Pages: 1 [2] 3 4   Go Down

Author Topic: Comparing spectros  (Read 18056 times)

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Comparing spectros
« Reply #20 on: June 17, 2017, 11:51:26 am »

Well that might be your opinion, but his question isn't answered; "which one is the more accurate"
drop below 1dE it's time to stop worrying about it and get on with making pictures. The CM tech will be good enough.

Yes, he wants to know which of his options is more accurate, so that's a "comparative" matter and the advice I gave him is very workable for his purpose without spending a bunch of money. That will be it from me on methodology, unless the OP wants any operational advice.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Rhossydd

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3369
    • http://www.paulholman.com
Re: Comparing spectros
« Reply #21 on: June 17, 2017, 12:35:46 pm »

Yes, he wants to know which of his options is more accurate, so that's a "comparative" matter
Only in the sense of comparing to a known accurate reference.

If you have two rulers and one reads 306mm and one reads 309mm you can compare them as much as you want, but the only thing that will confirm which is closest to the correct distance is to compare one to a known accurate distance.

Yes, he's going to need access to or buy a reference. There's no way to avoid it if he wants to know which is more accurate.
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Comparing spectros
« Reply #22 on: June 17, 2017, 01:19:32 pm »

Not necessarily, there are different workable approaches; but I'll leave that for you to figure out; I'm just plain out of time; sorry.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Comparing spectros
« Reply #23 on: June 17, 2017, 01:20:40 pm »

Only in the sense of comparing to a known accurate reference.

If you have two rulers and one reads 306mm and one reads 309mm you can compare them as much as you want, but the only thing that will confirm which is closest to the correct distance is to compare one to a known accurate distance.

Yes, he's going to need access to or buy a reference. There's no way to avoid it if he wants to know which is more accurate.

I actually had the same question come up when trying to, "accurately," measure printed ink distances. In addition to the mechanical issues of print head positional accuracy and repeatability there is the variation in paper since dimensions can change slightly over time, humidity, etc.

So how accurate were my steel tapes? And how consistent?

I had a SRX1 (surveyor's total station) which can measure angles accurately to about 1 second of arc but can only measure distance accurately to about 1.5mm.

By trading off smaller angles with increased distance one can find a point where an optimal accuracy is achieved. At 15m distance the angular and distance accuracy come out to about 0.1mm expected error across a 1m tape.

I found some steel tapes had cyclical variation of about .5mm but most were surprisingly good. I took the best one, which had errors of less than about 0.2mm, and use that with a loupe to measure printed ink distance.

I haven't found anything similar that would provide a reference standard for determining color measurement accuracy. There is considerable variation amongst the 3 I1's I have in the more saturated colors. But that variation decreases as colors desaturate. All are in reasonable agreement in or near neutral colors.

On another related point, X-rite, when it swallowed GMB, added some small changes in how the GMB (I1, etc) instruments read color so the company's product lines would more closely agree with each other. XRGA I think it's called. I have found essentially no documentation on the algorithms of this change but it involved differences around 2 to 3 dE on some colors.
« Last Edit: June 17, 2017, 01:24:19 pm by Doug Gray »
Logged

aaron125

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 55
Re: Comparing spectros
« Reply #24 on: June 17, 2017, 03:32:16 pm »

If you search hard enough, you might find X-Rite's white paper about their XRGA standard. I remember reading it quite some years back when they bought GMB.


Sent from my iPhone using Tapatalk
Logged

Rhossydd

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3369
    • http://www.paulholman.com
Re: Comparing spectros
« Reply #25 on: June 17, 2017, 03:50:42 pm »

I haven't found anything similar that would provide a reference standard for determining color measurement accuracy.
Last time I looked into this Hutchcolor offered the most credible affordable reference targets. I think Wolf Faust also offered similarly individually measured and referenced targets.
At more affordable levels the Colorchecker SG is probably the best option.
The CC24 seems to have gone through so many slight variations in manufacture process that getting the most accurate reference data is trickier, but is still a good starting point for amateur investigations.
Logged

Jeff-Grant

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 568
    • http://www.jeff-grant.com
Re: Comparing spectros
« Reply #26 on: June 18, 2017, 12:08:36 am »

The targets that you mention all appear to be scanner targets. What I need is a DTP70 formatted target that I can also read in spot mode with the I1p2. I'll attach a grab of a DTP70 target.
Logged
Cheers,
 Jeff  www.jeff-grant.com

Rhossydd

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3369
    • http://www.paulholman.com
Re: Comparing spectros
« Reply #27 on: June 18, 2017, 02:43:14 am »

The targets that you mention all appear to be scanner targets.
No, they're just reference charts. You can use them for scanner profiling, plus cameras and other systems.
The point is that they can be used to ascertain the accuracy of the handheld spectro and once you've an idea of limits of that, you can then measure your own autospectro charts to see how they compare.
It's not perfect, but the most affordable way to get an idea of what each instrument is capable of.

Quote
What I need is a DTP70 formatted target that I can also read in spot mode with the I1p2.
I've never found a DTP70 reference chart for sale anywhere. I'd assume that only X-Rite would have any. They might sell you one, but I would have thought it extremely unlikely that they still would have any they would be prepared to sell and if they did it would be uneconomically expensive.
Logged

Jeff-Grant

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 568
    • http://www.jeff-grant.com
Re: Comparing spectros
« Reply #28 on: June 18, 2017, 02:50:33 am »

Got it. Validate the accuracy of the i1p2 first and go from there. Here's a link to the DTP compatible chart. Unfortunately, it's $500 as it includes a subscription fee: https://chromachecker.com/info/en/page/A2

Do you have a recommendation of a chart to select?
Logged
Cheers,
 Jeff  www.jeff-grant.com

Rhossydd

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3369
    • http://www.paulholman.com
Re: Comparing spectros
« Reply #29 on: June 18, 2017, 03:14:13 am »

Here's a link to the DTP compatible chart. Unfortunately, it's $500 as it includes a subscription fee:
'on-line shop not available' Interesting, but the information shown suggests that chart might be $500 additionally to the $4k annual subscription.
It might be worth emailing them to see if they'd sell a chart subscription free with no support given. I'd be interested to know what they reply.

Quote
Do you have a recommendation of a chart to select?
Whatever you have available ? A CC24 comes with i1Profiler, colorchecker SG came with PMP5.
If you have neither, a CC24 or X-Rite passport would be the one of the easiest ways to get a reference. Both could also be used for camera profiling too.
X-Rite also supplied a reference target for scanner profiling with i1Match in the past. You could get involved with some reference file hacking and DIY to use it as a reference for the DTP70, but that would be quite challenging.
Logged

GWGill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 608
  • Author of ArgyllCMS & ArgyllPRO ColorMeter
    • ArgyllCMS
Re: Comparing spectros
« Reply #30 on: June 18, 2017, 09:32:34 am »

Ethan - even if only 10 or 20 of those samples are usable, it's still enough to make a useful average for the purposes of eliminating undue impact from outliers.
Number of samples doesn't mean much if each sample has a poor signal to noise ratio. Total light collected is typically what counts. What can improve S/N is slowing your scan down, but in terms of repeatability, the incandescent lamp in the i1Pro is its main limitation. Thermal effects cause noticeable reading to reading variation. In contrast, the ColorMunki spectro has much better repeatability, comparable or even superior to the SpectroScan.
Logged

GWGill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 608
  • Author of ArgyllCMS & ArgyllPRO ColorMeter
    • ArgyllCMS
Re: Comparing spectros
« Reply #31 on: June 18, 2017, 09:40:15 am »

XRGA I think it's called. I have found essentially no documentation on the algorithms of this change but it involved differences around 2 to 3 dE on some colors.
Persusing the xrga_equations table in spectro/xrga.c in the ArgyllCMS source code should provide all the technical information you are after.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Comparing spectros
« Reply #32 on: June 18, 2017, 10:20:58 am »

Persusing the xrga_equations table in spectro/xrga.c in the ArgyllCMS source code should provide all the technical information you are after.
Perfect!  No idea why I didn't think to look there instead of searching for pdfs.

Thanks.
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Comparing spectros
« Reply #33 on: June 18, 2017, 10:29:03 am »

Number of samples doesn't mean much if each sample has a poor signal to noise ratio. Total light collected is typically what counts. What can improve S/N is slowing your scan down, but in terms of repeatability, the incandescent lamp in the i1Pro is its main limitation. Thermal effects cause noticeable reading to reading variation. In contrast, the ColorMunki spectro has much better repeatability, comparable or even superior to the SpectroScan.

Yes, I've tested for reading to reading variation with my i1Pro2 under two conditions: (1) re-reading a patch without moving the spectro at all, and (2) taking numerous readings from different parts of the patch (in this case patches being at least a square inch large so one can move around in them such that each reading is unique. So I know where this matter stands and variances are there but livable. Approach (2) of course doesn't totally isolate the spectro because it is possible that ink laydown isn't totally uniform accross the patch, and indeed, the variances from (1) tend to be much lower than those from (2). I made such measurements on medium to high density patches, as well as on paper white. The outcome of all this was that with approach (1) dE(76) outcomes were within a narrow range of several 10ths of a dE, whereas with approach (2) the dE outcomes had a wider range but none greater than 1.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Rhossydd

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3369
    • http://www.paulholman.com
Re: Comparing spectros
« Reply #34 on: June 18, 2017, 12:02:42 pm »

with approach (2) the dE outcomes had a wider range but none greater than 1.
This is the sort of result (I've seen the same with testing here) that tells us that a degree of pragmatism is needed when idealising about which instrument is better.
If printers have a variability significantly greater than the instruments that measure their output, it's not worth getting too obsessed about a small amount of variability of readings.
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Comparing spectros
« Reply #35 on: June 18, 2017, 12:24:51 pm »

This is the sort of result (I've seen the same with testing here) that tells us that a degree of pragmatism is needed when idealising about which instrument is better.
If printers have a variability significantly greater than the instruments that measure their output, it's not worth getting too obsessed about a small amount of variability of readings.

Totally agree.

And it gets to be even more fun when you look at prints made with different printers, different profiles, but the same paper and have a hard time seeing the difference between them - because your colour management set-up is good enough to do what it is supposed to be doing.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Comparing spectros
« Reply #36 on: June 18, 2017, 01:40:47 pm »

Totally agree.

And it gets to be even more fun when you look at prints made with different printers, different profiles, but the same paper and have a hard time seeing the difference between them - because your colour management set-up is good enough to do what it is supposed to be doing.

This is the sort of thing I simply expect from good color management. When it doesn't happen it's almost always because I did something wrong in the underlying color management process. At least apart from significant BP variation and that is usually minimal with the same paper between different printers.
Logged

GWGill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 608
  • Author of ArgyllCMS & ArgyllPRO ColorMeter
    • ArgyllCMS
Re: Comparing spectros
« Reply #37 on: June 18, 2017, 09:41:07 pm »

Last time I looked into this Hutchcolor offered the most credible affordable reference targets. I think Wolf Faust also offered similarly individually measured and referenced targets.
At more affordable levels the Colorchecker SG is probably the best option.
The CC24 seems to have gone through so many slight variations in manufacture process that getting the most accurate reference data is trickier, but is still a good starting point for amateur investigations.
None of these charts strike me as being suitable for checking reflection instrument accuracy. None of them typically come with individually measured reference files measured using a reference grade instrument.
Logged

Jeff-Grant

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 568
    • http://www.jeff-grant.com
Re: Comparing spectros
« Reply #38 on: June 19, 2017, 12:51:22 am »

I've just heard back from Xrite. The DTP 70 is well and truly discontinued so I guess that is the end of this little exercise, unless someone out there is repairing them. I did the exercise of testing my new i1p2 on the ColorChecker. I'll attach the results.

Thanks for all the help .
Logged
Cheers,
 Jeff  www.jeff-grant.com

Rhossydd

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3369
    • http://www.paulholman.com
Re: Comparing spectros
« Reply #39 on: June 19, 2017, 01:07:06 am »

None of these charts strike me as being suitable for checking reflection instrument accuracy. None of them typically come with individually measured reference files measured using a reference grade instrument.
The key word was "affordable". Individually measured targets are VERY expensive.
Whilst not perfect, the charts mentioned are probably getting as close to spec. as is needed for the OP's purpose.
Logged
Pages: 1 [2] 3 4   Go Up