HowTo? fix semi-random PMP5 print profiles with highlights/borders turning gray?

[l_d_allan]

Continuation of topic:
http://www.luminous-landscape.com/forum/index.php?topic=75707.0

iSis with PMP-5, Win7, CS5/LR4, TC918 target, Canon Pro-9000-2 printer

I've been unable to get consistently usable printer profiles using PMP5 to generate printer profiles from test-chart targets scanned with my iSis. Perhaps 75%+ of the print-previews and prints end up with the highlights/borders turning gray, but the rest of the mid-tones and darker parts of the print are ok. I'm baffled.

As shown in the attached, with a flawed profile, the borders and highlights turn gray when I select the options in the CS5 Print-Preview for "Match Print Colors". They get more gray when the "Gamut Warning" option is selected. The rest of the print doesn't seem to change. It's not just getting darker.

I’m pretty sure it isn’t just the “gamut warning” in the Print-Preview. The print itself comes out gray. The gray can range from light gray to dark gray in the same print when different print profiles are applied.

I've measured a test-chart with PMP-5, and gotten a flawed profile. Then I'll re-measure and re-generate the profile with PMP-5, and it will sometimes be ok. Using the Measure-Tool's "Compare", the scanned values will be very nearly the same ... within 0.15 De2k ... which I understand is normal measurement variation.

When I use a text-editor to take a closer look at the MyProfileCgats-ok.txt and My ProfileCgats-flawed.txt, I can't see any difference in the TDL (target definition language).

I've tried pretty much all the combinations of PMP-5 preferences, and none seem to make a difference ... ver-2 vs ver-4, paper-gray vs neutral-gray, large-vs-default, etc.

I've also used i1Profiler 1.4.2 (demo) to control the iSis and make spectro measurements. I'll output to Cgats-PMP5 format and then use PMP-5 to generate a print profile from the i1Profiler measurements. Maybe 98% are flawed, but oddly a very few have been usable ... they don't have the gray highlights/borders. I haven't been able to discern the difference on my procedure that results in a usable vs flawed profile. It just seems semi-random.

The iSis seems to be working ok ... I sometimes do get a print profile that works fine, which suggests it is working ok. I've gotten some reference QPcards to check further ( see http://www.luminous-landscape.com/forum/index.php?topic=76304.0 )

I'm pretty sure I'm doing something wrong ... "some village is missing its idiot" ... but haven't figured out the missing key.

I've been wrestling with a work-flow using ColorPort + PMP-5 (seems better), and just starting to try ColorPort + ArgyllCms.

[Scott Martin]

One option is to just upgrade to i1Profiler which doesn't suffer from this problem. The quality of the profiles alone may make you wonder why you didn't switch earlier. PMP5 wasn't a great print profiling solution 5+ years ago and it's becoming increasingly problematic to try and keep using it today.

[Rhossydd]

One option is to just upgrade to i1Profiler which doesn't suffer from this problem. The quality of the profiles alone may make you wonder why you didn't switch earlier.

I assume you haven't been following L_D_Allen's posting here, if you had you'd know he's bought the iSis+PMP off eBay, so another $500 on a further software upgrade might not affordable. Especially so since his lack of results so far hasn't given him any confidence in the iSis anyway. If the iSis does prove to have faults it all becomes a very expensive mess if you buy a Profiler upgrade and then have to buy another working spectro as a result.

PMP5 wasn't a great print profiling solution 5+ years ago and it's becoming increasingly problematic to try and keep using it today.

PMP5 delivers very good profiles, yes i1Profiler delivers better results, but many people wouldn't notice much difference between the two.
Keeping it running ? absolutely NO problem on Windows systems like the OP runs.

[Rhossydd]

L_D_Allen: It sounds like you've got into a complete mess with this, trying too many different things before you have a decent understanding of the processes involved.

Here's a plan
Start again from page 1 and get what you've bought working properly and don't distract yourself with other software. Start a word document and record every detail of what you do, include screen shots. Make sure any print outs are marked with unique and full descriptions of how they've been printed. Then you'll have a record of what you've done and what the result has been.
Keep all new work separate from anything you've done before to avoid confusion.

Choose a one page iSis target from those supplied with PMP5.
Print out some copies of it, maybe two of each, on a couple different papers with a program of known reliability for printing targets correctly; PS CS4 or The Gimp.
Leave to dry overnight. If you've printed multiple copies, check they all match. Check the sizing is correct.
Go into PMP's Profilemaker module, load the appropriate reference file and measure the target from there. Save the measurements.
Then build a profile with the settings I recommended on your previous thread.
(Use version 2 profiles
Large profile size
paper grey perceptual rendering intent
Logo colorful Gamut mapping.
D50)

Then try the profile. Use a standard test image light a PDI or Northlight, not synthetic colour ramps or Granger patterns.

Then report back if you see a problem with that specific workflow.

Messing around with camera profiling targets like the QP cards is very unlikely to give you any useful information. You might get some data you won't really have enough confidence in to draw any conclusions from, at worst you'll trash the iSis.

[Scott Martin]

I assume you haven't been following L_D_Allen's posting here, if you had you'd know he's bought the iSis+PMP off eBay, so another $500 on a further software upgrade might not affordable.

We all have to consider the state of software when purchasing hardware. If an i1P upgrade wasn't in the budget then something like a ColorMunki would have been a better solution.

Especially so since his lack of results so far hasn't given him any confidence in the iSis anyway.

The lack of confidence should likely be with the software then, not the hardware. Test the hardware by making an i1P profile. Send a measurement file to someone with it to have a profile made.

If the iSis does prove to have faults it all becomes a very expensive mess if you buy a Profiler upgrade and then have to buy another working spectro as a result.

Test the hardware by making an i1P profile. Send a measurement file to someone with it to have a profile made. Play around with i1P in demo mode. As it is, an iSis with PMP5 might have seemed like a good deal but he's obviously paying for it in other ways...

PMP5 delivers very good profiles, yes i1Profiler delivers better results, but many people wouldn't notice much difference between the two.

Then get a Colormunki instead. Even it makes better profiles than PMP5. FWIW, every single person I showed PMP5 vs Monaco Profiler test prints to could tell that MP was better. Every single person. And that was 8+ years ago. i1P's profiles are even better today.

Keeping it running ? absolutely NO problem on Windows systems like the OP runs.

In the long run he will though. He'll get more devices and PMP5 won't support them. He may want to share data between different apps and PMP's limited formatting will hold him back - you get the idea. And there's the existing PMP problems like scum dot and  the poor perceptual rendering... PMP is not a smart way to go with the iSis considering the options out there. Having an awesome device like that without i1Profiler is like having a Formula One race car without tires. Get i1P or a ColorMunki and move on - that's my advise, FWIW.

[digitaldog]

PMP5 delivers very good profiles, yes i1Profiler delivers better results, but many people wouldn't notice much difference between the two.
Keeping it running ? absolutely NO problem on Windows systems like the OP runs.

I have to agree. While I think i1P's new engine is better, PMP should not be presenting these issues. I suspect the iSis since:

1. I've built hundreds of profiles with ProfileMaker Pro without issue (with several iSis and other supported instruments).
2. Until another device is added to the testing matrix, we don't know what causes the problems unless the measured data is used elsewhere.

[l_d_allan]

2. Until another device is added to the testing matrix, we don't know what causes the problems unless the measured data is used elsewhere.

Agree. I had thought by now I'd have use of the X-Rite i1Profiler Photo Pro-2 package for a week, which includes the Pro-2 spectro and i1Profiler software. The organizer of a Meetup camera group I belong to is an X-Rite distributor, and owns the Pro-2 package. He has indicated he'll allow rentals after a group training session (attended), but that has been slooooooow to actually happen. Frustrating, but real-world, I suppose.

My crossed-fingers hope is that the Pro-2 provides spectro scan numbers using i1Profiler that agree with numbers from the iSis using i1Profiler and then PMP-5. And also spectro scan numbers from ColorPort. If so, then that will suggest that the iSis is working fine. Whew.

If not, then I'll have some difficult, expensive decisions to make. Or give up on printing? However, I'm unclear on what are reasonable expectations on how closely measurements from the iSis and Pro-2 should be. Under 1.0 de2k overall? What would be a "threshold" for determining that the iSis is out of calibration sufficiently that it would have to be sent to X-Rite for $400+ re-calibration? Minimum $600 repairs. $1000+ to update to Rev E specs.

Hopefully, I'll be among the first, if not first, to get rental access ... so the Pro-2 spectro won't be too bashed up from ham-fisted renters. It doesn't strike me as the kind of delicate, expensive, electronic equipment that it makes sense to rent, imo. Also, I can picture frustrated newbies throwing it across the room after going through a lot of expensive ink and paper, with mediocre profiles. As I can testify, there are a lot of things that go wrong (some village is missing its idiot?)

LensRentals.com rents the ColorMunki for a reasonable amount, but by the time you pay for shipping, it jumps close to $100 quickly. My impression is that the ColorMunki is a tougher piece of electronics than the Pro-2 spectro. Or not?

[l_d_allan]

L_D_Allen: It sounds like you've got into a complete mess with this, trying too many different things before you have a decent understanding of the processes involved.

Mostly agree.

Here's a plan
Start again from page 1

I appreciate the advice, and your continued and patient "way-beyond-the-call-of-duty" assistance to this floundering CM newbie. However, I've sort of "been there, done that" as far as starting at the basics with the supplied TC918. At least once. Maybe twice?

Sorry if I'm being defensive and coming across as unappreciative. Here's a bit of a summary of my "saga" from an earlier post:
http://www.luminous-landscape.com/forum/index.php?topic=75707.msg610424#msg610424

And sorry for the following long post which may be tl:dr

Originally, I started off with the default TC918 supplied with PMP5 for the iSis. Perhaps it was self-inflicted, but I experienced an "off-by-one" alignment problem early on, if not immediately. Unfortunately, I didn't notice that symptom other than terrible profiles, so I figured the TC918 and the iSis didn't play well together. After the spectro scan, it showed a "Measurement done ok" and I believed it. Drat. In retrospect, I perhaps gave up on the TC918 too soon.

That lead me to some huge wasting-of-time, rabbit trails, and frustration in the weeds trying to get the Measure-Tool to generate test-chart targets. I was 99.44% sure that would work.  But, no. Then beating my head against the wall with i1Profiler 1.4.2 demo, then ColorPort, back to PMP5, even some ArgyllCms .... blah, blah, blah. Sorry for the whining.

I did return to using the basic TC918 + iSis + PMP5 a week or so ago. I did find the TDL iSis_INFO parameter issues that I think caused the "off-by-one" alignment problems. I made a checker-board black/white pattern of the TC918 that gives me confidence that the alignment on patch-reading is now ok, rather than the earlier "off-by-one". As delivered, you could see the b/w patches blending together about half-way across the page, which threw everything off. That's solved, and probably "operator error" out of ignorance on my part. But the flaws with gray borders/highlights continued.

If I say so myself, I tend to be pretty organized as far as making notes in my "lab book". That's probably an understatement, especially after all the problems encountered that could be explained by carelessness and getting in a hurry and not in the best state of mind from frustration. That includes putting "procedure notes" on the print-outs as far as time-stamps, how printed, how measured, patch-counts, CS4/Acpu settings, even barometric pressure lately here in Colorado at 6500', etc.  

But I fully concur it is excellent advice to be methodical about tackling "one degree of freedom" at a time, starting from "first principles". There is a lot to go wrong, especially with the work-flow spreading out over multiple days as prints dry.

The status now is that the combination of Acpu/CS4 + TC918 + iSis + PMP-5 continues to mostly result in unusable, deeply flawed profiles with highlights/borders becoming gray. Every so often, that combination provides what seems to be a good profile. With Acpu + TC918 + iSis + ColorPort spectro scan, and then PMP-5 to actually generate a printer profile from ColorPort spectro scan measurements, the printer-profiles results are significantly improved as far as fewer prints with gray highlights/borders. I'm actually getting what seem to be very good printer profiles ... hurray/whew ... which suggests that the iSis is working ok.

Further "status" ... I came across info that the PMP-5 choice of "LOGO Classic" for the "Gamut Mapping" choice on a PC may have bugs. That might explain a lot. A whole lot. From my software experience, I try to figure that the hardware and commercial software are ok, and the problems are caused by my PEBKAC. Or not in this case? Something else to try.

Also, I've been grappling with using ArgyllCMS to generate printer profiles from ColorPort spectro measurements of the TC918 target. That seems to be working very well, but it is a bit early to have confidence in that work-flow from dried test prints like the one from OutbackPhoto. That combination would remove PMP-5 entirely from the work-flow. Good riddance?

And I keep hoping I'll have rental access for a week of a relatively new i1Profiler Photo Pro-2 bundle from the organizer of a photography oriented Meetup group. That has been delayed time after time, which has thrown me off on several occasions.

And BTW, it's Allan rather than Allen. But you can call me Ray, or you can call me Bill, or you can call me "clueless newbie" if you can help get the iSis working. 

[Rhossydd]

The status now is that the combination of Acpu/CS4

If you've CS4 on your system, you don't need to use ACPU.
On Windows ACPU resizes the charts and that maybe what's causing problems with the iSis.

[Rhossydd]

Then get a Colormunki instead. Even it makes better profiles than PMP5. FWIW, every single person I showed PMP5 vs Monaco Profiler test prints to could tell that MP was better. Every single person. And that was 8+ years ago. i1P's profiles are even better today.

Yes, no one is saying there haven't been improvements, but they come at a price.

All through photography we make value for money judgements. Is it worth upgrading from a Canon 5Dmii to a mkii ? Is it worth upgrading an Epson 3880 to a 3880 ? Is it worth buying Leica instead of a Fuji ?

PMP may well be good enough for most people. Only precious few would be able to see the differences on regular images in non-comparative viewing. That $500 might be much better spent on something that gives the Op a more significant improvement in their photography.

[l_d_allan]

If you've CS4 on your system, you don't need to use ACPU.
On Windows ACPU resizes the charts and that maybe what's causing problems with the iSis.

Thanks for the helpful suggestion. I'll give it a try, carefully, but I'm less than optimistic that will help. Certainly worth trying.

I've tried both, and ColorPort seems to deal as well or better with Acpu made prints as those prints from CS4. ColorPort seems to carefully measure the black alignment bar at the top, and proceed from there. That seems to be the basis of the patch width. It also seems to identify the triangle alignment marks to figure out the vertical patch size. Not bad for freeware!

I'm actually kind of impressed with ColorPort. Unfortunately, ColorPort targets aren't compatible with PMP5 or i1Profiler. At least I couldn't get targets to spectro-scan-measure ok, even with tweaks to the ColorPort TDL in the MyTarget.xml.

To my satisfaction, I've checked this auto-scale alignment with a black/white checker-board TC918 target that should clearly show if the iSis is getting "off by one" out of alignment. ColorPort seems to cope fairly well with "auto-scaling" of prints made with Acpu, vanilla unmodified CS4, and even prints tweaked with CS4 to use the full width of the paper so the patches are spot on 6mm rather than reduced to 5.7mm or so.

My observation is that i1Profiler also "auto scales". PMP5 seems to auto-scale when you tell it to base the target identification and patch alignment on the bar-code, but much less well when you have it ignore the bar-code, and proceed based on the reference file itself. Or not?

IIRC, I've read in multiple places that CS4 on the PC doesn't really turn off color management completely. Eric Chan from Adobe? I think Acpu was put together when CS5 removed that option, but then was found to work better and more reliably than CS4 with "No Color Management".

But of course, you can read pretty much any darn thing on the internet? Anyway, I've made target prints with both Acpu and CS4, measured with ColorPort controlling the iSis spectro scan, and then used MeasureTool's Compare utility. They were further off than I expected ... like a factor of 0.5 to 1.0 for De2k?

[l_d_allan]

All through photography we make value for money judgments.

Agree, and thanks for mentioning that to the person making the reply. It certainly crossed my mind, although perhaps not so diplomatically expressed.

I recall going to a photography Meetup group led by a working pro who stated something like "if it bothers you to spend $100 to $150 for some geegaw gadget (like a rarely used LensAlignment device in that situation), then photography is the wrong hobby for you."

Whoa.

Is this the hobby for me? I'm retired on a budget with a keen sense of cost justification. Must be nice to have your toys be a write-off, tax-wise.

At the time, I was getting hints of displeasure from my otherwise cooperative and long-suffering wife. Then I had a Canon 1000D XS DSLR, which was even more entry level than a Rebel 450 XSi. Plus another lens. Shopping for LowePro fanny pack. Wife: "When is this going to stop?" One has to be careful to avoid W.O.W. (wrath of wife)

As I think you (or someone else) pointed out, the other issue is that it still isn't for sure that the iSis even works well enough to produce acceptable profiles, so getting the i1Profiler Publish software upgrade could be even more "good money thrown after bad". I think it is about a 75% probability that the iSis works fine, but that may be wishful thinking.

In retrospect, I'd probably have been much better off renting a ColorMunki from LensRentals.com, and then maybe watching for a sale.

Or worked out a return, if possible, with the eBay seller (Mac user) when I had the original problems. "Good luck with that."

But ... "that train left the station" quite a while ago.  I have at several points in time considered giving up and "cutting my losses". Then I step away for a week or so to calm down. Plus I've gotten lots of valuable suggestions from LuLa. There was a character in a movie who expressed" "Young man, I have determination for breakfast!"

[l_d_allan]

Logo colorful Gamut mapping.

Further "status" ... I came across info that the PMP-5 choice of "LOGO Classic" for the "Gamut Mapping" choice on a PC may have bugs. That might explain a lot. A whole lot.

Well, finally. Looks like progress.

The choice of "Gamut Mapping" options in PMP-5 between "Colorful", "ChromaPlus", and "Classic" seems to be the key to the flawed profiles with highlights/borders being gray. Essentially all the profiles I made were with the (default?) of "Classic", and almost all were flawed. I tried several iSis measured targets using "ChomaPlus" or "Colorful", and they seem fine. At least the highlights/borders are ok.

I saved the .txt files for the profiles that looked gray, and put them aside in a "..\FlawedGray" subdirectory. I'll re-run some of them with PMP-5 having the "Gamut Setting" other than "Classic" and confirm they are better. I'll make some prints of the OutbackPhoto test file and see how they look after drying.

I do believe I came across descriptions in X-Rite/GMB documentation that "Colorful" and "ChromaPlus" resulted in profiles that were in some sense less "faithful" and "neutral" in color in resulting prints, compared to "Classic". Since "Classic" was the default, that's what I used. My bad?

I did check, and several weeks ago I did make a series of profiles using the multiple alternatives of "Profile Size" (Large vs Default), Perceptual Rendering Intent (Neutral vs Paper-colored), and "Gamut Mapping" (Classic vs Colorful vs ChromaPlus). For whatever I was checking, the choices of "Gamut Mapping" didn't make any difference, so I stayed with the default of "Classic". So I apparently assumed that the "Gamut Mapping" choice wasn't the culprit  Drat.

[Rhossydd]

Well, finally. Looks like progress.

Good to hear, but I did tell you what settings to use three weeks ago.

...I tried several iSis measured targets using "ChomaPlus" or "Colorful",...

You don't measure with a gamut mapping option.
The profile building options; profile size, perceptual rendering intent, gamut mapping are only used when building the profiles from a measurement set.

If you know what you're doing building a set of test profiles with every option possible takes very little time in PMP5 using the batch option. Within half an hour you'll find the problem lies with the 'classic' (ie old) gamut mapping method on all paper grey 'perceptual rendering intent' option built profiles using either rel col or percep RI.
If you use neutral grey 'perceptual rendering intent' you see a similar behaviour only when using rel col RI.

In retrospect, I'd probably have been much better off renting a ColorMunki from LensRentals.com, and then maybe watching for a sale.

The smart thing to do would be to leave building custom printer profiles to the experts. With a custom printer profile costing less than $20 it seems crazy to spend big amounts on highly specialised kit that won't get used often and you haven't got the expertise to use effectively.

[l_d_allan]

Good to hear, but I did tell you what settings to use three weeks ago.You don't measure with a gamut mapping option.

Shortly after you suggested those, I did try the 12 combinations of Large-Default, Neutral-PaperColored, and Classic-Colorful-ChromaPlus. I don't know why, but I didn't see the Classic problem. Careless on my part? Or the semi-random nature of the issue, so it was bad luck that it wasn't obvious? Anyway, in retrospect I can see how my detective work could have been more effective, and saved myself a lot of frustration and time.

The profile building options; profile size, perceptual rendering intent, gamut mapping are only used when building the profiles from a measurement set

Sorry, I wasn't clear. I measured with the iSis, and used those measurements within PMP-5 to generate the profiles with Colorful and then ChromaPlus selected.

If you know what you're doing building a set of test profiles with every option possible takes very little time in PMP5 using the batch option. Within half an hour you'll find the problem lies with the 'classic' (ie old) gamut mapping method on all paper grey 'perceptual rendering intent' option built profiles using either rel col or percep RI.

I'm taking a closer look at "Batch". I doesn't appear to handle a directory at a time, like I hoped. It does seem to offer some advantages to prepare a group of profiles with different settings. Beats making one profile at a time, especially using the Large setting which takes longer. Thanks for the suggestion.

If you use neutral grey 'perceptual rendering intent' you see a similar behaviour only when using rel col RI.

I was seeing odd print-previews from Perceptual (often ok) compared to RelativeColor (often flawed).

The smart thing to do would be to leave building custom printer profiles to the experts. With a custom printer profile costing less than $20 it seems crazy to spend big amounts on highly specialised kit that won't get used often and you haven't got the expertise to use effectively.

Perhaps. I'm curious where one can get quality profiles for $20. There was "Cathy's Profiles" for $25 some time back, but that website has been closed for a while, and complaints about no profiles and no refunds. I've come across "specials" from a Photoshop book with a code to get a profile made for $75. Maybe I didn't look hard enough?

Again, thanks for your patient assistance to clear this up.

[Rhossydd]

Careless on my part? Or the semi-random nature of the issue, so it was bad luck that it wasn't obvious?

I don't think there's anything 'semi-random' about this, is easily repeatable.

Perhaps. I'm curious where one can get quality profiles for $20.

I run http://www.colourprofiles.com and we charge £15 per profile, fourth free after you buy three (so about $17 if you buy 4). There are also at least half a dozen other businesses in the UK offering very similar or cheaper deals. From what we've seen, postage doesn't have to add a lot to the bill, depending on how urgently you need the results.
It doesn't matter to us what the post mark on the envelope of targets says, we've emailed out profiles across the world for nearly ten years now.

[l_d_allan]

I run http://www.colourprofiles.com

I'll contact you via your www.colourprofiles.com website, and see what we can work out. I'd definitely be interested, especially if your company could provide the CGATS .txt file(s) for one or several profiles. That might allow me to compare patch Lab values on my iSis to your equipment in addition to the profile itself.

[l_d_allan]

2. Until another device is added to the testing matrix, we don't know what causes the problems unless the measured data is used elsewhere.

Just thought I'd mention ... I was able to acquire the book "Color Management for Photographers", and it cleared up a number of questions I had.

That book had something of a Mac owners' perspective, and I believe my iSis / PMP problem was specific to the PC. But other issues aren't as fuzzy to me, now, after reading the book.

[Schewe]

That book had something of a Mac owners' perspective, and I believe my iSis / PMP problem was specific to the PC. But other issues aren't as fuzzy to me, now, after reading the book.

Yep, Andrew answers a lot of the workflow questions...but while the specifics of location and use, the ICC workflow is cross-platform once you get past a few issues. Personally, I think most of your problems stem from your hardware...and your tendency to jump all about in an almost spontaneous manner to try to resolve issues that ultimately are hardware related. If the instrument you are using is faulty and incapable of accurate and consistent readings, everything else you are doing is an utter waste of time. Until you can confirm that the iSis is actually working within spec, everything else is useless...

[l_d_allan]

One option is to just upgrade to i1Profiler which doesn't suffer from this problem. The quality of the profiles alone may make you wonder why you didn't switch earlier. PMP5 wasn't a great print profiling solution 5+ years ago and it's becoming increasingly problematic to try and keep using it today.

PMP5 delivers very good profiles, yes i1Profiler delivers better results, but many people wouldn't notice much difference between the two ...

I have to agree. While I think i1P's new engine is better ...

So ... what are criteria for evaluating printer profiles? Are there one or several objective measurements to get a quantifiable "figure of merit" from alternatives such as PMP5, Monaco, i1Profiler, ColorMunki bundle, ArgyllCms, etc. ?

"Inquiring minds want to know" ... perhaps I'll start another thread to get input ... (but has such a thread already been thrashed to death?)