Newbie just getting into printer profiling

[DenverESullivan]

I'm learning as I go so if this seems like a stupid question I apologize in advance...

I recently expanded my photography interests and splurged on some new equipment that I'm just starting to learn how to use.  I ended up getting a Canon Pixma Pro-1 and the Xrite i1Publish Pro 2 system.

I'm now at the point where I want to profile my new printer to get the best performance out of it.  Unfortunately, I can't seem to get any definitive answers from anyone familiar with what I'm wanting to do, including Xrite.  I've read that over sampling with patches is detrimental to the process.  However, no where can I find at what level that becomes the case.

I'm going to be printing my targets on 8.5x11.  What would those of you who are currently using i1Profiler recommend??

I'm basically wanting the best quality I can get out of the printer without unnecessarily making files unmanageable.

[howardm]

As an aside, you didn't need the Publish version if you're not going to be profiling CMYK printers (the Canon & Epsons are CMYK but using their driver, they 'appear' to be a basic RGB printer).  So, if you haven't opened it yet, you might want to consider trading-down and saving some $$.

That said, I use 1215 patches which is 2 full sheets and extra near-grey patches.  The patch generator goes in-n-out of giving you extra greys depending on the # of patches.  1215 works for me.  Remember that you can then optimize that profile w/ another sheet if you want so the results can be better than blindly doing 1800+ (3 sheets).

Generally, 2 sheets is boring enough to try to manually scan

[PeterAit]

Profileth thee thou not thine printer. It is a wasteth of thy time. So sayeth - well, me.

Seriously - the canned profiles are fine - not perfect always, but good enough.

Go take photos. Make prints. Evaluate them, show them to people, get comments. Fussing with printer profiling is, IMO, just another form of mental masturbation. Let it go, and do photography!

[DenverESullivan]

Howardm,

Thanks for the response.  The reason I went the Publish Pro route is because I also have a color laser that is CMYK.  My local dealer gave me a fantastic deal and it cost me literally $100 more. Figured being able to do both was well worth the difference.

Do you find 1215 patches to provide better than the canned profiles?  It just seems from everything I've read that most folks seem to use more.  Is optimizing a lower number better than starting with a higher number?

I guess I was hoping there was some rule-of-thumb guideline out there that for x printer you use y patches.  Then folks warn that too many patches degrade quality but no where can I found what that 'magic' number is.

[Jim Kasson]

I'm learning as I go so if this seems like a stupid question I apologize in advance...

I recently expanded my photography interests and splurged on some new equipment that I'm just starting to learn how to use.  I ended up getting a Canon Pixma Pro-1 and the Xrite i1Publish Pro 2 system.

I'm now at the point where I want to profile my new printer to get the best performance out of it.  Unfortunately, I can't seem to get any definitive answers from anyone familiar with what I'm wanting to do, including Xrite.  I've read that over sampling with patches is detrimental to the process.  However, no where can I find at what level that becomes the case.

I'm going to be printing my targets on 8.5x11.  What would those of you who are currently using i1Profiler recommend??

I'm basically wanting the best quality I can get out of the printer without unnecessarily making files unmanageable.

A long, long time ago when I was a color scientist, profiling printers used tomake me smile...

Oh, sorry, the 40th anniversary of American Pie is doing funny things to my head.

Where were we? Anyway, when I was writing software to do this, you couldn't get too many patches. Yes, there was uncertainty in the data, but you smoothed all that out before you inverted the dataset. In fact, we'd read the same patch set several times and use all those readings.

When I profile a printer/paper set, I generally use about 2500 patches, but I'm reading them with an iSis (Boy, I'll bet Xrite wants that name back), so there's no manual scanning penalty.

I find my own profiles preferable to the vendors' profiles, but I like to be able to tweak things, and sometimes I've been known to use weird papers.

Jim

[DenverESullivan]

Jim,

Thanks for the response.  I do imagine Xrite wishes they'd never seen the name Isis... lol.

Does using that number of patches make the profile unnecessarily large?  Being new to this I'd hate to 'oversample' and then end up with something too large to deal with.  Xrite even warns of going too large but no where have I found where the line is drawn.

[Jim Kasson]

Does using that number of patches make the profile unnecessarily large? 

I see no reason why the number of patches used to create a profile should have anything to do with the size of the profile. The patches are used to populate 3D lookup tables (LUT) that are used for interpolation. Maybe the size of the metadata in the profile is affected. The things that mainly determine the size of the profile are the sampling density (8x8x8 vs 16x16x16, vs 32x32x32, say), and the precision of the entries (8 vs 16 bits, say).

That's assuming there's no data compression before the profile is stored. I don't know if that's currently done or not.

Jim

[howardm]

on some level, I think this is an excellent educational opportunity.

Try it yourself and see !   

Get a pack of somewhat decent paper and try  2, 3, 4 sheet profiles and maybe a 2 or 3 sheet optimized.
Can you see the difference by comparing something Andrews Printer Gamut File?  If so, is it 'worth it' given extra paper and time?

Bottom line is that the line you seek shifts like sand and isn't a hard number.  The Colormunki does a
respectable job w/ only 100 patches

[DenverESullivan]

Jim,

I kindof agree with your thinking that patches shouldn't make a profile unmanageable.  However, given the number of references I've seen that appears to be the case... or at least the perception.

HowardM,

I agree that this is definitely going to be a learning experience.  I'm just trying to get to where I'm starting from a reasonable position and hoping not to waste a ton of paper and ink unnecessarily.  Truth be told, I'm really only planning on using 4-5 different papers.  Figured it's better to go for a few high quality choices and only expand from that if I encounter something completely new or different.  No sense in having multiple choices that are essentially pretty much the same just from different vendors.

Any other tips anyone want to share on profiling before I get too far into it?

[Jeff-Grant]

If you work you way through this, on page 2, Andrew Rodney comes up with a number: http://forum.luminous-landscape.com/index.php?topic=99611.20

[Wayne Fox]

Profileth thee thou not thine printer. It is a wasteth of thy time. So sayeth - well, me.

Seriously - the canned profiles are fine - not perfect always, but good enough.

Go take photos. Make prints. Evaluate them, show them to people, get comments. Fussing with printer profiling is, IMO, just another form of mental masturbation. Let it go, and do photography!

Yeah, sort of standard reply for a photographer doing anything technical.

But profiling a printer/paper combination takes a relatively small amount of time and effort.  Canned profiles can be pretty good, but some not so good.  It's not like taking 30-60 minutes to make a custom profile is going to interfere with shooting time. Making my own profiles has never cost me the opportunity to shoot something.

The print is the ultimate expression of your art/craft.  Seems like any small investment to ensure the quality would be worth while.  

[Rhossydd]

Fussing with printer profiling is, IMO, just another form of mental masturbation. Let it go, and do photography!

You're wrong. Getting printer profiling right is just a way of ensuring that digital printing is as easy as it can be and is delivering the best possible results. It's usually a simple 'once for each paper' task that allows you to stop worrying about printing and concentrate on the important aspects of image making.

I'm not sure that I would recommend spending nearly twice as much on a profiling system than the printer as the OP has done here. Getting a printer profile from a remote service will give the benefits of custom profiling without the huge investment in cash and time learning how to use it.

[howardm]

The OP wanted opinions and I gave him mine. 

You're free to give your own.

I still believe that fewer sheets w/ optimization yields better results than more straight-away sheets since the software
can actually analyze the space post-profile creation.

[Jim Kasson]

Jim,

I kind of agree with your thinking that patches shouldn't make a profile unmanageable.  However, given the number of references I've seen that appears to be the case... or at least the perception.

I didn't say anything about managing profiles, although, now that you've brought it up, I see no impact of patch set size there, either. What I did say (or tried to say, if I was unclear) was that

1) the size of the patch set should not adversely affect the accuracy of a well-designed profile generation algorithm, and

2) the size of the patch set should not materially affect the size of the completed profile.

Any other tips anyone want to share on profiling before I get too far into it?

Another poster has recommended an optimization stage after the first profile is computed. I concur with that recommendation. In fact, though I haven't done it in years, for really critical work you can do an optimization stage for just the colors in a specific image.

Jim

[hugowolf]

I don't know if an increased number of patches creates a larger profile, but I can't see why it would. Maybe it does if you embed CxF data in the profile?

In my limited experimentation, I have found two 8.5" x 11" sheets followed by one optimization sheet to produce more accurate profiles than four sheets with no optimization. And 1 +1 to be better than 2 originals.

The ColorMunki Photo does remarkably well given its meager 50 + 50 patches.

But at some point, a profile made from large number of initial patches will show little or no change after optimization. However, without an i1iO or i1iSis (which will not read some thicker papers), manually scanning a large patch set is more than a little tedious - especially with the machine/software instructing the human operator (oops, too fast, go back an do that row again, etc).

Keith Copper's review of the i1Photo Pro2, includes a tutorial run through, and also usefully shows the missing neutral ramp problem that can occur when trying to maximize the number of patches per sheet.

http://www.northlight-images.co.uk/reviews/profiling/i1_pro2_photo.html

Brian A

[Rhossydd]

I don't know if an increased number of patches creates a larger profile,

It does. From some tests I did recently;
1005 patches 2,463kb
2033 patches 2,799kb
3155 patches 3,164kb

From what I've seen just over 2k patches seems to be the sweet spot for performance.

[Jim Kasson]

It does. From some tests I did recently;
1005 patches 2,463kb
2033 patches 2,799kb
3155 patches 3,164kb

Interesting. I went back and looked at some of mine and found that the optimization adds to the size, in spite of what I said earlier. I guess it must store all the readings...



Jim

[Rhossydd]

Is profile size an issue anyway ? 1mb difference in size isn't going have a significant impact on storage now and I doubt that difference will have any impact on the speed of printing.

Sure, it's a pain to manually scan multi page targets, but as a rarely done task, is it really a problem ?

[digitaldog]

I see no reason why the number of patches used to create a profile should have anything to do with the size of the profile.

It will if the measurement data is written into the profile which has been the case with GMB/Xrite over the years.

[GWGill]

I see no reason why the number of patches used to create a profile should have anything to do with the size of the profile.

It does when the patch data is included in the profile itself, which is often done to preserve it (very useful for making possible to regenerate the profile).

But typically there should be a correlation between patch count and the resolution the profile is re-sampled to - there's not a whole lot of point in generating a high res. table from sparse sampling of the printer, and vise versa.