Sunday, June 11, 2006

Gutenprint / Gimp-Print inkjet RIP review Part 1 — RGB

I have started to look at the Gimp-Print inkjet drivers (now called Gutenprint). In conjunction with other preinstalled components such as CUPS and Ghostscript, this open source system project forms a RIP system, supporting 8 and 16 bit RGB and CMYK files as well as PDF and Postscript, and it's present on every Mac, and can run on every Linux box.

Today, I'll just discuss my findings when using the beta Gutenprint V5.0 RC3 as an inkjet RGB driver for printing images from Photoshop.

Nutshell report: Very easy to install and very good results, on a par with the proprietary Epson RGB driver. However, it's painful to configure. Gutenprint's excellent functionality is obscured (read obfuscated) by an interface overloaded with undocumented options.

Equipment used: Epson 2400, connected to a Dual 2.5 Ghz PowerMac G5 printing on Epson archival matte. Profiling with XRITE DTP-70 spectro and Monaco Profiler 4.8, eyeball evaluation by using the Chromacity target, and male and female eyes.

Installation: On the Mac, an update installation of the latest GutenPrint is easy, via a package that can be downloaded from Sourceforge. Creation of a print queue is done via the Mac's Printer Setup app, and is well documented in a PDF supplied with the package. Gutenprint does as well here as the Epson driver.

Printer Configuration: On the Mac you set up the printer by pointing a browser at the CUPS web interface, located at
http://localhost:631 and there are zillions of incomprehensible parameters. Hint: Set the printer to Line Art if you are doing profiled printing with Photoshop working as the color engine. I'd say this is the only bad aspect of this driver.

Printing: You can print to your newly installed Gutenprint device from Photoshop via the usual Print with Preview dialog. Although the margins are different, everything ends up at the right place when you click the "Center Image" checkbox. Nothing new here - and that's exactly the way we want it.

Print Speed: On my Epson 2400, Gutenprint was slow. Not unusably slow but much much much slower than the Epson driver. Long pauses between passes. Maybe I need a USB 2.0 cable ?

Profile Quality: I profiled both Gutenprint and the proprietary driver with Monaco Profiler, using the simple 343 patch target that results in very smooth inkjet profiles. Looking at the target indicated Gutenprint has a better linearization out of the box. The gamut looked fairly similar to the Epson driver, except in the reds where there may be a bug. The Gutenprint gamut seemed better in the shadows.

Print Quality: Looking at the Chromaticity target, rendered perceptually.
  • -Neutrals were neutral, shadow gradients smooth.

  • - Gutenprint had better shadow detail.
  • - The Epson shows a smidge more contrast (due to the blocked shadows ?) and slightly more sharpness.
  • - Highlight definition of colored zones looks better with the Gutenprint driver.
  • - I'd say the reds are significantly better with Epson's driver.
  • - Examination with a loupe showed a weave/stripe pattern on the Gutenprint sample, while the Epson sample showed more random fill patterns. However this is not visible to the naked eye.
  • - The Colorchecker looked good on both samples, except - you guessed it- for the red patch. The red is better but not perfect as rendered by the Epson driver. But then, of course I printed with perceptual rendering.

Daylight/Indoors: Disparities between the samples seemed stronger in incandescent light. They matched more closely in daylight.

Suggested improvements: A minor quibble, there is a slight gamut issue with the reds on the 2400, at least on matte paper. I guess this is something that could be easily fixed by adjusting the linearization settings, but I don't know how to do it. Apart from that, to solve the zillion settings issue by giving people decent starting points, I suggest bundling some pre-configured settings into the install package, one of which would be a good-quality no-color-correction settings-set for the CMS crowd, and another for normal users wanting to print photos. I'd also recommend some sort of HTML documentation for the settings options. Apart from a very slight deficit of sharpness, I don't see much in the functionality that needs improving. Some tools for linearisation might be a useful adjunct however. Oh yes, and then there is the speed problem.

Summary: The Gutenprint RGB printer driver, when profiled, offers equivalent quality to Epson's printer driver. Indeed Gutenprint seems particularly well suited to portraits because of good shadow and highlight rendering. In the next part of this review we'll be looking at what Gutenprint can do that Epson's driver cannot - did you say CMYK and Postscript ?

Addendum: Yes, I will make custom profiles for this driver on request. Provided you already know how to use it.

9 comments :

Anonymous said...

Some other "bad aspects" for anyone seriously considering dumping their RIP for unproven open source alternaware:

1. No control over head alignment or paper feed

2. Can't do a thing with spot colors

Anonymous said...

Hi Edmund,

Thanks for the information.

I have recently acquired an R2400 and thought I'd pass on a few comments.

I have no experience with the Mac, as I am using Gutenprint with Cinepaint, the 16 bit version of the 1.x.x Gimp, on Linux.

While it may lack some of the filters and features in the current Gimp, the "killer" feature in Cinepaint besides it's high bit depth, is that it is colour managed and has a plugin which allows realtime feedback from changes to be displayed in iccexamin. This app allows a high resolution 3d display of the colour gamut, which can be rotated and zoomed through 360 degrees and displays the image data, the working space and proof space.

Kai Uwe Behrmann has put a considerable amount of work into both projects.

Anyway, your comment about the printing speed being very slow makes me think that you are using an old version.

If you are unable to get a package you can use, built from the current source available from CVS, you will be well advised to re-test once Gutenprint 5.0.0 is finally released (as opposed to the release candidate you are currently using), as a number of significant changes have been made, including printing speed ups and ink density changes for matte papers.

As far as the lack of docementation for all the variables - go here:

http://sourceforge.net/project/shownotes.php?group_id=1537&release_id=418068

and have a look at sections V and VI.

I hope to soon have a profile made for Epson Premium Glossy and if you are interested, I can send you an iccexamin gamut display comparing it to Epsons public profiles.

Thanks again for the post.

Regards,

Richard

Anonymous said...

Sorry, the URL for the documentation has been cut off, so here it is again, on 2 lines:

http://sourceforge.net/project/

shownotes.php?group_id=1537&release_id=418068

reghards,

Richard

Edmund Ronald said...

Richard,

Thank you for the information about Cinepaint. However many will be apprehensive about the fork from the main GIMP path and do not understand why the GIMP does not integrate those features.

As for gamut diagrams, I have at least 10 pieces of software that make them, including Steve Upton's excellent ColorThink.

I chose on purpose to put in that matrix image because ANYONE can read it in one glance.


Last, not least, I have an offer for profiles on the main page - all you need to do is ask me for the chart.

Edmund

Anonymous said...

In regards the first reply to this review:

Gutenprint certainly is not the be-all and end-all of drivers. However, "unproven open source alternaware" is not a useful comment to make. If you want to criticize Gutenprint, do so directly and specifically rather than via innuendo.

As for your specific comments:

1) I'm not sure what you mean by "no control over head alignment or paper feed", but Gutenprint does include a command line utility to perform head alignment. We haven't included it in the OS X package, but I've asked our OS X packager to investigate including it.

2) Gutenprint can actually handle spot colors just fine; we just don't have a way to get it through CUPS.

--Robert Krawitz, Gutenprint project lead

Anonymous said...

Hi Edmund,

I appreciate many of your readers may not wish to go down the Cinepaint path, but thought it may offer something for those who enjoy getting their hands dirty.

Cinepaint with Gutenprint offers a few advantages - one could use it for final printing after image manipulation in Photoshop if required.

1) It offers 16 bit printing, as the Cinepaint print plugin works directly with the printer, (bypassing CUPS). I do realise that the latest version of CUPS is now 16 bit.

2) The Cinepaint Gutenprint plugin (and Gimp print plugin) offer extra adjustment parameters that are not available from CUPS - see entries K (OUTPUT LINEARIZATION CURVES) and L (DETAILED COLOR CORRECTION CURVES) in section VI of the documentation reference above.

I understand that release 5 of Gutenprint will be the last where this situation exists and all these features will will appear in future versions of the CUPS drivers.

It seems to me that for someone with profiling equipment and the knowledge to use it, that extremely fine tuning of the ouput is possible, to a degree the Epson driver does not allow.

That this has been achieved and freely distributed by Robert and many others, is remarkable to me and hugely appreciated.

I look forward to reading the rest of your investigation into Gutenprint.

Out of curiosity, what are "spot colours", mentioned in previous posts?

Regards,

Richard

Anonymous said...

OK, here's a specific criticism:

The attitude of open source developers is parochial to an extreme, with their primary impetus for working stemming more from the undeserved sense of superiority they derive from repeatedly bashing everyone over the head with their utterly irrelevant personal preferences in operating system software rather than from wanting to develop genuinely useful tools that can be comfortably trusted in professional environments where people have deadlines to meet and do not work out of their parents' spare bedrooms for free. There are certainly exceptions to this rule (Apache comes to mind immediately), but with nonsense like

"There's not much I can say about Photoshop, since I don't use it (I want to create an entirely FOSS-based workflow), so you're on your own for that."

coming out of your eMouth, you have little to no hope in the world of ever being taken seriously by anyone other than your fellow Linux nerds. If you want to develop free software, good for you, but this aloofness you exude in choosing to pretend as though you don't even know commercial software exists does nothing for anyone but your ego. You could very well be the best programmer and/or color scientist the world has ever known, but that doesn't necessarily mean you know anything about the design, photographic or pre-press industries, and until you drop the airs and start paying attention to the way the rest of the world operates around you the only reason any Mac user ever needs to use your drivers is if they have a printer that isn't supported by its vendor which they are too cheap to simply replace. It doesn't matter how many machines it comes pre-installed on, until Gutenprint either becomes easier to use or offers some clear and immediately apparent benefit, it's never getting adopted by the Mac community at large. We have our own bizarre little ideologies about software we like to flaunt, and notions like "I'd rather pay a little bit of money once than have to sift through 20 pages of obtuse, outdated documentation on sourceforge every single time I want to print a picture" are quite commonplace.

And yes, not including a given tool in a given distribution is a great way to make it look like it doesn't exist. I guess I should have setup a Linux box and tried printing out some giant posters of Richard Stallman from there before shooting my mouth off about that, but hey, whattyaknow, I don't care enough to bother; it's a ton of effort for little to no reward, and I'd be mighty surprised if any (profitable) print shop in the world went anywhere near Gutenprint, Gimp or the IRC chatrooms where their usage gains any 12 year old child as much instant social acceptance as smoking and using the oh-so-clever abbreviation "M$". I'm sure the project will look great on a resume and will garner you a minor amount of fame in certain circles, but as far as I (and pretty much everyone I know who actually makes any money printing pictures) am concerned, it's an overly-politicized novelty act at best. Maybe that'll change someday, but I'm not holding my breath.

Edmund Ronald said...

Richard, I'm not sure who you are, could you email me ?

Anonymous poster - Please remain polite even if you disagree with other posters.

Anonymous said...

I was hoping that you (anonymous poster) would have specific issues with Gutenprint. My reasons for working on Gutenprint are my own, and they're documented on our site; the initial reason was simply that I didn't like spending $70 for a large print to put up on our wall and I wanted a fully Linux/FOSS-based solution for this. I was as surprised as anyone when Gimp-Print took off on the Mac. I won't say that I was displeased in the least; it's been a lot of fun, but at the same time I do have a very specific motivation in mind, and I'm not asking to be taken seriously except on the merits or demerits of the package itself.

My goal is still to produce the best printer drivers out there, period -- not copy anything existing. I don't mind learning from what's done elsewhere, but most proprietary software has license clauses that don't allow trying to figure out how it works under the hood, and I like to get my hands dirty. I do get a bit ornery when someone tells me that I'm not even allowed to figure out how something I own works or try to improve it. License clauses that read "you're not allowed to reverse engineer, disassemble, or otherwise attempt to discover how the program works" are flat out offensive to me, and I will not personally use such software.

I'm willing to accept -- not necessarily use, but politely accept -- reasonable suggestions from anyone who knows what they're talking about and/or can give me experimental evidence that they can do something better than I can. A lot of people who know a lot more about color science and dither algorithms than I do have given me excellent suggestions, and I've even used a few of them :-) At the same time, I do have my overall goals, which I'm not willing to compromise.

That all said, a lot of people in the Macintosh community have been very helpful indeed. Edmund's review is very insightful, and there's a lot of very useful material in it.