t related. Some users are moving from Tiger to Snow Leopard, so their problems might be related to the changes made in Leopard, which most have resolved.

What strikes me as odd is there are so many users not having problems. I have had no problems printing to a 3800, 7900, and 11880 with Snow Leopard. I just inserted the Snow Leopard installer disk, did a standard install, and everything continued to work fine. Frustrated users are trying all kinds of things, but there doesn’t seem to be much commonality - why does it not work for everyone?

The one thing I’ve picked up on from various threads might be related to how “clean” the  Snow Leopard system is. It appears users that have upgraded without fresh installs from previous versions of OS X are having more problems, especially if they also have printer drivers related to older Epson printers.

This actually makes some sense, from the perspective of ColorSync. If you open the ColorSync Utility on a Mac, you normally find tons of extra “stuff” laying around. It seems the OS and ColorSync never forgets anything you’ve done, so stuff from printers past still lurk in your system. As an example, here are some screen shots from my MacPro tower in the ColorSync utility.

I have only 2 printers connected to this computer, a 3800 and an 11880. As you can see, pretty much every Epson Printer I’ve ever owned shows up in this box ... even my 9600 and 2200 which have been gone for many years. This leaves the question of how to clear this up.

At this point, I strongly recommend to everyone upgrading to Snow Leopard to “clean up” their Mac printing system. To do this I recommend the following steps.

1. Uninstall one of your Epson drivers. To do this, you need to launch one of the installer programs for an Epson printer. You can download the most current one, or use one you have. Once you run the installer, click the Pop up menu and select the uninstall option.

You only need to do this with one printer. This will delete some random Epson information that would get missed in the next step.

1. Open the folder located at “YourHardDriveName/Library/Printers/”. In that folder should be a folder called Epson. Move this entire folder to the Trash.
2. Open the folder “YourhardDriveName/Library/Caches/”. If there is a folder there labeled Epson, move it to the trash.
3. Restart your Macintosh. After it finishes booting up, empty the trash.
4. I recommend you launch Disk Utility at this point, and do a repair permissions on your hard drive. This step is optional.
5. Reset the Mac printing system. This is perhaps the most important step. To do this, open the Printer and Fax Preference Pane. Right click (control click) in the printer list, and select Reset printing system.

This will remove all printers from your system and remove all registered printer devices from ColorSync. One downside, all custom saved setups as well as saved paper sizes in your printer dialogs will be deleted.

Your printing system is now basically at the same level as a clean install of the OS. If you launch the ColorSync Utility you should see no printer devices in the list ... this is a good step to verify the preceding steps have worked.

At this point you should be ready to reinstall the driver, and then setup your printer in the Printer and Fax preference pane. I recommend you download the most recent driver from Epson’s site.

This may or may not help you with your printing challenges. I know it has worked for some. Drop me an email if it does or doesn’t work for you, including your printer model.

If you are trying to print targets to make profiles from your Epson printer, there is still an issue, which fortunately has now been recognized by Apple, Epson, and Adobe. The issue and a current workaround is in this article by Mark Dubovoy ....Solving Recent Profiling Issues with Apple Computers and Epson Printers. This issue, while recent to Mark, is actually something we have been fighting for quite some time, but this work around from Eric Chan (Adobe Engineer) is probably the most reliable of several out there.

New firmware for epson 79/9900 printers

Fri, 16 Oct 2009 13:25:35 -0600

Last week Epson released a new firmware update in the UK and Australia for the 79/9900 printers. Those in other markets could either download from the UK or Australian Epson sites, or use the Epson Printer Utility which would also find the update and install it. After installing, the printer should show firmware version HN01699.

Initially there was no documentation, and little was said as to what was changed. Epson Australia has since released a document that describes the changes in the firmware. There are 3 key changes in the Auto Ink Detection and Auto Nozzle Check functionalities.

First, the AID (auto ink detection) system has been improved to provide more accuracy. On some printers nozzle cleanings are frequently occurring despite printed nozzle patterns which appear perfect. This should help with that as well as make other printer functionality more reliable - assuming your AID system is not actually faulty.

The second key change is the functionality that occurs if you choose to set ANC (Auto Nozzle Check) to off in the setup menu. Prior to this firmware update, disabling ANC simply prevented nozzle checks before printing a print but had no affect on any other times the printer performed auto nozzle checks, such as after finishing a nozzle cleaning cycle. A faulty or over sensitive AID system would then continue to clean, which could result in large amounts of wasted ink. Now if you disable Auto Nozzle Checks in the setup menu, it actually disable the auto nozzle checking for almost all functions, meaning it is up to the user to monitor the nozzles with printed nozzle checks and manual cleans. The only time it will do a nozzle check if ANC is set to off is after the initial printer fill and after a black ink swap. (I’m not sure why they left it in there for black ink swaps ... should have disabled that one too). Most importantly this means the printer will no longer do a nozzle check after performing a clean if this option is set to off.

The third change is a new menu, Auto Cleaning Times, which has been added to the maintenance menu (which you access by holding the pause button down when you start the printer).

This menu allows you to change the printers action when Auto Nozzle Checks are enabled. (note that if you choose to disable ANC as mentioned in the previous paragraph, this setting is irrelevant since no auto nozzle check’s will be performed before printing.)

Prior versions of the firmware would perform a nozzle check after each clean. If there were failed nozzles, the printer would attempt to clean again, repeating the process up to 3 times. If after 3 cleans the AID system still detected faulty nozzles, the printer would provide an error message indicating cleaning had failed and ask if the cleaning process should be repeated. This new menu gives you two other options. It is important to note that option 3 now is the equivalent of the prior versions action and is NOT the default. The other two options now available to you are ....

1. When Auto Cleaning Times is set to 1, if the AID system discovers a problem when performing a nozzle check before printing, it will provide a message on the LCD indicating there were clogged nozzles, but then continue to print - THIS IS NOW THE DEFAULT
2. If set to 2 and the AID system detects a clog when performing an ANC, the printer will execute a “cleaning cycle appropriate for the level of blockages detected”*. It will then perform another nozzle check and if there are still clogs, it will provide an informational message on the LCD, and then continue to print.

*(I find the language intriguing ... and hopefully meaningful. Does this mean if the printer finds 2 colors clogged it will execute two separate channel cleans, and not an entire head clean? I have no way to verify this, but to me this would represent a “cleaning cycle appropriate for the level of blockages detected”)

Personally I have a problem with the decision to let the printer go ahead and print once clogs are detected. Not much point in that, with this printer output expectations are so high I can’t image allowing the printer to complete a print if there are blocked nozzles .. at least without knowing how many and which colors are clogged. This isn’t logical ... it should at least ask you if you want to continue or abort printing. So something that could be terrific is probably pretty much useless.

My biggest concern is it appears all of Epson's attention is on the AID/ANC systems, and not trying to understand why this printer (which was supposed to be the best in regards to clogging) appears to be a step backwards from the 78/9880 series. I’m OK with my second printer, but only because of the steps I’ve taken to prevent massive and unnecessary cleaning. I’m not OK with how often I actually have to spend time cleaning it, even though it doesn’t take that much ink anymore. The bottom line is the printer almost always needs something cleaned when powering it on and this is true for a great many users. These “clogs” are not typical Epson clogs ... for example several times now my printer has had 100% of the cyan channel clogged when I start it up, with all other colors being perfect. A simple CL1 of that channel is all that is required to clear up this problem. How does every single nozzle get clogged in a couple of days when powered off, yet not one single nozzle of any other color get clogged? How can you clear all of them with such a low volume cleaning cycle? Then today, the printer showed every color clean except about 50% of the cyan channel was clogged. I wasn’t in service mode and when ahead with a channel clean on the cyan/magenta channel. Afterwards, the cyan channel was fine, but 100% of the light cyan channel was now “clogged”. Huh?

Something else is going on ... air bubbles, head not sealing tight enough, perhaps the new pump assembly has a problem (I’ve heard they went from multiple pumps in the 11880 to a single pump design in the 79/9900’s)... I don’t know, but the real issue here is excessive and odd clogging problems. Yes these changes may make it better if you want to manually manage the situation, but they wouldn’t be necessary if the printer clogged as infrequently as my 3800 or 11880.

(I finally decided to call Epson, and the tech agreed completely that losing an entire channel like this isn’t normal, and told me it sounded like a pressurization issue. They are going to send a tech to take a look at it, and she was concerned when I told her many users seem to be experiencing similar “unusual” clogs. She said she was going to escalate this to the support management team, and they would review it and may want to call me back. Here’s hoping someone will listen...)

So does the firmware update help? Does it change anything? The new Auto Cleaning Times menu appears to be pretty useless ... if I have a clog I at least need the opportunity to print a nozzle check to see how serious the clog is. I certainly can’t let the printer go merrily forward only to find out the entire cyan head is clogged so I just wasted ink, paper and time when the printer knew there was a problem. Give me an opportunity to resolve it or at least understand how serious it is.

This means the only good option is to continue manually controlling the process. If disabling ANC actually disables it after cleaning cycles, then setting ANC to off is an acceptable option and a nice change. This means what you get when you choose this is what you expect to get, and you can print nozzle checks and control which heads get cleaned just like you can when using service mode. Using the service menu cleans still offers the greatest control since it gives you access to a CL1 clean which appears to use less ink that normal cleans in normal mode. However, the normal cleans may only use slightly more ink, and since you can do a single channel clean without fear of continued cleaning from auto nozzle checks you have complete control of when and what gets cleaned ... something that before was only available using service mode. So I’m going to try using standard mode and disabling ANC for a while.

As far as black ink swaps I still recommend doing those in service mode and then manually checking the nozzles. The printer will still perform a nozzle check if done in standard mode, and nearly every user reports the printer cleans almost every time when it switches. The only way to make sure the printer is only cleaning when it needs to is to use service mode to switch, and then printing a nozzle pattern. I’ve only done a black ink swap twice, and I found a normal clean didn’t remedy the problem either time but in fact it made it worse - a CL1 cleared it up both times.

As a final note regarding the challenges with cleans, there seems to be some anecdotal evidence that aggressive cleans actually aggravate the “clogging” problems causing increase in failed nozzles quite frequently, requiring repeated cleanings. Hey Epson .... what’s up with that?

The Turning gate Lightroom Addons

Tue, 28 Jul 2009 12:28:03 -0600

I’ve never worried too much about an internet site to sell my work. Perhaps it’s just an excuse to be lazy, but I’ve always felt images like mine have to be seen in person to be appreciated ... they look OK on the web, but not much different than the millions of others all over the net. I don’t even actively pursue sales or promote myself ... I enjoy the freedom of retirement and I really don’t want to be committed to any particular activity other than what I feel like doing at the time. I guess this website is a good example ... sometimes I feel like writing about something so I do.

Be that as it may, recently I was asked to do a small show by a friend. I sent out a few invites, and was surprised by a request as to where my work could be purchased online. I decided it was time to create some type of site, not really to solicit business but more to allow those who may have seen my work an avenue to make, or at least consider, a purchase at a later date.

The first step in this process was to investigate other landscape photographers sites and observe how they work. I spend time every week visiting photographers web sites to admire and learn from their work, but I’ve never really payed much attention to how their sites are organized. My goal was to have a little site up and running by the time this show opened, so I also checked out all of the hosting type sites such as smugmug and zenfolio. Because I was short on time and only needed 9 images in the initial site I decided using iWeb was a quick and easy solution. I really don’t need a “shopping cart” per say, since most of the time the buyer is someone I know and we’ll just talk on the phone.

In about an hour with iWeb I hacked together a little site basically creating a page for each image, and linking them with some menus and previous/next buttons. The end result wasn’t too bad, but to update and maintain a gallery like this is too time consuming and too prone to errors. iWeb galleries are nice but I wanted something even easier ... something I could do right from Lightroom would be perfect. A google search led me to The Turning Gate and after trying out the demo I went ahead and purchased the TTG Highslide Gallery Pro. To use this for a commercial site I also had to purchase a license for Highslide, the engine which allows the gallery to display the images the way it does. It is very configurable, and even includes the ability to utilize a PayPal shopping cart. Total cost was $69, but in all honesty I really didn’t need the features in the Pro version (yet anyway), so I could have just bought the Standard TTG Highslide gallery for $20 plus the Highslide license for another $29.

Once you load the module into Lightroom, you can build a gallery quickly. The end result is very customizable, including colors, what is (or is not) included, thumbnail size (most of the web modules I tried created thumbnails that were too small) as well as image size. In addition this isn’t a flash gallery but instead uses javascript ... it works fine on my iPhone. (Ok, not a big deal. But it’s kinda cool to show this to a friend on an iPhone.)

To set up a gallery is pretty easy. First I created a collection for each of the various galleries, and I added a Title and Caption to each images metadata, which is used for the images when the gallery is created. I modified some colors and border sizes, created a few menu items and added the links so the pages would work inside the main site, and finally set up the upload options. Once it was configured I saved a template so it was easy to create new galleries. If you want to see what I ended up with, you can check it out here ... Gallery 1.

I’ve also been playing with another option, TTG Client Response Gallery. This web gallery enables viewers to make comments about individual images as well as the overall gallery. I’m trying this as a way for my kids to order portraits when I do sittings of their families. No shopping cart needed, I just need a nice easy way for them to let me know what they want.

All in all a very simple clean and yet effective way to create web galleries.

Some tips on using an Epson 7900

Wed, 1 Jul 2009 00:35:54 -0600

UPDATED 7/25/2009 in yellow below.

The Epson 7900 is a fabulous printer from both a mechanical/ease of use perspective as well as a print quality perspective. The new roll feed system is outstanding, the rugged engineering seems stellar, and menus and interface both on the printer as well as from the driver of OS X are simple and easy. The print quality is fantastic ... better than I originally thought, especially if you have very high quality files. Unfortunately it isn’t without issues.

BUY AN extra LK INK CARTRIDGE ASAP !

The LK ink in the 7900 will need replaced very quickly. Buy one when you buy the printer, or shortly there after. Once you do the initial ink fill you will have about 60% left in all of the colors ... except for LK. It will only have about 38% left. When you run out of LK ink, you will probably have about 40% left in the other colors, depending on how often you need to do nozzle cleans. If you are going to run the 350ml cartridges in your printer, I would recommend using the 700ml LK cartridge. There are three reasons the LK needs replaced so quickly.

Two are due to LK ink sharing the channel with the PK/MK inks. The printer provides one of these two inks and LK ink to a channel in the head. This consumes extra LK ink in two ways. First is in the initial fill. The printer will load that channel with ink, pushing the MK and LK inks through the lines to the head. After the MK is loaded, it will switch and pull the PK ... and you guessed it ... more LK ink through the line. Too bad Epson can’t design the start up process so you don’t insert the LK ink until it has pulled the MK ink (and air in the LK line) into the printer, then when it’s ready to initialize the PK ink you then insert the LK ink. Maybe it wouldn’t work, all I know is when you are ready to start printing you will have dumped about 25ml of LK ink into the waste tank (which by the way will already be at 70%).

This also means whenever you change from PK to MK or back you will also lose some LK ink as well. I don’t use MK ink much at all, so I haven’t made the switch to see how much LK you might lose when you do this.

There is another and very acceptable reason the LK goes down much faster. Checking the job sheets, the printer uses LK ink more than any other color ... substantially more. On many jobs it took as much LK ink as all of the other colors combined - these are full color images, not black and white. I would guess the screening process utilizes LK to lower saturation. Whatever the reason, this is why I recommend using the 700ml size if you are using the 350ml cartridges for normal operation in your printer. I have no problem with LK ink being used faster than other colors ... makes perfect sense. I don’t mind ink that goes onto the paper ... I just hate dumping it into the waste tank.

CAN YOU SAY CLOG!

Speaking of ink going into the waste tank, with this printer there seems to be a lot more of it getting dumped into the waste tank than there should be. Despite the many outstanding features and output quality of this printer it does seem to have an Achilles' heel affecting quite a few users. Clogging nozzles in pigment based inkjet printers is a fact of life. Unlike Canon’s and HP’s where these clogs are effectively hidden from the user, with an Epson you are usually pretty aware of any clogs. Over the past few models Epson has made great improvements in their printers to lessen nozzle clog issues, and the technology offered in the new 79/9900 printers was supposed to make this even better. With Auto Nozzle Checking to detect clogs and clear them automatically, as well as several new head improvements to reduce clogging, the experience should be very satisfying - in fact much like the competitor’s models with this printer clogging should be virtually invisible to the user. Unfortunately for many it isn’t working out that way.

A great many users report excessive ink consumption with these printers. Most believe it relates to the ANC (Auto Nozzle Check) function, and think if they “turn it off” it is disabled and things are good again. It isn’t quite that easy however.

I estimate my first 7900 consumed anywhere from 1,000 to 2,000 ml of ink clearing “clogs”. It was a learning experience, and to Epson’s credit they were very responsive and aggressive trying to address and fix the problems including enough replacement ink to more than make up for the waste. At least 50% of the time when starting the printer I would have massive amounts of missing nozzles. Despite ANC being “disabled”, cleaning the printer would almost always result in multiple cleanings. I was only able to print about 10 to 15 prints before having to replace most of the inks, and not long afterwards I had to replace my maintenance tank.

I know others have reported their 79/9900’s are fine, but after a couple of phone calls I’ve found that some believe just disabling ANC resolves the problem (and in fact that will help sometimes). I ask them how often they have to actually perform a clean when they start up the printer and most say 1 or 2 times a week. Isn’t that “normal” for Epsons?

NO. Absolutely not. I have an 11880 printer that is over 1.5 years old that has only required cleaning a handful of times. I have started up this printer and run a nozzle check HUNDREDS of times without a problem. My current 7900 (the replacement for the original one) requires a cleaning over 50% of the time.

It appears after breaking in two of these printers (Epson replaced my original 7900 because they couldn’t seem to resolve the problems) there seems to be two issues going on with some of them. The relative challenges each user has will depend on whether their printer suffers from either or perhaps both of the problems.

The first problem is Auto Nozzle Check. While great in theory it doesn’t seem to deliver reliable functionality. Almost every user I know has “disabled” the function in the menu (which doesn’t really disable it, more on that coming). Symptoms of ANC not working are two fold. First if you have ANC enabled you will find the printer “cleaning” at odd times before making a print. If you have ANC set to check before each print you are more likely to see this problem. The second symptom is the printer cleaning multiple times when it does need a nozzle clean, often followed by an error message that cleaning failed and asking if you want to clean it again. You should answer no to this and verify the printer does have missing nozzles with a printed nozzle check. Most likely you will find the printer is fine, the nozzle check circuitry just isn’t working right and is erroneously reporting a problem.

The second problem is excessive failed nozzles. The main symptom is when powering up the printer and printing a nozzle pattern the printer will have missing nozzles ... some times massive numbers. They may affect one color or many, but when a color is affected it often it is a large number of nozzles ... 30-70% of them. I’ve never had this massive of nozzle failures in any Epson printer, except perhaps the 9600 that sat in the basement for over 2 years without being turned on. Obviously this means the printer needs to be “cleaned” far more than it should, again dumping ink into the waste tank. On the first 7900 I had, I consumed all of the original waste tank and 50% of a second. A lot of ink.

I hope Epson is aware of the problems and will come up with a way to resolve it. The ANC function should work as advertised. After all it is a highlighted feature of the printer and one of the reasons these Epson printers were the next generation in reducing clogs. In addition, whatever is causing the massive amounts of failed nozzles when printer is powered off or sits idle for a day or so should be resolved. This is the worst Epson I have ever owned as far as clogging issues, and I’ve owned nearly every model they’ve made.

In the meantime I do have some advice to keep things manageable. You can take control of the cleaning process manually and if your ANC is causing problems with unnecessary cleanings you can manage to bypass it completely.

First disable ANC in the menu. All this does is prevent the printer from doing a nozzle check before printing a print (a great feature if it worked). What it doesn’t do is prevent the printer from doing an ANC anytime you actually start a manual cleaning process. So if the printer has a clog and you go into the maintenance menu to initiate a clean, when the cleaning is finished, the printer will do an Auto Nozzle Check. If the ANC circuitry is defective it may indicate a clog when none exists and initiate more cleaning cycles, perhaps to the point where you get the error message that cleaning failed.

To get around this, don’t initiate a clean while in the standard mode. Instead when needing to clean nozzles, startup the printer in service mode, which allows you to initiate several different levels of cleans. In service mode, the printer does not do an auto nozzle check when cleaning, which means you can completely control the cleaning process manually without fear of the printer cleaning itself unnecessarily. As an example, if you have colors in just 2 different channels clogged you can’t use the channel clean function in standard mode to clean them one at a time. If you clean one channel, the ANC will report clogs and the printer will go ahead and clean itself until it the ANC reports no problems. In service mode you can clean individual channels or the entire head, and you can perform 4 different levels of cleans (CL1 to CL4, CL1 being the least aggressive). It seems to me CL1 uses a little less ink than the standard mode cleans, taking about 1-1.5ml of ink per color, and I have never needed more than CL1 to completely resolve all nozzle problems when I have them.

To get into service mode, hold down the center(menu) button, the down (paper advance) button, and the right (menu) button simultaneously while powering the printer on. Once the printer is on, you can verify you are in service mode by hitting the menu button ... you’ll see a much different menu than you are used to.

WARNING: Use service mode at your own risk. Don’t do anything else while you are in service mode. There are calibration and adjustment routines that are performed by technicians, and if you start them you may create problems (which Epson may not fix under warranty). I’m not particularly worried since it is pretty straight forward once there ... just select the cleaning menu and stay away from the other choices.

Finally, some users report that the printer seems to “break in” and settle down after a period of time. This second printer seems to be following that trend ... this week for example I have had only 1 color clogged (about 30% of the yellow), and doing a single channel clean resolved it. I’ve printed quite a bit this week with no other issues, and I have been running with ANC on for every print now for a couple of weeks without fault. So at this point the printer is getting somewhat close to what I would expect.

(UPDATE: I’m happy to report that after about 3 weeks, this 2nd 7900 stabilized quite nicely. It’s now been nearly 1 month since I had a clog. Even a 2 week hiatus without turning the printer on revealed no clogs when I finally started it up again.)

DON’T REPLACE THAT INK CARTRIDGE TOO QUICKLY!

The printer will start giving you a warning about ink running low when there is about 10% left in the cartridge. On the first printer I went ahead and replaced the cartridges when they got down to 2 or 3 %, but I no longer do this. The printer itself will stop when it runs out of an ink and ask you to install a new cartridge. It will do this in the middle of the print and doing so will not cause an artifact in the print. This is a stated feature of this printer, and I’ve done it a few times now without problems.

Why should you do this? Well, I’m currently at 1% of LK ink in this second 7900. I have printed 20 to 30 square feet of media since I hit the 1% mark. Since this is a 110ml cartridge, 1% is only about 1ml of ink, yet since I hit the 1% indicator in the printer I have printed jobs that consumed over 10ml of ink. This means when the printer was reporting only 1% remaining, I probably still had over 10% of the ink remaining in the cartridge. Easy fix ... just wait until the printer stops and waits for you to replace the empty ink cartridge.

Changing the Printers Default ColorSync profile

Tue, 7 Apr 2009 17:06:11 -0600

UPDATED TO SHOW CHANGES IN SNOW LEOPARD ... SEE YELLOW TEXT.

Apple dramatically changed the mechanism used by the OS to communicate with the printer in Leopard (OS X 10.5). Most printer drivers and color managed applications had problems and required updates, but despite the updates residual issues still remain. There is anecdotal evidence that Leopard, CS4, and Epson print drivers are still not playing together as well as they should. I certainly do not know who is to blame ... of course it’s always the “other guys fault”. Generally speaking the changes bring some needed functionality to OS X and ColorSync so it seems the change by Apple is a good one overall, but since these issues are rare and seemingly random I’m not sure there is much effort to resolve the problems.

Sometimes the workaround to resolve printer problems involves setting the printer’s default ColorSync profile to one other than that set by the printer manufacturer. It is important to note this is completely different than simply selecting the printer as the default printer for the OS. How to accomplish this and how to verify whether or not the change has actually worked isn’t very well documented ... I spent some time with Google without much success. Hopefully this article will help those trying to do this same thing.

First a little more discussion of the problem.

One of the key changes for Apple in Leopard was to empower ColorSync. As an example, for most of it’s life iPhoto has been a pretty poor performer when printing to higher quality inkjet printers. The OS didn’t provide the printer driver a simple mechanism to use a profile other than the manufacturer’s “default” profile, which worked just fine if you used the manufacturers recommended paper. With Leopard if you choose ColorSync under the Color Matching tab in print dialog box, you see a popup menu which allows you to tell ColorSync what paper profile you would like it to use. This then functions basically the same as letting Photoshop manage colors when printing, the main difference being the use of Apple’s CMM engine rather than Adobe’s ACE engine.

I don’t know all of the behind the scene changes (and wouldn’t understand them anyway), but one of the problems is older applications and printer drivers didn’t work correctly with this new system of OS color management. Currently I see sporadic complaints on forums about their printers not producing expected results despite successful printing prior to some change. The most common complaints sound much like “double profiling” which means both the application (Photoshop) and the OS (ColorSync) are applying the printer profile. The print driver should be able to tell the OS to disable ColorSync, and it certainly appears it does but somehow the communication isn’t working 100 % of the time and ColorSync is still applying some modifications ... perhaps even on rare occasions applying the full profile.

As I mentioned, most of the time things seem to work. There is one very specific example which can be replicated easily, which to me indicates there is still a problem within the OS, application, and printer driver that may cause unexpected results at random times. This problem was first brought to my attention on a thread on Luminous Landscape, and after quite a bit of testing it is a confirmed bug.

This problem occurs when attempting to print with no color management from Photoshop CS4 using OS X 10.5.6 and Epson’s most recent print drivers (I personally confirmed this with the 7900 and 3800). For example when printing targets for making profiles you must disable both application and OS color management. If the target is printed using a different paper type than the paper type used for the current ColorSync default profile (which is set to Epson Premium Luster) the target will print inaccurately. This is especially true if the target requires a different black ink (MK/PK) than the ink used for the default profile paper type. I haven’t determined if it just the black ink type that matters or if other factors require matching the paper type. If color management is disabled completely, any settings in ColorSync should be irrelevant - this should never happen. But targets printed using various combinations of Photoshop and operating systems (CS2, CS3, OS X 10.4 and Windows) yield identical results .. the only ones that are different (and obviously not as good) are those printed from OS X 10.5 and CS4 where the printers default profile was built using a different paper type than the one being used in the Printer drivers dialog box.

As an example, I am currently profiling Breathing Colors Chromata White canvas for the 7900. While I found their profile for the 11880 performed very well, their profile for the 7900 produced surprisingly different (and bad) results. When I first printed the target, it looked a little “washed out”. I remembered the bug, so I set my default color sync profile for the Epson 7900 to Watercolor Paper Radiant White MK (the paper type used when printing to this media), and printed the targets again. The results were as expected, and different than the original targets. Personally I fail to understand how any information in the default profile would even cause such a problem, not to mention why any information from the default profile is even accessed and used at all. Interestingly enough, CS3 seems to print targets under Leopard accurately.

This seems to be a bug only when trying to disable all color management ... as long as Photoshop is managing colors things seem to print normally (for most people that is).

OK ... enough already. Here’s how to change the default profile ...

The printer manufacturer establishes a “factory default profile” which is used by ColorSync when printing. Until now I guess Apple assumed most consumers used their printer manufacturers recommended paper, which resulted in somewhat satisfactory prints.

As I mentioned this has nothing to do with the default printer. Changing this default profile isn’t particularly straight forward, and what’s worse, it isn’t obvious that the change has even worked. Begin by opening the ColorSync Utility (Applications/Utilities/Colorsync Utility) and click on Devices.

You will have to resize the list by dragging the dot so you can read the entire profile name. You will notice here that my 7900 is not the “default” printer, the HP laserjet in my office is. If you scroll down the list of profiles for the printer you will eventually find one with a blue dot, and if you click that profile you will then see the next screen (you may have to resize the window so you can read all of the information about the profile).

Here we see the default profile for this printer is set to Epson Premium Luster, as noted by the blue dot. This is the first confusing part ... the “Factory Profile” information on the right side has nothing to do with the “Default Factory Profile”. If you click on any profile in the list, it will then show up as the “factory Profile and the “current Profile”. Sort of meaningless.

update: Please note that if you are using Snow Leopard and the latest Epson Drivers, the default profile is no longer Premium Luster. For example for the the Epson 7900 it is now sRGB Profile ... here is a screen shot

If you are using Snow Leopard and your screen does not match the one above, INCLUDING both the Factory Profile and the Current Profile you might try resetting the default profile by clicking the arrow next to Current Profile and selecting set to Factory. If that doesn’t resolve your problems, then you might try setting the default profile to the profile you are going to print from following the rest of the steps outlined below).

Our goal here is to change the current default profile from the factory default to a different one, in this case to a profile that was made using the paper type we intend to use when building our new profile. To do this make sure the profile with the blue dot is highlighted, click the arrow button next to Current Profile, and select Other...

which then opens this dialog box ...

In the case of Epson printers the profiles are buried pretty deep in Epson resources, and are not in the ColorSync folder inside the Library folder. However, the OS seems to know where they are. If you don’t see them, here is the complete path to navigate. (If you still don’t see them you need to reinstall the printer driver).

In this case I’m going to set the default to Water Color Paper- Radiant White_MK ... the setting I’ll be using when printing the target and printing to this medium.

When done you now see this listed as the “Current Profile” ...

At this point, you are done and can quit ColorSync, but if you are like me, you want to confirm what you just did worked, and are puzzled why the blue dot indicating the default profile didn’t move. So you click on the profile you just chose and are surprised to see it listed as the factory profile and current profile. In fact, if you click on any other profile in the list, they all display as the Factory and Current profile.

HUH?

Here’s the problem .. the blue dot does not indicate the current default profile. The blue dot represents the manufacturers default profile. Basically it is the one that will be used if you reset the Printers Default Profile to the Manufacturers default profile. The blue dot doesn’t ever move. In addition, the Factory Profile and Current Profile listed aren’t really connected to the default profile. What in effect this screen does is show you the factory profile for the highlighted choice, and show you the current profile. You can literally take any profile in the list, and using the same steps tell ColorSync to use a different profile (the Current Profile) any time you choose the highlighted profile. I’m not sure what functionality this offers, but for example, I could create my own profile for Epson Premium Luster, and use these steps set the Current profile for Epson Premium Luster to use my profile. If using ColorSync and choosing Epson’s Premium Luster profile, ColorSync would instead use my Premium Luster profile. Yeah, I know, way too confusing ... why would you ever want to do that? I don’t know ... I personally can’t think of a reason. The only real purpose for any of this functionality would be to change the factory default ... why can’t we just click a profile, and click a “Set as Default” button, and then the blue dot moves to show the new Colorsync default profile for that printer? Seems logical, but it doesn’t work that way.

There are two ways to verify what profile is currently set as the default ColorSync profile for your printer. One is to open Colorsync Utility and click the profile with the blue dot and then see what profile is listed as the Current Profile. This is what ColorSync will read as the default profile. I’m not sure this setting is very sticky ... it appears some conditions may cause it to reset. Another way to verify the default profile is to select Print from the File menu while using an Apple application, such as Safari or Preview. Select the printer, select the Color Matching tab, select ColorSync, and make sure Automatic is displayed. The displayed profile is the current Colorsync default for that printer.

As mentioned, once I set this paper type as the default my targets printed accurately. I verified this by reading targets printed from CS3/OS X 10.4 and comparing with the ones from 10.5.6/CS4 ... virtually identical results. It is obvious Apple has work to do - this entire process should be much more transparent and easier to manage - it shouldn’t even be necessary. In my example, if I have told the OS, Photoshop CS4, and the printer driver to not color manage a document, any settings such as this shouldn’t affect the output anyway. The fact it does affect the output indicates there are still residual issues remaining, and hopefully someone at Apple, Adobe, and in my case Epson are listening.

There may be other occasions where changing this default profile is worth trying to see if it cures a problem you are experiencing. When doing this I had to muddle through it ... I couldn’t find it documented very well anywhere. Hopefully this will help.

Color Management Urban Legends

Wed, 25 Mar 2009 15:14:17 -0600

So here is the final chapter in my Color Management series. My approach from the beginning wasn’t really to teach how to do color management, but to help those that don’t understand it (which includes most of us that use it) to trust it enough to actually use it correctly.

As I mentioned in the first article, there are many Color Management Urban Legends floating around. On the surface they all seem logical. As a photographer, you have a choice. You can believe these urban legends, and use them, or you can trust the science of color management even though you don’t fully understand why it works and use correct principles. The choice is certainly yours to make, and unfortunately there isn’t any real way to “prove” to someone through a bunch of web articles that they can trust it. Anyway, here goes ...

Color Management Urban Legend (CMUL from now on) #1: If your output is to a device or facility that requires sRGB or just for the web, you should just use sRGB as your working space in Photoshop.

As with most of CMUL’s, they sound logical and seem to make sense unless you understand the purpose and capability of a complete color managed workflow. When I hear this, my normal response is why? There is nothing simpler about using sRGB, it’s not inherently better or easier . As I’ve discussed, the choice of a working space doesn’t have much to do with your images final output choice, and using a better workspace doesn’t change the workflow or make processing the images more challenging. Instead what you are doing is constantly forcing your image into a “lowest common denominator” state, and instead of using color management correctly, you are just using an output profile as a working space to constantly clip and force your data into sRGB. Using this logic, would you also use your printer’s profile as your working space? That’s basically what you are doing when using sRGB as a working space because your lab requires sRGB files. Instead, the process is just the same as when printing your own prints... wait until the last step, then output files as sRGB images to send on to the printer.

CMUL #2: If your image is an sRGB jpeg, you should use sRGB as your working space.

OK, I’ll admit this is a tough one. Once you’ve baked the information into a jpeg, most of the image data has been discarded. There isn’t much you can do to the image without causing some serious issues. Most likely it really doesn’t matter what space you are using in this case, because you’ve already severely restricted what you can do with image anyway by not having the RAW file. I don’ think using a wider space will hurt, and I certainly wouldn’t change my default workflow to sRGB. As long as you have Photoshop confirming color management mismatches, if you choose to use the imbedded space (sRGB) instead of the default working space (such as ProPhotoRGB) when you open the file, you effectively set the working space for just that image as sRGB.

CMUL #3: The camera can’t capture anything outside of the AdobeRGB space, so there is no reason to use a wider space (such as ProPhotoRGB).

Another “logical” conclusion. My question is how do you know your camera doesn’t capture things outside of the AdobeRGB space? Most try to prove this by showing a camera profile mapped against AdobeRGB that was made by shooting a target in an artificial situation. There is nothing wrong with using camera profiles, but unlike printers where we can send information that we know exceeds the printers capabilities and then map how the printer responds to that data, shooting a chart like this doesn’t do the same thing. Even if it were true (which it isn’t) this assumes that you will never do anything in the RAW converter that might push that data outside of the AdobeRGB space. The simple fact is most of the time it does. I’m not sure there is a technique that can actually measure the maximum limit of a cameras sensor, but I do know you can’t do it by shooting a color target lit by a strobe. More importantly, by the time the RAW data is processed and reaches Photoshop, very few images will fit inside the sRGB space, and most won’t fit inside the AdobeRGB space.

CMUL #4: Your monitor can only display a space similar to sRGB ... you can’t see things in the wider spaces anyway.

This is a common argument lately ... if you can’t see a color on the screen, then how can you know what adjustments to make? It is important to understand you aren’t seeing those colors on the screen anyway - you never view the actual colors of the image when editing it in Photoshop. What you see is the colors of your image mapped through the monitor profile. The color management system is trying to display the image as it was intended to be seen, not as it literally is. It’s goal is to maintain the relationships of density, color and saturation to the other colors of the image within the confines of the current monitor so as to preserve the perception of the image. It is more about how saturated is this yellow than the yellow next to it. Perhaps that yellow isn’t as saturated as the one your printer can make ... no problem. When it maps your image data into the printer space, it can take advantage of the ability of the printer to produce a more saturated yellow, and modify all of the other colors to maintain the same perception of the image. True, side by side you may discover that the yellows of the print are more saturated, but the overall effect is still the same because the relationships have been maintained. (If anything, the print may be just better than the display). One additional thought ... almost all LCD’s have colors that exceed the sRGB space. By using sRGB you aren’t even achieving the maximum quality your monitor is displaying. Here is my Apple 24” LED LCD screen compared to sRGB ...

As you can see, the sRGB space (solid color) fits completely inside my monitors space (wireframe). My monitor, even though it isn’t a wide gamut display, exceeds sRGB in every possible color. Most monitors today exceed sRGB in quite a few places. In reality the world would be much better off if sRGB passed into the history books, because there are not many devices left ... cameras, monitors, and printers ... that do not exceed sRGB in at least some colors.

CMUL #5: ProPhotoRGB is so much bigger than your data it has to stretch the data and can cause banding in some transitional tones.

Another “logical” theory. With this approach we decide to throw away some of our data even though it won’t fit into our container, rather than use a container that appears way too large. If you’ve noticed on my site there are quite a few images of golf holes ... I guess it’s pretty obvious that I’m into golf. I remember one time waiting in line at a driving range where you purchased tokens to insert into a machine. The machine would dispense 3 different quantities of balls, based on which of the 3 tokens you purchased. There were only 2 sizes of baskets however, and I watched as a person in front of me decided to select the smaller basket, but then inserted the token for the middle quantity of balls, incorrectly guessing that it would be large enough. Pretty soon golf balls were spilling out of the top. When it comes to working spaces, theoretically it might be best if we could use a space that barely contained our data. That’s pretty impractical however ... so either you have to lose some information or make sure it all fits in your space.

The idea of the space being “too large” has more to do with the bit depth we are working with. If we are only working with an 8 bit file, we are limited to about 16 million colors. In sRGB or AdobeRGB this won’t be a problem ... this many colors spread throughout the space will still be close enough together that we shouldn’t see any problems such as banding. If we choose to use 8 bit data in ProPhotoRGB, we have to expand all of those colors to fill the new space. Same amount of colors, spread farther apart. Indeed they may be too far apart in some cases, resulting in visual issues.

However, if we move to 16 bit image data, we now have 281 trillion colors to work with to fill that space. With this many colors, despite the bigger space, the colors themselves are actually closer together. This means an image in 16bit ProPhotoRGB can render transitions every bit as well, and perhaps better than the image in 8 bit AdobeRGB. Bottom line, if you choose ProPhotoRGB as your working space, you should also be using 16bit image data. If you are only using 8 bit image data, you should stay in AdobeRGB . Should you use 16 bit data in AdobeRGB? Won’t hurt anything, but not much point in it. Why use 16bit data and then clip it into a small space anyway.

CMUL #6: My printer is only an 8 bit device, so why work with 16 bits?

First, even though the printer, or printer driver can only handle 8 bit data, that doesn’t mean it’s gamut fits inside of AdobeRGB. I’ve already shown that AdobeRGB will result in clipping colors that many of todays printers can actually print. More importantly the main concept is to not restrict any data of your image file until the final step .. output. If you work in a small space using only 8 bits, you are constantly “clipping” the information as you work on it ... compressing information and altering other information. It is better to not have these artificial restrictions limiting your editing. Let the color management system do it’s job, and map the colors into the output space when the time comes.

Enough urban legends ... for now. There are plenty more, and I may add some to the end of this article, and even update it as I think of or see others.

As a closing statement to this color management discussion, I offer one last thought to those that just don’t trust it. Lightroom has proven that a fully managed workflow from 16bit/ProPhotoRGB conversions works. Color Management, once you set it up right can be just that transparent ... it just works. I have yet to see anyone complain that Lightrooms workflow is causing issues with the printed output. Adobe engineers did not place options for sRGB or AdobeRGB workflow into Lightroom ... there is no need at all. The same workflow can be used in Photoshop, and if you are seeking maximum image quality, using 16bit/proPhotoRGB in Photoshop is the way to go.

sRGB, AdobeRGB OR ProPhotoRGB? My Color Management Journey

Tue, 17 Feb 2009 17:24:44 -0700

Perhaps the most misunderstood aspect of color management is the choice of a working space, and Color Management Urban Legends about working spaces are plentiful. I stated in the previous article I feel the ultimate color managed workflow requires setting the working space in Photoshop to ProPhotoRGB and conversion of RAW captures into 16bit ProPhotoRGB files. This is the workflow built into Adobe Lightroom, and works very well.

What is a working space and how do you use it? The only analogy I’ve ever come up with is in comparison to most sports. The working space is sort of like the rule book and field of play for the Color Management System(CMS). It defines the colors for the CMS, which using those rules modifies all of the color data to whatever device profile the data is being viewed in, and sets the boundary or limit to the available color pallet (gamut) for the application. As we work on our file within the editing program any change which creates a color that is “out of bounds” is remapped back in bounds by the CMS. The only way we can be sure we are not limiting ourselves is to make sure our color space is large enough to contain all of the colors we are working with. Currently ProPhotoRGB is the only standard working space which is large enough to do that.

For many years my prints didn’t match my monitor very well, so I was still working the same way as when I used a darkroom ... print a test, correct and print another test, rinse and repeat until it looks good. In the past few years, my personal focus has changed to landscape photography, with my goal always being very large and high quality images. I set about trying to get a better monitor to print match and found there is an incredible amount of information out there, with plenty of people offering all kinds of advice - much of which is contradictory. To make matters worse, nothing ever dies on the internet, so things written several years ago remains, even though technology advances require a different approach now.

Like many photographers, I didn’t pay much attention to working color spaces for a long time. Ignorance is bliss they say, and I really didn’t understand much about color management. I started with sRGB as my working space in Photoshop because it is the default when you install it. At some point I discovered my camera had a setting for AdobeRGB, so I switched my working space to match. I made the assumption that the “working” space was whatever space I captured the image in. I didn’t even realize RAW files do not have a color space (in the context of this discussion anyway), and the AdobeRGB camera setting I enabled only applied to any jpegs the camera made.

During this search I found several very helpful articles which I have linked at the bottom of the page. As far as working space choices, an article by Adobe Camera Raw guru Jeff Schewe demonstrated very effectively why ProPhotoRGB is important, even if the destination is for a web sRGB jpeg. I never miss attending Jeff’s classes at Photoshop World, he has a great knack of demonstrating the practical application of somewhat technical information. You may want to take some time and look it over now, or refer back to it after reading how it affected my thinking.

Schewe Photo -sRGB VS ProPhoto RGB

As I reviewed this information, a few things started to clarify in my understanding of color management.

1. The working space doesn’t have an obvious visual relationship to the various device spaces in our workflow, such as our monitor or our printer. It’s more like a container for our image data.
2. The camera can definitely capture colors which are outside of the sRGB and AdobeRGB spaces, and normally you will expand that range further when rendering the image in the RAW converter.
3. If you choose a working space smaller than the colors of your image when rendering from your RAW convertor, or as a working space in the editing program such as Photoshop, the Color Management System will compress and clip the colors so all colors will fit in the working space. Any color information outside of the working space is gone for good, and some of the data that was inside the space is permanently modified based on the smaller space.
4. Many normal steps in Photoshop will not be able to use any colors outside of the working space,artificially compressing and restricting your colors. Because the monitor is gamut limited, you may be deceived and think it doesn’t matter since it isn’t as apparent on the monitor, but in reality what you are doing is restricting all output to fit inside the limited gamut of your monitor. One of the most common mistakes in trying to set up color management is assuming the monitor is a good guide.
5. Most printers since the introduction of the Epson 2200/7600/9600 can print some colors which are outside of the AdobeRGB color space (and sRGB space). Current printers have a large number of colors outside of AdobeRGB, and it can be assumed this trend will continue as printer, paper, and ink technology progress. If you limit your editing to sRGB or even AdobeRGB, you cannot take full advantage of current and future output technologies. If you don’t limit your editing by using ProPhotoRGB you can easily take advantage of future advances by simply using the new device profile ... the CMS system will map your image colors into the new space and any advances will automatically be realized.
6. The Color Management System, if you set it up right and trust it, does just what it says .. manages the entire process. Your image will be mapped through the appropriate device profile whenever it is viewed, such as your monitor while editing or your printer when printed. The job of the CMS is to maintain the visual appearance of the image regardless of the gamut of the device. You don’t have to be overly concerned with the overall gamut of the space and any mismatches ... the CMS takes care of it for you.

With these thoughts in mind I continued to do quite a bit of research trying to understand the role of the working space. The more I felt like I understood how things worked and the more I began to trust it, the simpler it actually became. I’m not an expert, but my color managed workflow functions very well because I now understand what it is trying to do and how the pieces interact with one another. In my current workflow I still print a “test” print ... an 8x10 on Epson Premium Luster paper using an Epson 3800 printer. My monitor to print match is extremely close, and usually when I make a correction it is because I see a flaw in some of my editing, not a problem with the overall image color or saturation. When I am happy with the test print (very often the first one), I print the finished print on my Epson 11880 or Epson 7900 using Epson Exhibition Fiber paper. These printers have a larger gamut than my 3800, but by using good printer profiles, the match is very close. Side by side you can see some differences ... the expanded color range of the two large printers often reveals a slightly better image. Despite the subtle differences, the prints from all 3 printers are terrific and completely acceptable.

One of my obstacles to understanding and trusting color management was a failure to realize the working space is not an output space and nothing is ever viewed in the working space. It’s the container for the data for our image, and establishes the rules for the application when we open the image for editing. The Color Management system is the referee, and forces a color management aware application (like Photoshop) to play by the rules of the working space. As long as our image is opened in a program that is color managed, the image will be rendered in the appropriate output space when it is viewed.

The key is to actually let the system work as intended and not short cut it. In the case of the article by Jeff Schewe, he is comparing whether we make the edit’s in the confines of a limited working space that happens to be the same as the destination space (web sRGB), or if we perform those edits in a wider space and after we are done we then allow the color management system to map the results of all of our edits into the destination space. If we remember that even devices with much lower capabilities can still appear visually similar to wider gamut devices, we understand how editing on a gamut limited monitor doesn’t necessarily restrict our ability to maximize quality for devices with larger gamuts, such as our printer.

One of the problems with researching this information is while ProPhotoRGB has actually been around for a long time, the need for using it as a working space hasn’t been as important until more recent years. If 100% of the image data could be contained inside the AdobeRGB space, and if non of the output devices could exceed the AdobeRGB space, there wouldn’t be much reason to use a wider space when working on images. It really wasn’t until a few years ago, with the advent of the Epson K3 Ultrachrome inks as well as advances by Canon and HP, that printers began exceeding AdobeRGB enough to have a visual difference in output.

This is a graph of the AdobeRGB space (wireframe) against a profile for the Epson 9800 printer on Epson Exhibition Fiber paper. As you can see, the only place where the Epson exceeds AdobeRGB is the saturated yellows ... the small point in the graph. For the most part AdobeRGB is much larger in all other colors.

Printing an image on a 9800 would result in great prints. If our working space was AdobeRGB all of the colors outside the wireframe would have to be pushed into the wireframe. With the 9800, such a small number of colors are affected very few prints would exhibit any differences (although some certainly might).

Because of this, there is a lot of information out there still recommending AdobeRGB floating around. As I said, nothing ever dies on the internet. In the past few years however this is being challenged for a very good reason ... it no longer applies.

Here are two comparisons of the new Epson 7900 printer on the same paper against AdobeRGB. As you can see, this printer is capable of producing a large number of colors well outside the limits of AdobeRGB, from greens to blues as well as oranges and yellows.

The original capture of this image is a RAW PhaseOne p45+ file, which has been rendered into a 16bit/ProPhotoRGB working file with Adobe Camera Raw. In Photoshop, some additional work was done (not much) until I was satisfied with the resulting image. For display on this web page, it was then converted into the sRGB space.

Does the web version lose anything? To be honest, it isn’t a lot different than the image on my monitor in Photoshop. (Of course, my monitor’s color space is very similar to sRGB so that makes sense.) But I can tell you it isn’t as good as a print ... not even close.

ColorThink is a great tool to compare this type of data. First, here is a graph of the actual colors of this image(solid area) compared to ProPhotoRGB (wireframe).

As you can see, the image colors easily fit inside this working space. No matter what I would like to do with my data, the color management system won’t have to impose any limitations. Some criticize that the space is too large and can cause posterization. This is why it is important to also be working with 16 bit files.

Here is the same image data graphed against the AdobeRGB space. As you can see, about 20% of the colors of the image exceed AdobeRGB. Had I used AdobeRGB as my working space, all of these colors would have been clipped or pushed into the confines of AdobeRGB, and some of the colors inside the space moved closer to the center to make room for the colors being pushed in.

Here is the image compared to the sRGB(wireframe) space. Even more of the colors are outside of the space. The actual photo on this page has been converted to this space, so all of the colors of the image outside the wireframe graph have been pushed in so they fit inside the sRGB space for display on the web. The Color Management System has done a credible job of doing this so the web image looks OK, but in reality there are colors missing from the image that you cannot see in a browser. The important thing is rather than clipping and constraining the colors as I worked on the file, which is what happens if we set our working space to the smaller space, the conversion is done after the file is finished, allowing a more pleasing rendition of the file. Since the image itself is stored as a 16bit/ProPhotoRGB image, all of this information is still available for use on other devices that have better gamuts than sRGB.

So what happens when using a better output device, such as my Epson 7900 printer? This is challenging to show on the web ... it is much easier to see when you can drag the graph around in ColorThink. However, this graph at least gives you some idea of my point. Here we have plotted 3 spaces. Adobe RGB is the blue wireframe, the image data itself is the solid grouping of dots, and the 7900 space is the magenta wireframe.

You can see there is a large number of colors in this image between the blue and the magenta wireframes. These are colors which are in my image, and can print on my printer. Had I used AdobeRGB as my working space, all of this color would have been lost ... pushed inside the AdobeRGB space. The image would have still looked OK, but much like the web jpeg those printable colors wouldn’t have been on the print. True there are some colors needing pushed into the printer space, but even so many less colors are affected, and most of the colors being pushed will still be outside of AdobeRGB when moved into the printer space.

Does it make a difference? In this case it absolutely does. There are some delicate tones in the aspen leaves, pale oranges and reds that are not visible unless the file is printed on a newer printer. I first noticed it when I printed the image on my Canon ipf6100 ... I was surprised how much different it was than the print from my Epson 9800. The difference between a print from the Canon and the new Epson 7900 is not as striking, but even there I can see a little more color in the leaves. Because more of the colors can be printed, fewer colors need to be pushed into the space, allowing more detail in the yellow leaves. Yes ... the printer can render more subtle gradations because it has more colors to work with instead of a big mass which has been clipped into the same or nearly the same color.

I often see discussions claiming you don’t need to use ProPhotoRGB, some even showing various “examples”. It’s a little humorous, because how can you really demonstrate this with a bunch of pictures that are limited by the gamut of a web browser? Normally what you find is they are just bypassing parts of the color managed workflow ...pushing everything to the lowest common denominator such as the monitor profile or sRGB. The final part of this series, Color Management Urban Legends, talks about some of these and why they really don’t add up.

The most common argument against ProPhotoRGB has to do with the space being too big. Each color space is described by the same amount of reference points. If we make the space larger, the distance between those points increase. Andrew Rodney uses the analogy of a balloon covered with dots. As we make the balloon larger, the dots get further apart. If the dots were actually distributed 3 dimensionally inside the balloon, you can visualize how these would also get further apart. The assumption is as these reference points get further apart, the space will introduce banding, and be detrimental.

Most of the articles that really try and legitimize this theory are pretty dated, and they don’t supply any answer for those that want to take advantage of newer printer technology. It does little good to expand printer gamuts if we intend to force all of our image data into a space that clips the extra color anyway. This entire issue is actually “theoretical” ... while it may actually be true, it is so small you can’t actually see it.

As I said, I’m not a color scientist. I’m not sure exactly how the data in the profile is used. However, it seems that for this to happen, the space would actually have to describe every color, not just reference points, and those colors would have to spread out far enough so that a visible difference can be seen. Each pixel of a 16bit file contains 1 of a possible 281 trillion colors ... yes trillion. I don’t think a profile can map each of those individual colors. What it can do is contain reference points, figure out where a particular color falls in relation to various reference points, and use that information when mapping to a new space using that spaces corresponding reference points. I’m sure it isn’t as simple as that, but to cause banding would require the distance between the colors of the file to be far enough apart to be visually observable, not the space between the various reference points. The colors are not forced to reference points, they are all available. Personally, I don’t think this is possible if we have 281 trillion colors to work with. With an 8 bit file where we only have around 16million colors to spread around perhaps. Working with an 8 bit file in ProPhotoRGB may indeed be problematic ... AdobeRGB is probably a better choice here. But anyone seeing banding or other similar issues in subtle tones when using 16bit ProPhotoRGB images and ProPhoto as the working space in Photoshop has a problem somewhere else in the workflow ... the RAW convertor, the original camera data, some poorly done adjustment layers. Using ProPhoto RGB as the working space is not the cause. Sure you may “fix” it by forcing your image data into a smaller working space, but this is a work around, and places limitations on what you can do.

Enough about this from me. Here are some good articles that were very helpful to me in my learning process ... better written than mine. Hopefully my journey of color management understanding has been a little helpful.

The Role of Working Spaces in Adobe Applications by Andrew Rodney

Understanding ProPhotoRGB by Micheal Reichmann

Why Use the ProPhotoRGB color space? by Uwe Steinmueller

The Ultimate Color Managed Workflow

Wed, 4 Feb 2009 14:11:02 -0700

OK, “ultimate” is how I might describe it anyway. Once we have established and acquired the basic tools discussed in the previous article to implement color management, we then need to establish a workflow insuring we are correctly applying appropriate color management steps. To me the “ultimate” method is one that delivers the maximum quality possible to those viewing the image.

The goal of using color management for photographers is to achieve as closely as possible a match between our monitor and our output. There are several possible workflows, and the quality of our output depends on the choices we make. It is important to understand this concept ... choices in our workflow may limit the final quality of our output. These choices can happen at almost any stage in the workflow, from our choice of capture equipment and capture type (RAW or jpeg for example), to the choice of output device. I’m not saying there is only one way ... I just think it is important to understand the ramifications of those choices, so we knowingly choose to accept the trade-off between maximum quality and a different result if we make those choices.

A common mistake is to start backwards and make choices in the workflow based on some limitation later in the workflow. If your destination output is a web/sRGB file, it sounds logical that an in camera sRGB Jpeg is all we need. After all doesn’t that match the destination? The problem with this is we discard most of the data from the capture itself and allow the camera to provide a “baked” jpeg of our data(as it has been described by some). Once something is baked, you can’t use the ingredients to obtain a different result. Unless the camera jpeg is nearly perfect it will be very difficult to achieve the best results in Photoshop (or other editing software for that matter). If we decide to use RAW files for our capture, we no longer have that limitation. It’s sort of like building a good sound system. There are several components, but if any of the components are weaker than the speakers (the final component) the speakers themselves cannot perform at their optimum level. To achieve the highest quality with our final images, having the final output step as the most limiting is ideal. If we artificially limit the image in the workflow before we are ready to convert it into the output space, we will have compromised the final quality. Is it discernible? Good question ... sometimes it might not be, but sometimes isn’t always. Is it good enough? Each photographer must answer that question for themselves, but before deciding it helps to understand the trade-offs.

There are circumstances were trade-offs between ultimate quality and workflow demands are acceptable. There are all kinds of photographers, and doing a little league shoot probably doesn’t require ultimate quality. A little league shoot probably wouldn’t even be opened in an image editor. For this sufficient quality might be obtained using in camera jpegs as long as we light it correctly and choose the right camera settings so the camera firmware can get us close enough. Each photographer must make these decisions for themselves, and indeed some photographers have multiple workflows based on the type of work they are doing at the time. Again, fully understanding the trade-offs allows one to make informed decisions.

So what is the “ultimate” quality workflow?. Our original image needs to be from the highest quality device that we have available which fits the needs of the shooting situation (hard to shoot an eagle with an 8x10 view camera). With film cameras, quality (especially with regards to resolution) was mostly a function of format ... generally a medium format system achieved higher quality than a 35mm system. While format still impacts digital cameras as well, there also exists within each format a wide range of capabilities. Often budget is the determining factor in our capture device. While that may impose some limitation as to the final size of the prints we make, it certainly doesn’t impose many other limitations as far as the impact or quality of the work we do.

No matter the capture device, perhaps more important in the workflow is the choice of capture type. Here we must choose RAW. Choosing a jpeg workflow seriously limits our output possibilities. In fact, of all the choices we make this would be the most limiting. Some complain about storage and file sizes, but with the capacity of todays storage mediums I don’t see that as in issue if building the maximum quality workflow.

We need a good RAW convertor to render our file into a visible image. Personally I use Adobe Camera Raw (either in Photoshop or Lightroom) most of the time. That choice is based on convenience and the amazing interface and tools it provides. Some claim other convertors do a better job, and perhaps they do for them. From a purely technical perspective it may be possible other convertors have some edge and create better results. Here again we find the tradeoff in our workflow. Personally I haven’t found this ... at least with my skill level. I am trying various other convertors to find out if they will yield better results for various problem files I have, but ACR is just easier to use and offers so many great tools that my results are almost always best when I use it.

The next step in our workflow may vary. There are basically two choices. We can use a workflow tool, such as Lightroom or Aperture, or we can use an image editing program such as Photoshop. We may choose to use a combination of the two. The real decision here isn’t which program, but which bit depth and working space we want to work with. (Remember, I am trying to define what I feel is the highest quality workflow). Lightroom makes that decision for us ... and I advocate using the exact same choices when using Photoshop.

So the short answer is duplicate Lightroom’s settings in Photoshop. Lightroom is built on a ProPhotoRGB/16bit workflow. You can’t change it, and because it is transparent it works very well ... even on 8 bit jpeg files. It’s quite simple to duplicate this in Photoshop ... just set your RGB working space as ProPhotoRGB, and tell the RAW convertor to send files to Photoshop as ProPhotoRGB/16bit files. There are a few convertors that don’t allow ProPhotoRGB. If the RAW convertor can only send AdobeRGB or Jpeg, I strongly recommend changing to a different convertor.

Many of the Color Management Urban Legends out there center around working space choices, and in fact this is probably the least understood part of a color managed workflow. This one decision is the subject of my next article, “sRGB, AdobeRGB, or ProPhotoRGB”, so we won’t go into much detail here. I’ll just leave it with one thought ... .any choice other than ProPhotoRGB will artificially force your image into a smaller space, with some affect on your final output. Even if your final output is just to the web with an sRGB jpeg, if you do not use ProPhotoRGB as your working space you may clip your data with visible results. But more on that later.

Of course the last step is output. If we have followed this workflow so far, we have the highest possible image data. Our output device is what it is ... some are more capable than others. Because we have taken all the right steps in our workflow, we are ready to let the color management system map our results into the output space in a way to maximizes the devices capabilities, and our workflow hasn’t imposed any artificial limits on our image - the only limit is the capability of the output device itself. Some devices are more limiting than others, so the soft proof function in Photoshop allows us to visually see how those limitations will affect our output, and if necessary apply some small adjustment layers to optimize our image. The beauty is it doesn’t matter what output device we choose ... because of our workflow we can easily optimize our image to a variety of output devices so all of our edits apply equally as well to a web jpeg or a large print from an Epson 7900 printer.

As I mentioned, this to me is the highest quality workflow available as of right now, and a great majority of photographers seeking ultimate image quality agree with this. There are those which advocate other choices, but understand those choices involve trade-offs. Many times the trade-off doesn’t work very well. I know the downside to choosing sRGB as my working space, and I really can’t think of a good upside. If there is not a beneficial upside to a trade-off, why should I accept the downside?

So there you have it. As I mentioned, in the next article I will discuss the choice of a working space in Adobe Photoshop and why I feel 16bit/ProPhotoRGB is the best option.