Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Comparing the HP Z3200ps on-board profiling vs Spectrolino  (Read 1310 times)
teddillard
Sr. Member
****
Offline Offline

Posts: 664


WWW
« on: May 15, 2010, 05:38:00 AM »
ReplyReply

...just started a new gig at Parrot Digigraphic, and I'm getting to play, er, work with some fun new stuff.  I'm pretty quickly developing a good deal of respect for the HP Z-series printers, although it's a bit of a learning curve, as well as addressing some of my prejudices.  

For one, I was interested in just how good the onboard i1 profiling is, and I also wanted to play (for my first time) with the Spectrolino that was sitting in the corner...  so I snagged a profile from the printer, then build one with the same paper on the Spectrolino.  The results were almost identical profiles- a fast and dirty comparison, I know, but still pretty impressive performance, especially for the money.  

I posted the video here- http://www.youtube.com/watch?v=lioBBT2gc6w
Logged

Ted Dillard
terrywyse
Jr. Member
**
Offline Offline

Posts: 90



« Reply #1 on: May 15, 2010, 04:18:20 PM »
ReplyReply

The spectro embedded in the HP "z" series printers is essentially an iSis. I got into a mild argument the other day with a colleague that claims it's EXACTLY like an EyeOne under the hood (it's not). Of course, they call it "i1 Color by XRite" or whatever but it's more like an iSis than an EyeOne, with the LED illumination being the key element tying it to the iSis. Because of this, it's natural behavior is that of a UVcut-filtered spectro since the LED illumination gives off no UV. As a result, I assume you'd find fairly big differences between it and an unfiltered Spectrolino, especially with certain papers. The more OBAs present in a media, the larger the difference.

the HP "z" printers maybe aren't the hot ticket for fine art printing compared to Epsons, especially the x900 Epsons, but they make fine proofing printers when driven with a RIP (GMG, EFI XF, etc.) that supports the embedded spectro for fully automated calibration. I think they're nice machines, all things considered, and find them to have more consistent behavior than most Epson printers I've used.

Regards,
Terry
Logged

Terry Wyse, WyseConsul
day job...Color Management Consulting
on the side....photoWyse, photography and fine prints
G7 Certified Expert (but that depends on the day)
terrywyse
Jr. Member
**
Offline Offline

Posts: 90



« Reply #2 on: May 15, 2010, 04:59:32 PM »
ReplyReply

Hate to be a nit-picker here but around 2:20 into the video, the video states something to the effect that the Spectrolino is measuring the profile chart and comparing the "measured values to the actual values" via a LUT. In this context, this is incorrect. When measuring a chart to create a profile, there's no expectation of what the values SHOULD be, only what the values ARE at the time you measure the chart.

Using "measured values vs. expected values" would be more appropriate in the context of an ICC transform/conversion between a source profile and a destination profile; the source profile would represent the "expected" values of the ICC conversion while measurements taken from a print of this ICC conversion would be the "actual" values that resulted from the conversion. The degree to which the measured values in the print agree with what was expected from the source profile would be the level of accuracy you've attained. Naturally, to have any expectation of accuracy the first requirement would be that the destination (printer/output) profile would have to be able to fully "contain" the source profile. In an RGB workflow, this is typically not the case.

Also, simply overlaying the 3D gamut volumes of the two profiles is not a good predictor of accuracy. It only tells you that the two spectros captured roughly the same gamut volume, which is what you'd expect, but it tells you nothing of the accuracy or agreement between the actual grid points in the two profiles. "Inside" the two profiles, there could be massive disagreements between specific data or grid points.

If this is about comparing the two spectrophotometers, the only real test would be to take the measurements generated from the HP's internal spectro and compare those measurements *from the same exact chart* as measured by the Spectrolino. The requirement to measure from the same printed chart would be damn difficult given the different chart layout requirements for each spectro.

I think I understand what you were trying to accomplish here but the methodology used and the understanding of what goes into making a profile is flawed in my opinion. If I simply missed what you were trying to do, then please straighten me out.

Regards,
Terry
Logged

Terry Wyse, WyseConsul
day job...Color Management Consulting
on the side....photoWyse, photography and fine prints
G7 Certified Expert (but that depends on the day)
teddillard
Sr. Member
****
Offline Offline

Posts: 664


WWW
« Reply #3 on: May 16, 2010, 05:48:53 AM »
ReplyReply

Thanks Terry- I think you're making valid points here...  nit-picking is certainly welcome...  the goal here, though, was to see if, in fact, the two devices were actually able to read the same gamut from the target.  In comparisons I've done in the past, in particular the original version Color Munki, I got substantially different and limited gamuts from the Munki.  Capturing the same volume isn't what I'd expect, necessarily, although it's what I'd hope for.

But it clearly is simply a starting point in looking at how the two systems work.  The interesting thing to me is that they can bundle this onboard system into the printers without adding a huge price tag, and that the thing works so well.
Logged

Ted Dillard
Mussi_Spectraflow
Full Member
***
Offline Offline

Posts: 171


WWW
« Reply #4 on: June 18, 2010, 02:11:38 PM »
ReplyReply

It's an interesting test, but seems to be more of test of the profiling software than the spectro's. I did some tests with the Z3100 when it was released and found the spectro to be a pretty decent device, it is essentially a modified Rev D i1. The repeatability is on par with that of the i1. With fluorescent measurements though the LED(with no UV) source is probably going to produce slightly different results from the Spectrolino(with a UV cut tungsten) source. Also I'll take a bet that it's been a while since your spectrolino was calibrated  I know the 530 sitting on my desk is more than a bit due for calibration.

So comparing profiling packages. Well I talked to some of the guys in Barcelona and they explained to me that the profiling software under the hood is basically the Monaco package optimized to the gamut of the Z. Since they figure that the profiles built on the printer are essentially only going to be for this printer(yes I'm ignoring the APS option) they were able to hard code in a few variables and get good looking profiles with fewer patches. I'll also second Terry's point about gamut volumes....gamut volumes are an awful metric, since there is no standardized way to calculate them, and often times you'll find that the colorimetric vs perceptual rendering will vary more with one package and printer than the other.

I thinks it's great that you posted this test, it certainly interesting stuff but the devil is in the details and unfortunately there are a lot of variables that make isolating and testing the spectro a tough task. There used to be a utility that let you export the measured patches from the chart, exporting that and then doing some spot measurements with the spectrolino from the same chart would give you a better idea of the device agreement. Good luck with your testing.
Logged

Julian Mussi

Spectraflow, Color Workflow Solutions
www.Spectraflow.com
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad