Averaging with i1Profiler

[Jeff-Grant]

While reacquainting myself with Profiler, I found the description of how to average measurements while building a profile. http://www.xrite.com/product_overview.aspx?ID=1397&Action=support&SupportID=5709

I tried it and got a normal looking profile. Having tried it again, I now get a profile that looks like someone has attacked it with a knife. I used three measurement files, which, by themselves, produce a normal looking profile.

I'm baffled. but that's no surprise with this stuff. I'm attaching a screenshot of the profiler display. Does X-rite have a support organisation for this stuff? I raised an issue a while back which got me a couple of replies indicating that what I wrote had not been read, and then silence.

[digitaldog]

Yeah, that's no good.

Something data wise is hosed big time. I don't know if it's a bug or whatever. I use another product (ColorAnt) or the old MeasureTool, they work.

[Jeff-Grant]

Thanks Andrew. I'm assuming that ColorAnt is expensive. They make no mention of price. Measure Tool won't work on Mavericks. Thank heavens I have an old Mini with Snow Leopard on it that I can use.

[digitaldog]

Color Ant is expensive but does a lot of cool things <g>.

Just get MeasureTool running on an old system. It is just so much easier to get some old MacBook on eBay, let it run 10.4 and you'll be able to also run Classic. I've got OS9 stuff from the really old days, like ColorBlind that run today on such an old beast. Get MeasureTool, it should work just fine as waiting on X-rite to fix i1P is a huge waste of time. Stuff we had years ago was better in many respects. You even have to options for averaging in MT depending on how much data you're using.

[Jeff-Grant]

Thanks Andrew, will do. It must be wonderful having a near monopoly.

I've gone back to 1.5.0 but I still get the same result. The set that initially produced a good profile now produces junk.

[digitaldog]

It must be wonderful having a near monopoly.

As long as you have customers.

[Jeff-Grant]

Are there choices? Isn't profiler the only game in town?

[digitaldog]

There's ColorLogic who make ColorAnt. And a few others but not much. If you own the old stuff, sometimes that's the way to go. Depends. I've built profiles from i1P, Copra, MonacoPROFILER and ProfileMaker Pro and depending on a lot of factors, no one was best.

[Jeff-Grant]

Thanks Andrew, good advice.

[Some Guy]

Something has gone kludgy for me since I updated to the newest Profiler and Colorport software.

Sometimes I press the button for a just a spot reading and the LEDs flash red?  Other times I press it on white and it shows up black in Colorport.  If I recalibrate again it acts normal - until I unplug it and try later.

Very odd why it is happening (Windows 8.1, i1 PhotoPro 2) and seem worse since the May 2014 updates.

However, I've never seen wireframe as bad as yours.  Looks like you did a painted rock or something.  Have you ran the head through their "i1 Diagnostics 4" software?  I did it on mine and got a clean bill of health, plus it worked for a while too.  Next day, red LED's on spot readings right off the calibration tile.  I dunno.

SG

[digitaldog]

However, I've never seen wireframe as bad as yours.

It's not 'real' data. The averaging screwed up the numbers in multiple places and the 3D map shows this. It's kind of useful to see that: oh, this profile is absolute garbage. Of course you could soft proof and see really awful results too. Or convert the data.

Take one of your own data files, open it in Excel, move some numbers around and build a profile. You'll be able to produce a similar profile depending on the mixing of values. 

[Jeff-Grant]

FWIW, I have now gone back to 1.5.0 and the averaging problem remains. What's weird is that it worked once. I suppose that I shouldn't hold my breath waiting for a fix.

[Rhossydd]

FWIW, I have now gone back to 1.5.0 and the averaging problem remains. What's weird is that it worked once.

Any idea which version it worked on ? I think your post may be the first time I've read that it worked for anyone at all.

[Jeff-Grant]

It only worked once for me, and then started producing junk. I just tried using ProfileMaker 4 to get an averaged file. It doesn't work either. Is there any way that an averaged file can be used? I'm assuming that averaging is still a good idea. I have downloaded ColorAnt which can also average. I can't try it till I get the trial licence code.

[Rhossydd]

It only worked once for me, and then started producing junk.

So maybe you just got lucky once ?

I just tried using ProfileMaker 4 to get an averaged file. It doesn't work either.

As far as I know PMP's averaging function works(ed) fine. However not every PMP licence unlocked that feature (mine doesn't).

I'm assuming that averaging is still a good idea.

I guess this depends on the reliability(repeatability) of the spectro. I'd also guess that some substrates like textured art papers may benefit more.

The bottom line is to try your trial of Color ant and make some profiles, some with averaged data, some without and see if you can see any difference. Then decide if the cost is worth any benefit.

On a zero budget it's possible to average data with a spread sheet(Excel). It's all just asci text data and cut 'n paste can be your friend.

[Jeff-Grant]

Sorry, what I meant is that I averaged three files in PM4 and then tried to build a profile in i1Publish with the averaged data. The result is junk like i1P averaged files. I suspect that the ColorAnt files will be the same.

I downloaded both PM4 and PM5. My i1Pro is licensed for Publish but wouldn't let me average in PM5. PM4 will let you average without a spectro. I don't know how repeatable the i1Pro is on an io. There always seem to be some differences but I just don't know whether I should trust a single pass so I do three. My threshold of pain would kick in with Excel. I would pay someone before I went that way.

[Rhossydd]

then tried to build a profile in i1Publish with the averaged data. The result is junk like i1P averaged files

Have a look into the *.txt files and see if there's anything obviously different with an averaged data file. You might find it's as simple and daft as the wrong text format or an extra line of metadata that Publish isn't happy with.

I don't know how repeatable the i1Pro is on an io.

Easy enough to find out, just do several measurements then compare the sets and evaluate the results.
If you are routinely making three sets of measurements you'll already have the data.

My threshold of pain would kick in with Excel. I would pay someone before I went that way.

YMMV, it's pretty simple maths by Excel standards.
It all comes down to which you have more of, time or money.

[Jeff-Grant]

Thanks. I'll take a look at the files, and see is there is anything obvious. I'll also look at the measurement files. I suppose that it would be too much to ask for Xrite to say what it should be. I do remember, after buying an i1Pro, being sneered at by an Isis user and told that it was crap. If he was right, it was very expensive crap.

I have a number of Excel spreadsheets from my days with ColorBurst. I'll dust them off and take a look.

I haven't been near any of this stuff in a few years. It doesn't seem to have been made any simpler in that time.

[Rhossydd]

I suppose that it would be too much to ask for Xrite to say what it should be.

Just use a good standard data set as reference for comparison.

I do remember, after buying an i1Pro, being sneered at by an Isis user and told that it was crap. If he was right, it was very expensive crap.

The i1Pro is a fine and versatile instrument. The Isis may be slightly more accurate for some things, but there's not a lot of difference in the end result, if any.

I have a number of Excel spreadsheets from my days with ColorBurst. I'll dust them off and take a look.

Just start from scratch again. Trying to re-use old spread sheets may be error prone unless you documented them well.

I haven't been near any of this stuff in a few years. It doesn't seem to have been made any simpler in that time.

I don't think there's been much improvement in the last ten years. Profiler might build slightly better profiles and have a flashier interface, but as Andrew has already said there's not much between it and PMP5, or anything else of that vintage.
Documentation is still dire.

The one thing that has improved is manufacturer's profiles now seem better and some third party materials suppliers seem to be making progress in providing quality profiles for their products too.

[Rhossydd]

Having piqued my interest, I thought I'd see how hard it is to average data with Excel. Averaging two data sets took less than eight minutes from scratch.
The maths are as hard as =(C1+N1)/2  <you might want to work with more data sets, but this was just with two>
auto fill makes propagation of the calculation on the sheet dead simple, format the sheet to two decimal places, copy and paste the result into one of the original data sets and save it with a different name..... job done.
i1Profiler has no problems parsing it and a decent profile comes out of the end.

OK it'll take longer if you have to copy 'n paste more and/or bigger data sets, but it really is dead simple for anyone that has a passing knowledge of spreadsheets and has a clue about what to look for in the measurement files. It's all pretty obvious.