Pages: [1] 2   Go Down

Author Topic: Good vs. not-so-good output profiles  (Read 14237 times)

Mike Arst

  • Full Member
  • ***
  • Offline Offline
  • Posts: 148
Good vs. not-so-good output profiles
« on: February 02, 2008, 08:25:34 pm »

I have seen more vibrant reds and yellows with Ilford Gold Fiber Silk than I've seen in prints of the same image on Harman Gloss. Might the difference be due to innate characteristics of the two papers? Or is it more likely affected by the quality of the printer profiles? Surely varying quality of the profiles must have a big effect. I haven't yet taken one participant's suggestion to download a Harman gloss profile from Booksmart.com. When I do, I'll be somewhat sobered if the resulting prints are noticeably better than what I'm getting from a custom profile I have made with an X-Rite Pulse system.

Which leads me to the question: if all appears to go normally during profiling -- no hardware errors, no apparent software errors -- but the custom profile is not as good as someone else's: what has made the major difference?

Some error in procedure that isn't immediately noticeable? The fact that the people making the other profiles are using better equipment? (Which would be a bit disappointing, considering what X-Rite's MSRP for the Pulse system was!) If the profile as first made by the Pulse ColorElite software isn't good as someone else's profile, can editing the profile improve matters considerably?

Is so, then: where might a person find guidelines on how to edit profiles to best effect? I can image it's possible to boost some hue that seems deficient, only to produce an even worse profile because it has now exceeded the paper's own gamut.
Logged

AaronPhotog

  • Full Member
  • ***
  • Offline Offline
  • Posts: 128
    • Dygart Photography
Good vs. not-so-good output profiles
« Reply #1 on: February 02, 2008, 10:55:01 pm »

Quote
I have seen more vibrant reds and yellows with Ilford Gold Fiber Silk than I've seen in prints of the same image on Harman Gloss. Might the difference be due to innate characteristics of the two papers?
[a href=\"index.php?act=findpost&pid=171874\"][{POST_SNAPBACK}][/a]

Mike,

In my recent testing of the new FB type papers by the numbers, I've let QTR send calibration ramps of each ink to each paper, mainly to study their black and white response, but I also just checked these two papers for Magenta and Yellow maximums (no profile involved, just telling each ink to lay down a 20 step ramp to the maximum) with my Epson 3800.  The difference between the two papers you mention is virtually nil, with perhaps a tiny bit lighter response from the Gold Fiber Silk.  Both papers should give you equally rich reds.  

Prints I've made of a volcanic fountain (from a scanned Kodachrome) with strong glowing orange reds are quite rich on both the Harman and the Gold Fiber Silk, such that the differences are mostly in the surface texture and gloss differential.  Black densities (dMax) are also practically identical.  It comes down to personal preference.  

Almost all really good profiles have gone through some editing, and there may be profiles tuned not only to specific papers but to specific purposes as well.  For a good discussion and example of profile editing see Bruce Fraser, Chris Murphy, and Fred Bunting's "Real World Color Management" second edition, published by Peachpit Press, 2005.

Aloha,
Aaron
Logged
Aaron Dygart,
Honolulu

Mike Arst

  • Full Member
  • ***
  • Offline Offline
  • Posts: 148
Good vs. not-so-good output profiles
« Reply #2 on: February 03, 2008, 03:03:04 am »

Quote
... good discussion and example of profile editing see Bruce Fraser, Chris Murphy, and Fred Bunting's "Real World Color Management" second edition, published by Peachpit Press, 2005.
Aaron -- thanks for your reply. I've stayed away from the book up to now, thinking that it must be intended for an audience way more sophisticated than I am. Now I'll have to pretend I'm sophisticated enough to qualify for reading the book. :-) I can imagine that the ability to edit profiles would be a huge 'plus' for specialized purposes. Intriguing...

I'm glad to hear that the two papers' responses are near-identical per your test. That would eliminate a question of 'inferiority' in the Harman paper and point a finger at the output profile. (I don't know if these papers respond much differently to Ultrachrome K3 ink, compared with the Ultrachrome ink used in the R1800, which I print with. Seems unlikely, but then I haven't ever used the K3 inks and haven't got a way to compare them.)

... a while after writing the above ... now I've printed the same shot on the Harman gloss using both Harman's canned profile and one I created. I'm relieved -- the print done with the custom profile is considerably better in the bright/deep reds. The reds and oranges on the Ilford sample are still a tiny bit brighter (as you noticed). What's more interesting is that the Ilford print is showing somewhat better-looking midtones than either Harman's or my profile for the Gloss FB AI. Hmm.

What a night-and-day difference between the two canned profiles. Ilford's is much better than Harman's, which blocked up the shadows. Some shadows in the image that are a deep red are dark and muddy (and with not much red showing) in the print done with the Harman profile.

If I can get a batch of the Ilford paper that doesn't curl strongly toward the back side (which so far has made using it a hit-and-miss proposition with the R1800 <grumble>), I'll be able to make a custom profile for it and then compare again.

Excellent papers, these. I hope Harman makes good on something they wrote about in e-mail some months back: to have Wilhelm test their baryta papers with Epson (or other) inks.
Logged

Pete Berry

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 445
Good vs. not-so-good output profiles
« Reply #3 on: February 03, 2008, 05:04:01 am »

Quote
Mike,

In my recent testing of the new FB type papers by the numbers, I've let QTR send calibration ramps of each ink to each paper, mainly to study their black and white response, but I also just checked these two papers for Magenta and Yellow maximums (no profile involved, just telling each ink to lay down a 20 step ramp to the maximum) with my Epson 3800.  The difference between the two papers you mention is virtually nil, with perhaps a tiny bit lighter response from the Gold Fiber Silk.  Both papers should give you equally rich reds. 

Prints I've made of a volcanic fountain (from a scanned Kodachrome) with strong glowing orange reds are quite rich on both the Harman and the Gold Fiber Silk, such that the differences are mostly in the surface texture and gloss differential.  Black densities (dMax) are also practically identical.  It comes down to personal preference. 

Almost all really good profiles have gone through some editing, and there may be profiles tuned not only to specific papers but to specific purposes as well.  For a good discussion and example of profile editing see Bruce Fraser, Chris Murphy, and Fred Bunting's "Real World Color Management" second edition, published by Peachpit Press, 2005.

Aloha,
Aaron
[{POST_SNAPBACK}][/a]


Your findings are very interesting and fall in line with my recent testing of 16 different papers from Hahne., Innova, Crane, Red River, and Harmon. I used a controlled setup also, printing from the driver using the same media type - "Special #5" in my iPF5000, which works with all feeds paths, for the eight "glossy" variants, and the all-path "Premium Matte Paper" for the true mattes.

In each group, each paper got exactly the same ink applied. I was looking to see the differences, which I thought would be obvious, and reveal something about the individual paper characteristics. But as the test progressed with two prints each of separate unmodified test images, I was amazed not only to see papers of a closely related type virtually identical in response (any differences directly correlating with paper whiteness or warmth), but also essentially identical output with papers as different as RR UltraPro Satin 2.0, Harmon Fb Al Gloss, and Hahne. Photo Rag Pearl and Fine Art Pearl.

Surfaces varied, of course, but with light behind the observer, each group was incredibly uniform, except for Hahne. Photo Rag Satin in the "glossy" group, which was very muted; and Crane Museo (original version) in the matte group, which did not take the ink well, but looked good except slightly muddy shadows after drying overnight. The two very inexpensive photo mattes (RR Polar and Premium Matte Plus) came off very well, the Prem. a little lacking in contrast and sat, but easily tweakable to be indistinguishable from the virtually identical premium fine art rag papers such as Innova Smooth Cotton High White 315 and Hahne. Photo Rag 308 (slightly more texture than the Innova), but much less that the German Etching, Museum Etching, and Wm Turner which also had identical color responses to my eyes.

There were differences between the two groups - all the "glossies" had significantly brighter reds, and again, virtually no difference I could see not related to paper tone. The mattes had identical slightly orangy reds by comparison, but Atkinson's luscious strawberries still looked plenty good in all. Basically, every print except with the Photo Rag Satin was more than acceptable.

I'm looking forward to getting Ilford Gold Fibre and Innova F-type series samples to complete my testing, then getting someone to do some objective measurements. Would you be interested Aaron?

My conclusion? Papers are incredibly more similar within their general group than I had believed possible. Certainly closer with this standard test than could be achieved with canned profiles from different sources; and likely, in my opinion, more consistent paper-to-paper than with custom profiles from the same source considering the many different surface characteristics and their effect on surface reflectiveness.

For more details, see my thread in the DP Review printer Printers/Printing forum:

[a href=\"http://forums.dpreview.com/forums/readflat.asp?forum=1003&thread=26589047]http://forums.dpreview.com/forums/readflat...thread=26589047[/url]

Pete Berry
Logged

Geoff Wittig

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1023
Good vs. not-so-good output profiles
« Reply #4 on: February 03, 2008, 10:07:23 am »

Quote
Your findings are very interesting and fall in line with my recent testing of 16 different papers from Hahne., Innova, Crane, Red River, and Harmon.
Pete Berry
[a href=\"index.php?act=findpost&pid=171949\"][{POST_SNAPBACK}][/a]


Pete-
Thanks for posting this; results of systematic tests are always more useful than unsupported assertions.
Inkset and printer are a major variable in the equation, of course. I've noted significant differences in output from the Epson K3 printers compared to the HP Z3100 on the same paper type, sometimes to the extent that my preference reverses.

I would note that Hahnemuhle photo rag satin is a unique paper, and doesn't belong with the other semigloss/satin paper types with their high contrast and neon reds. Its surface looks just like "plain" HPR before printing, but takes on a really unusual surface sheen once ink hits it. And it's quite sensitive to ink type. At least in my experience prints from the Z3100 on HPR satin look much better than those from a K3 printer.

Geoff.
Logged

AaronPhotog

  • Full Member
  • ***
  • Offline Offline
  • Posts: 128
    • Dygart Photography
Good vs. not-so-good output profiles
« Reply #5 on: February 03, 2008, 12:16:25 pm »

Mike,

Another comment.  There are things that can affect a papers response besides the profile.  For example, with the Epson 3800 and others, one can choose between output at 2880 dpi and 1440 dpi.  Also, you can control whether it is uni-directional or bi-directional.  Visually, the 1440 dpi prints look similar to the 2880 dpi prints in terms of sharpness and smoothness on a specific paper.  But...the 2880 prints will reach a higher dMax on some papers, such as the Ilford Gold Fiber Silk.  You will get different response curves.  I've stuck with 2880 for the tests.

One of the papers I tested, a matte type paper, was said to also take PK type (glossy) black inks.  So, I gave it a try.  It took the ink alright, with no smearing of the black ink, but it produced much flatter curves, and a much lower dMax.  It was a disaster waiting to happen for anyone who blindly followed the advice to switch away from the MK inkset.  

There are so many variables to play with when using the regular printer driver and photoshop that one can easily "program in" an inconsistency when attempting to make a comparison.  I know, I've done it.  So, stick to one resolution, and take as many variables out of the test as possible.

Paper characteristics can also affect a given paper's response.  A certain paper that I tested was comparatively thin, and exhibited a nasty curl, but worse than that, it had a severe reversal in the deepest blacks.  Several papers seem to reverse when you push the ink beyond the point at which the hit their maximum densities.  This one was the worst (it shall remain unnamed for now).


Pete,

I read your posting on your tests.  Excellent.  Yes, your conclusions are right in line with what I'm finding.  Not sure how I could help much.  I'm also using the same Gateway monitor as you, plus the Spyder 3 and Datacolor spectrometer, so it takes some time, but the spectro is very good at making measurements with good consistency and accuracy.  I probably would have gone down the path of trying off-the-shelf canned profiles, except that I have a printer that came uncalibrated and simply wouldn't work with them.  Two "replacement" machines arrived with much worse problems and went back.  I eventually wound up using similar "ink offsets" as you describe, to make and use my own profiles.  

I think that any printer that isn't calibrated properly to begin with probably could produce better profiles by first getting it to print the targets such that they come within a reasonable range of well-calibrated screen images (my early attempts to calibrate the printer with ColorBase had failed).  Fortunately, none of this comes into play with the QTR calibration sheet tests.  

I'm considering StudioPrint RIP, which can linearize printer response, as a later option to my inking offsets approach, but for now, the profiles I'm getting really work quite well.  Bill Atkinson's test image and others print very nicely and match the screen visually.  Traditionally correct approaches, though, just don't work for this particular printer.

Aloha,
Aaron
Logged
Aaron Dygart,
Honolulu

Mike Arst

  • Full Member
  • ***
  • Offline Offline
  • Posts: 148
Good vs. not-so-good output profiles
« Reply #6 on: February 03, 2008, 03:24:32 pm »

> take as many variables out of the test as possible.

I'm aware of the variables and with the R1800 have standardized on uni-directional printing and on what Epson calls "Best Photo." I assume that means 2880 DPI. The "Photo RPM" setting might be useful for something but I have yet to see an improvement when I've used it. (If the improvement isn't visible without using a loupe, it seems hardly worth using the additional ink). A few years ago I did a few tests in which I changed paper settings while using a single (matte) paper. I noticed distinct differences in color "bias" in the different paper-type settings. It's too bad Epson doesn't provide documentation about the expected effects of changing these settings with a given paper. They never have that I know of, and I've never run across anyone who has real-world information about those subjects. Surely there's someone at Epson with all of the information. But for whatever reason they keep it close to the corporate chest.

It's a shame that the Epson drivers -- those I have used so far, anyway -- aren't "modular," enabling you to add or remove paper types. There isn't a way to add even Epson's own papers as new ones become available -- IMO, a pretty serious shortcoming in the software. (Perhaps the drivers for the larger "pro" printers are "modular" in that way.)
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Good vs. not-so-good output profiles
« Reply #7 on: February 03, 2008, 03:41:08 pm »

Quote
Almost all really good profiles have gone through some editing, and there may be profiles tuned not only to specific papers but to specific purposes as well. 

I disagree. A good profile will rarely if ever need profile tuning.

The few times I've needed to do so is for cross rendering (make an Epson simulate press sheet) or minor editing of a soft proof table. But again, that's really rare. And profiling tuning is absolutely not for the faint of heart and is a great way to waste hours and a lot of media. Fun when you're paid by the hour, using OPM (Other People's Media). Don't go there.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

papa v2.0

  • Full Member
  • ***
  • Offline Offline
  • Posts: 206
Good vs. not-so-good output profiles
« Reply #8 on: February 03, 2008, 04:28:12 pm »

A couple of things to consider

Drying time from print to profile. This can have a effect on colour measurement.

Repeat measurements five times and then an average of the measurements to go to the profiling software.
Instrument calibrated properly. Check every thirty readings.

Diferent software packages build their profiles in different manners.

How many patches on the test target. Usually the more the merrier for the LUT builder.

Is there UV brighteners in the papers? This can effect the instrument. Has it a UV cut option?

Did a test not so long ago where i profiled a desktop inkjet using two top profiling packages and compared the results with the supplied profile from epsom. The manufactures profile did better in the round trip testing than my own.

Conclusions. Profiles supplied by manufacturer are pretty good nowadays for their own ink paper combinations. Only need to profile for non documented paper ink combos.

Oh and as Andrew says Profile editing is not for the faint hearted.
« Last Edit: February 03, 2008, 04:29:14 pm by papa v2.0 »
Logged

Mike Arst

  • Full Member
  • ***
  • Offline Offline
  • Posts: 148
Good vs. not-so-good output profiles
« Reply #9 on: February 03, 2008, 05:43:34 pm »

Quote
A good profile will rarely if ever need profile tuning. [...] Don't go there.
I assumed that profile editing would be quite an adventure time-wise, to say nothing of the cost ink+paper-wise.

The printer I use (R1800) isn't fine-tuned at the factory in the way a 3800 is. I'm sure there's liable to be more variation from one R1800 to the next, and I would think that canned profiles for the R1800 are liable to be much less accurate than canned profiles for a printer like the 3800. Correct assumption?

One thing I haven't delved into is how the output will vary depending on the ICC (v.4 versus v.3) setting, and the illumination setting (D50, D65, etc.) when the profile is saved. I don't use a D50 viewing source and have been selecting D65 up to now. I tend to prefer the way the prints look under tungsten illumination (warmer), but that seemed a bit limiting. All of the profile data is saved in .txt files, so it would be easy enough to build alternative profiles without having to reprint targets and measure them again.
Logged

Pete Berry

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 445
Good vs. not-so-good output profiles
« Reply #10 on: February 03, 2008, 05:46:41 pm »

Quote
Pete,

I read your posting on your tests.  Excellent.  Yes, your conclusions are right in line with what I'm finding.  Not sure how I could help much.  I'm also using the same Gateway monitor as you, plus the Spyder 3 and Datacolor spectrometer, so it takes some time, but the spectro is very good at making measurements with good consistency and accuracy.  I probably would have gone down the path of trying off-the-shelf canned profiles, except that I have a printer that came uncalibrated and simply wouldn't work with them.  Two "replacement" machines arrived with much worse problems and went back.  I eventually wound up using similar "ink offsets" as you describe, to make and use my own profiles. 

I think that any printer that isn't calibrated properly to begin with probably could produce better profiles by first getting it to print the targets such that they come within a reasonable range of well-calibrated screen images (my early attempts to calibrate the printer with ColorBase had failed).  Fortunately, none of this comes into play with the QTR calibration sheet tests. 

Aloha,
Aaron
[a href=\"index.php?act=findpost&pid=172001\"][{POST_SNAPBACK}][/a]


I printed with the driver of my i9900 for about four years with as good result as I could imagine - limited more by artistic factors and my PS abilities. Never knew what a paper profile was, and wouldn't have cared much if I did.

With the iPF5000 a year ago, I started with the driver tweaked  to my papers (with mimimal changes) and was very happy with the output. Then I read Andrew's book (most of it , anyhow!), which got me hoping for possibly more. So I tried a number of manufacturers' and better canned profiles. All needed tweaking to approach the driver's results.  So why tweak the tweaker rather than going straight to the source? Some looked artificial, and one had definite banding in the linear spectrum in Digigog's test image indicating probable small patch sample or mis-reading (page 243!).

 ColorHQ owes me four custom profiles, so I may collect and add another set of prints on my most used papers from these. If they're better, Ill use em'.

Pete
Logged

Mike Arst

  • Full Member
  • ***
  • Offline Offline
  • Posts: 148
Good vs. not-so-good output profiles
« Reply #11 on: February 03, 2008, 05:58:23 pm »

papa v2.0 >Drying time from print to profile.

I let them dry at least a couple of hours -- usually overnight.

> Repeat measurements five times and then an average of the measurements to go to the profiling software.

That's something I haven't done yet. I'll have to review the X-Rite documentation for the procedure.

> Instrument calibrated properly.

I have been calibrating it before each reading. (The software almost always insists that the previous calibration is out of date, anyway.)

> How many patches on the test target.

I always go for the maximum number of patches.

> Is there UV brighteners in the papers? This can effect the instrument. Has it a UV cut option?

I don't see this option in the software's various dialogs. Unfortunately, X-Rite 'orphaned' this system and there's no longer a way to upgrade to v.2.0 of the Pulse ColorElite software (grumble). If the brighteners can be compensated-for in that version, too bad for me as it's out of my reach.

> Conclusions. Profiles supplied by manufacturer are pretty good nowadays for their own ink paper combinations. Only need to profile for non documented paper ink combos.

Judging by my experience with the Harman 'canned' profile for my printer: that profile isn't good enough. It tends to muddy the shadows. Reds, yellows, and oranges are less vibrant than they should be. The custom profile improved all of the above quite a bit. Ilford's canned profile for the Gold Fiber Silk paper seemed to be a lot better than Harman's profile for its Gloss FB AI. And the shadow detail is better yet with the Ilford profile that either Harman or I could provide for the Harman paper! I haven't yet made my own profile for the Gold Fiber Silk for comparison. (I won't be able to use the paper again with this printer until they solve the unpleasant paper-curl problem. The curl has ruined too many sheets.)

I have wondered if ambient illumination can affect the scanning. I keep the room lights on -- they aren't super-bright -- but turn off any nearby desk lamps in case they could have an effect.
Logged

AaronPhotog

  • Full Member
  • ***
  • Offline Offline
  • Posts: 128
    • Dygart Photography
Good vs. not-so-good output profiles
« Reply #12 on: February 03, 2008, 06:23:04 pm »

Dig. Dog and the late Bruce agree that it's not a trivial pursuit.  Sometimes the best approach with a bad profile is to ditch it and start over.  

Papa's right, too.  For well calibrated printers, the manufacturer's and purchased canned profiles are now being made for them by some real experts, and work quite well... except for my printer.  It can't use them.  So, I've made my own using the Datacolor spectro.  The program has some ways to tweak profiles that aren't that tricky, though.  You can even add curves, and make other adjustments easily, and it isn't a horror show at all if you do it carefully.  Yes, you'll go through some ink and paper.  I usually don't have to do much adjusting, but I've done it with successful improvements.

Drying time is indeed important.  Give it time.  You'll get significantly different results with some of these papers after a day of drying time vs. an hour (depending on your humidity I suppose).

As for UV, you can still build a good profile for papers that use a lot of optical brightening agents (OBA's).  Good profiling programs take that into account fairly well.  In my tests, it's easy to get an idea of the extent of OBA use.  I know that Andrew (who also has an excellent book on color management) advises to stay away from OBA's but they are there, like it or not, in some measure, and it's hard to avoid them altogether.  The brightener salesmen are doing their jobs (silver print papers have them too).  The Harman and Ilford papers discussed in this post use them sparingly.  The Epson and Hahnemule FB papers use them a bit more.

There is another reason to avoid papers with excessive brighteners, though.  In some limited perceptual testing that I did by just showing my test sheets to different people and asking questions, excessive use of brighteners can cause another variable to enter the printing equation.  Some people are less able to respond to OBAs than others, perhaps because of age, contact lens use, or just different color sensitivities.  For example, some people have correctly identified the Epson Exhibition paper as darker than the Harman, while others see it as brighter.  A friend who recently had cataract surgery saw the Epson as brighter than the Harman, while her daughter (wearing contacts) correctly saw the Harman as the brighter of the two.


Aloha,
Aaron

"It's all about the seeing."
Logged
Aaron Dygart,
Honolulu

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Good vs. not-so-good output profiles
« Reply #13 on: February 03, 2008, 06:52:02 pm »

Quote
So, I've made my own using the Datacolor spectro.  The program has some ways to tweak profiles that aren't that tricky, though.  You can even add curves, and make other adjustments easily, and it isn't a horror show at all if you do it carefully.

They provide that because, out of the box, often their profiles kind of suck (and having a true Spectrophotometer would help).
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

AaronPhotog

  • Full Member
  • ***
  • Offline Offline
  • Posts: 128
    • Dygart Photography
Good vs. not-so-good output profiles
« Reply #14 on: February 04, 2008, 02:07:05 am »

Quote
They provide that because, out of the box, often their profiles kind of suck (and having a true Spectrophotometer would help).
[a href=\"index.php?act=findpost&pid=172097\"][{POST_SNAPBACK}][/a]

Andrew,

To be sure, a spectrophotometer enables the use of larger numbers of patches, which is sometimes, but not always, better.  A spectrophotometer such as the Eye-one is not necessarily any more accurate than the Datacolor spectrocolorimeter.  Their own claims for tolerance and repeatability indicate that it's approximately a dead heat.  My own tests show the Datacolor unit to be very accurate.  Nor have I found that the Datacolor profiles often "suck."  With care in reading, my experience has been that they are often excellent.  The ability to easily adjust them is a bonus, not a compensation for inadequacy.  Their greyscale chart adds excellent control of near-neutral tones.  Their value for the money is amazingly good for the little guy like myself who isn't going to be using spectro's a lot, but who needs to take personal control of color accuracy.  No, I don't work for them, if it sounds too much like a plug.  If I could afford an iO I'd probably go for that, and may do so later, but right now I'm happy with what I have.

Mike,

Here's another opinion, for what it's worth (I've said it here before);
The business about the 3800 being fine-tuned or calibrated at the factory is primarily advertising hype.  They cannot identify their "special manufacturing process" and you don't get a certificate of calibration as you would with any truly calibrated device.  You do get one saying it meets class B radio emission requirements.  There are good 3800's and bad 3800's, based on my experience, posts in this forum by others, and from my discussions with Epson.  They turn out a lot of good printers and some that are in need of some serious adjustment, and a few that are train wrecks.  Maybe they get a little beat up going from China to Japan to Epson headquarters, then to a dealer, and then finally to the customer's location.  

Keep in mind that all such devices can drift over time, too.  If you check Eric Chan's website, you'll see that he frequently recalibrates his 3800 with ColorBase, an Epson program that mysteriously is not avaliable from the Epson USA site.  The US service department persons didn't even know about it when I called.  Now they do.  Check the UK website.  There's a way to use the Datacolor spectro to populate the tables it reads, but an eye-one would be easier to be sure.  It would also be my procedure if it worked with this machine.  It turns out that Epson's service technicians do have a way to re-calibrate the machine at the hardware level, but I don't think I'll fly one to Hawaii just for that.

Meanwhile, use what works for you and enjoy the images.

Aloha,
Aaron
« Last Edit: February 04, 2008, 02:16:26 am by AaronPhotog »
Logged
Aaron Dygart,
Honolulu

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Good vs. not-so-good output profiles
« Reply #15 on: February 04, 2008, 09:25:01 am »

Quote
Andrew,

To be sure, a spectrophotometer enables the use of larger numbers of patches, which is sometimes, but not always, better.

Got nothing in this context to do with number of patches but with the data being measured.

In the  future, you'll see an instrument that makes profiles of superb quality with 10% of that rivaling those you get with 1700.

Quote
A spectrophotometer such as the Eye-one is not necessarily any more accurate than the Datacolor spectrocolorimeter.  Their own claims for tolerance and repeatability indicate that it's approximately a dead heat.  My own tests show the Datacolor unit to be very accurate.

What reference Spectrophotometer did you use to measure this accuracy? There's a difference in accuracy and repeatability.

Quote
Nor have I found that the Datacolor profiles often "suck."

There are reports all over the web of this (and to be fair, of those who have made very good profiles too). I suspect its ink or paper dependent.


Quote
With care in reading, my experience has been that they are often excellent.  The ability to easily adjust them is a bonus, not a compensation for inadequacy.

One would question why you need to tweak the profile and why doing more work (and slower) and printing more media isn't a compensation for an inadequate profile.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Good vs. not-so-good output profiles
« Reply #16 on: February 04, 2008, 09:35:41 am »

ColorBase was a product designed to make multiple Epson's in one environment match. And I have to say, the US product managers were probably correct that it was more work and unnecessary than need be, based on the work I've done and that of others like Mac Holbert etc. When we built the profiles for the Exhibition Fiber paper, we measured 5000 patches off multiple machines (3800 included and not only that, 2400!). The average deltaE00 of the various targets from various locations was well below 1! And some of these machines were brand spanking new, others (like my 3800 and 4800) were older. The consistency of the units is startling and this has been demonstrated going back to older models where Bill Atkinson built and supplied profiles that I suspect thousands if not more users found to be superb in terms of quality.

Initially Epson wasn't going to have us do the 2400 but apparently that unit too undergoes some consistency checking on manufacturer and so that product manager asked if we'd build profiles for that unit too. Again, the consistency of supplied measurements off multiple machines, all at best behavior (heads aligned, inks all firing) are amazingly consistent from unit to unit. I have the measured data to back this up.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

allan67

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 69
    • http://
Good vs. not-so-good output profiles
« Reply #17 on: February 05, 2008, 02:26:01 am »

Quote
There are reports all over the web of this (and to be fair, of those who have made very good profiles too). I suspect its ink or paper dependent.
[a href=\"index.php?act=findpost&pid=172197\"][{POST_SNAPBACK}][/a]

Hello,

I have to agree with Andrew on this. I have Datacolor unit and built some profiles for my R1800 with it.
Some paper-ink combinations profile without problems, while others just don't, whichever settings I try (and with the help of Datacolor tech support).
For example I was never able to get a consistent profile that even approaches canned Ilford's for Ilford Gallery Smooth Pearl paper using Epson inks.
On the other hand, Ilford Gold Fibre has profiled beautifully. I was able to get a better gamut and shadow and midtone separation from my custom profile (without any tweaking) than from canned Ilford's profile.

Allan
« Last Edit: February 05, 2008, 02:28:20 am by allan67 »
Logged

AaronPhotog

  • Full Member
  • ***
  • Offline Offline
  • Posts: 128
    • Dygart Photography
Good vs. not-so-good output profiles
« Reply #18 on: February 05, 2008, 02:38:03 am »

I didn't mean to push your buttons there dawg.  I respect you and the work you do.  I have your book too.  Meanwhile, though, I guess I've been lucky(??) to have had, so far, two Epson 3800's that overinked severely and in the same general direction, and another one that was just plain busted.  I learned that under the overinking conditions, the Datacolor spectro couldn't produce a working profile.  No canned profile, Epson's or anyone elses, worked either.  Talk about delta e's, these were big enough to saddle and ride.  I eventually figured out that the patches had to be within the correction range of the device to begin with.  Once I dialed in the right "inking offset" corrections, I was able to immediately see a major - night and day - difference.  Then the profiler could do it's corrections accurately, and without any adjustments most of the time.  

As to the accuracy of the device and its program, I did some round trips and found it to be very accurate that way, not with a reference spectrometer, but with known Lab values, and what the spectro read and what Photoshop reported.  The original results had not been close.  After profiling and printing with my inking offsets, the results were, as I said before, excellent.  That's one reason I'm happy now.  Another is that I didn't have to spend more than the cost of the printer to get where I am now with it.

There was a review of several profilers (I think it was on Luminous Landscape), including the PrintFixPro, and it did surprisingly well.  There were a couple of better ones, and a few worse, but once you get me within range, I can usually get what I want, after having had over 52 years in the black and white and color darkroom.

OK. Here it is:
  http://www.luminous-landscape.com/essays/f...graphics2.shtml

Here's a discussion of ColorBase:
http://www.luminous-landscape.com/reviews/...colorbase.shtml

And here's a comparison, including cost, on spectrometers and the DataColor Spectrocolorimeter (admittedly their own):
http://www.colorvision.com/product-compare1005.php

As I said before, if I were able to afford it, I'd go for the better ($$) units and profiling programs, but for now, this works to the extent that I will say that yes, my 3800 is an excellent digital printer.  It always has been in black and white, and is now in color as well.  

What about your Epson profiles with my offsets?  Would it be an improvement, or not?  Maybe I'll give it a try and see how close my machine is to the ones that are your standard.

As an aside, I still prefer the darkroom, though (another topic for later perhaps), but I feel that it's important to get a handle on the digital world as it continues to evolve.  

Meanwhile, "good negs."

Aloha,
Aaron
« Last Edit: February 05, 2008, 04:03:27 am by AaronPhotog »
Logged
Aaron Dygart,
Honolulu

Pete Berry

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 445
Good vs. not-so-good output profiles
« Reply #19 on: February 05, 2008, 04:25:42 am »

Just finished running some Ilford Gold Fibre through my standardized driver managed paper test. As with the other eight glossie variants, it printed beautifully - slightly warmer than the cooler base Harmon Fb Al Gloss, slightly cooler than the warmer Hahne. FA Pearl. We're talking gnat's eyelashes here, for the most part.

Then I tried Ilford's GFS profile for the 5000 - what a mess! Andrew's model was pale and yellowish. All colors weak, with a very curious transition from blue to cyan in the linear spectrum with a dark magenta zone actually to the right of blue and interposed with a sudden bright cyan zone.

This did nothing but confirm my past experiences with canned profiles. Seeing only a print without the evaluation tools Andrew has put into his test image, I would have rejected the paper out of hand as truly lousy rather than the result of a bad profile.

Driver printing with the uncorrected media setting the profile used - Photo Paper Plus S-G, was much improved, and seems identical on first look to "special #5" which works with all feeds. The same corrections gave a very nice image with the PPP SG setting.

I'm done with this evaluation project, every paper like deja vu all over again. Time to try it out in the real world.

Pete
Logged
Pages: [1] 2   Go Up