Re: Epson canned profiles
Re: Epson canned profiles
- Subject: Re: Epson canned profiles
- From: Andrew Rodney <email@hidden>
- Date: Mon, 16 Apr 2018 12:18:55 -0600
On Apr 16, 2018, at 12:03 PM, ben <email@hidden> wrote:
>
> On Apr 16, 2018, at 10:18 AM, Andrew Rodney <email@hidden> wrote:
>
>> THREE different settings in the print driver to make 918 color patches.
>
> Just to make sure I'm understanding your workflow...
You don't. At all!
> you have a 306-patch chart that you print three different ways?
918 patches per target. Three different print parameters changed in the Epson
driver as outlined for English speakers, between the two print samples.
Measured on an iSis XL and not totally dried down. The differences (Avg dE)
between the two targets are INVISIBLE. Hence, no problem as you suspect, assume
and are incorrect in stating.
> And then you read each chart independently, merge all samples into a single
> results file with 918 samples, and generate your profile from that?
Have you ever measured a target for a profile? Ever saved CGATs data and used
software to compare the dE of the measurements? Seems not.
> (Seems weird and highly questionable without some solid colorimetric data to
> back up such an approach...so either you've got that sort of data or I'm
> misunderstanding your workflow or your workflow is invalid.)
Because you're so confused!
> 5000 patches, the average dE was 0.5!
>
> Yes, that's maybe a tenth or two of a dE better than what Argyll generates on
> an iPF 8100 with similar-sized charts.
You are confused again. The difference (distance actually) between 5000 color
patches, the average from two of the same Epson printers is 0.5 dE 2000.
Invisible.
> But you're still not even pretending to answer some really, really basic
> questions.
You're asking silly questions and I refuse to answer with silly answers.
> You've told us how tight a fit your profiles are to the measurements you've
> made, which is probably good.
No I did not. You're confused a lot. This has ZXERO to do with ICC profiles.
There's no reason to introduce that into the mix. I'm telling you the measured
differences in either 918 or 5000 color patches from printers that are
measured.
> (It's possible to over-fit to the detriment of profile quality...probably not
> a problem with a thousand-patch chart on a modern printer, but a decade ago I
> would very much have suspected over-fitting with the numbers you supplied.
> And, if you really are only sampling a few hundred locations in the printer's
> color space, I'd definitely start to suspect over-fitting -- especially since
> the triple redundancy may well trick the profiling engine into
> over-emphasizing those locations.)
Go onto the topic and don't go into areas that will only confuse the discussion.
>
> But I'm asking a trivial question completely unrelated to profile fit.
This has NOTHING to do with profiles Ben. It has to do with device behavior
before profiles are built. Data that is measured to create a profile.
> What, exactly, in D50 L*a*b*, is the darkest color the printer generates with
> the most economical setting? And what does it generate with the highest
> quality setting?
Doesn't matter. You're asking about ONE color and I'm provding data about 918
or 5000! It appears you have no idea how to read a dE report, what the average
of thousand of meaured colors tell us about color differences (distance) of two
sets of samples.
> You've got your just-printed chart there in front of you, right? And you just
> measured it, no?
Yes. How else do you think I made the dE report in ColorThink Pro? Unlike you,
I'm actually measuring data on an iSis XL and comparing actual colorimetric
data.
You're making stuff up.
> Unless you're doing something utterly bizarre, at least one of the patches on
> that chart will be R=G=B=0.
I'm not doing any bizarre utterly or otherwise. What's bizarre is that you'd
come onto this list with such a poor understanding of colorimetry.
> All you have to do to answer this very simple question from me is find that
> patch in your measurements, once for the one setting and another for the
> other setting, and report what the measured D50 L*a*b* (or XYZ or other
> absolute coordinate system) values are.
I answers 917 and 4999 questions of colorimetric data.
> Is it really so hard for you to understand what I'm asking for?
Is it really so hard for you to understand how we measure color and produce dE
reports of output?
> Really so hard to answer it? Really so hard to understand why it might
> matter? Really so hard to believe that, whatever Epson is doing today, it's
> been a source of challenge in the past?
Is it so hard for you to understand that what you state is rubbish and unproven
colorimetrically? This from a fellow with NO experience with Epson printers.
Sad.
> I mean, it seems like not that long ago that forum posts were about nothing
> but printer dMAX this, dMAX that -- obsessing over which made the darkest
> blacks, which ones were problematic because the lowest L* colors they could
> print were so far off the neutral axis, and so on. Sure, technology has
> improved dramatically...but do you really expect us to take your perfectly
> unremarkable error fit report as evidence that Epson's dMAX is the same
> regardless of driver settings?
Posts you appear not to have understood.
> ...er...you _do_ know what that error report is, right? You supply your
> profiling engine with the RGB values you printed and the L*a*b* (etc.) values
> you measured from the print. The engine builds an ICC profile that maps
> requested L*a*b* values to RGB values to send to the printer. Your report is
> the result of taking your measured L*a*b* values, running them through the
> profile, seeing what RGB values the profile generates, and comparing them
> with the actual RGB values you sent to the printer. In other words, all
> you've done is tell us how tightly your profiling engine was able to fit the
> profile to the measurements; you haven't told us anything about the
> measurements themselves....
Bla, bla bla. The more you write assumptions and FUD, about the behavior of
Epson printers, the more there is no doubt:
It is better to say nothing and be thought a fool than to open your mouth and
remove all doubt. - Abraham Lincoln
Andrew Rodney
http://www.digitaldog.net/
_______________________________________________
Do not post admin requests to the list. They will be ignored.
colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden