typeface testing

nicolaj's picture

Hi there, I'm about to reinvest in a printer for testing typefaces. Both in small size bodytext and bigger size design testing.

Does anybody have any good advices about printer type, postscript (software) rip, testing methodes etc. How do I simultate ex. offset book printing?


Nicolaj Bak

John Hudson's picture

Hrant, that's presuming that the typeface is in a fit state to be seen in public. I'm talking about testing during the design phase, even with incomplete glyph sets.

Randy's picture

I'm back with another PagePro 9100 vs. Offset.

A couple things to note:
1. In my first sample I printed from the type itself. In this sample I printed from a 1200 dpi photoshop bitmap. The results are much better. I think this may have to do with Postscript emulation. No crappy dithering anymore. This is an anoying extra step, but it is nice to be able to get more out of a cheaper printer.

2. I realized I was comparing Minion Pro with Minion. This is a useful thing because there are slight design differences between the two. See the italic r. This is clearly visible in the printout. This is what I was getting at with being able to see pretty fine structural differences at 1200.

3. The two problems are the traps and the overall darkening. You might be tempted to overdo the traps if you only looked at the laser, or didn't look for this phenomenon. The darkening is definitely greater than offset (which what totally minimal). This is partly the fidelity of the laser print, and partly the nature of toner vs ink. Toner is like laying black paper over white paper, while offset is like black paint on white. Toner is a blacker black.

Anyhow, I hope this helps people make decisions. I think this shows you can get rather far in the design and testing process with 1200dpi, but again it depends on you and your comfort/skill level. All things considered this seems pretty good for an $800 printer at 9pt.

dezcom's picture

Thanks for posting your scans. You show another proof of the weight gain difference. This still seems to be the main bugaboo in Laser print vs offset. Printing in reverse also tells the tale but inchoking way.


hrant's picture

Randy, thanks (again).
It's funny, in the overlays the gain of the laser seems minimal, but when you look at the results on their own it's clear as day. On the other hand, as I think you've said, weight gain is something you can account for "blindly", while in the case of outline infidelities you're lost. The thing is, don't you think the laser has [more] troublesome bumps?

> I printed from a 1200 dpi photoshop bitmap.

Don't you think this might have introduced some gain?
Like how convert-to-outlines does.

BTW, does the Minion-Pro italic lc "r" have a trap where the non-Pro didn't? Huh, maybe Slimbach's long-standing anti-trapping stance has a kink after all...


What would be nice now is to compare Randy's 1200 laser to a 2400 laser, although the emulation/rasterization issues have to be handled carefully. Probably the only way to get an accurate comparison is to also rasterize the type on the 2400 output too. Assuming it matters, just in case: Randy, which version of Photoshop did you use?


ebensorkin's picture

Thanks Randy! Therse were very useful - especially looking at the offset with digital overlay.

levonk's picture

Randy, what type of paper are you using? I just realized that the grains are different. That also may account for some difference in weight gain. Or maybe it is just the scans that came out like that.

Hrant, and anyone interested, I am willing to print some pages for you. If you send me the PDFs I'll rasterize them in photoshop at 2400 and print. I can also scan the prints and post them or send them, or mail the actual prints.

hrant's picture

Levon, very kind of you to offer.
Question: What optical resolution does your scanner have?


Randy's picture

The thing is, don’t you think the laser has [more] troublesome bumps?
Yes for sure. Looking at the bottom of the e. Also the counters of the s. The laser at this size sometime lookes squarish in places that are really round. Remember that this is 9 point. I wouldn't trust 9 point prints for much other than getting the general feel of the type. It just happened to be the sample in front of me (a veer print)

Don’t you think this might have introduced some gain? (like convert to outlines)
I don't think so. I typed it directly into the photoshop (7) 1200dpi bitmap from the font file, so I think it uses the hinting in the font -- or at least adobe's rasterizer. Conversely, sending outlines to a printer, there is no hinting at all. At any rate, the tests show that p-shop's rasterizer is a heck of a lot better than Minoltas.

does the Minion-Pro italic lc “r” have a trap where the non-Pro didn’t?
No, actually. I think the terminal shape is slightly different. I think the above offset (uncoated) shows that Minion doesn't need traps, even in the italic. Slimbach is a giant. Everyone heralds Matthew Carter (rightly so), but Slimbach's body of work is staggeringly good (IMHO). I'm glad adobe has a badass on their team.

Randy, what type of paper are you using?
I've since thrown out the paper wrapper. It's ultra bright laser paper (28lb i think), that I got a Staples. It came in an orange shrinkwrap. Sorry I can't get more specific. I'm using the same paper for all my prints. The offset is on a cream colored uncoated stock (and was very lightly printed).


hrant's picture

> I think it uses the hinting in the font — or at least adobe’s
> rasterizer. Conversely, sending outlines to a printer, there
> is no hinting at all.

The reason to worry is not [just] the hinting. Convert-to-outlines for one comes out thicker (by an average of half a dot) because the rendering algorithms for outlines are different than for type... And the same could be the case in Photoshop.

And to find out for sure, you can't even use your setup, because we don't know if the emulation is gaining (or losing) in addition to that half-bitting. So maybe send something to Levon?

> p-shop’s rasterizer is a heck of a lot better than Minoltas.

Overall, I agree. It remains to be determined if Photoshop and/or the PS emulator are altering the weight.

> Minion doesn’t need traps

1) In the offset, the crotch of the "r" seems fine, but not any of the other crotches... A fluke?
2) Traps aren't just for gain, they help optically as well (especially at smaller sizes).
3) Not all printing is offset! :-)

Slimbach: great work; would be even better with judicious trapping.
And it's interesting to note here that Myriad -which was co-designed
by Slimbach and Twombly*- does have gentle trapping... which is
probably one manifestation of their historic bickering.

* Designer of the boldly trapped Mirarae.


levonk's picture

Hrant, my scanner is 600dpi optical. I use it for low-res scanning to use for proofing. I have access to a Linotype-Hell drum scanner, but that would cost. I can also send it to someone here to scan. But I wouldn't want to scan the text for the final test. If someone wants to test their font I prefer sending them the film/paper output, so they have the real thing in their hands.

Randy, offset ink tends to spread on uncoated paper more than on coated paper because it is oil based and paper absorbs it and spreads. On a laser printer, the ink is powder and not oil based. The spread is due to the way it is applied to the paper, which should be the same on both coated and uncoated paper. Imagesetter output is exact, there is no spread. There could be some weight change when burning to plate from the film, so CTP gives the best result (exact until applied to paper).

My point in all this is: what is it that you really want to test? If the font is to be used on offset, then offset is the best test. It is possible but pretty expensive. I can do that at a reasonably cheap price if you are not in a hurry in getting your tests. I can add some text around jobs. But you have to wait unti I get a job that uses a paper and ink within an acceptable range of your inteded use. I can also do the imagesetter output for you, no wait necessary.

Maurice Meilleur's picture

This is a bit OT, but the answers might bring up things relevant to nicolaj's and others' questions and thoughts on printers:

I've recently asked in the Build section if someone might be able to account for a phenomenon I've recently noticed: using an HP LaserJet 1300 in Mac OS 10.4, characters in documents printed from InD CS and Word seem all to be aligned exactly on the baseline (hence giving the rounded characters like a and o, and vs, the subtle but noticeable appearance of floating higher than the other characters).

I wonder if anyone else has noticed this--the 1300 and its cousins, non-PS all, are pretty popular--and might be able to account for it. If this were something widespread, it would make 1300s rather unsuitable for any sort of font testing, no?

ebensorkin's picture

It sure sounds like it! What happens in other apps? Also, HP lasers have been bashed on typophile for these purposes before. Well maybe not bashed - derided? This was partly because of soft looking output in general and partly because they lack real PS - they just have an emulator. And the consensus is that the emulator is poor.

Maurice Meilleur's picture


The problem seems to be global. I have traded a couple of message with T Phinney on another thread in this section and he seems to recall the same thing happening with a type designer he knows using an HP; Thomas is pretty certain it's the emulator. I gave things a go with the printer at work, using a PC, and found the same problems. Thomas would be better at explaining the problem he suspects than I.

So, here's a question (that perhaps requires another thread of its own?): Is there out there a $500 (or sub-) laser printer of reasonable quality, with 1200x resolution, that has native Adobe PS capability? That might allow reasonable quality for type testing (and some home-grown final production) without breaking the type designer's bank.

Barring that, has anyone tried to write their own (better) drivers for printers without native PS?

ebensorkin's picture

This is the closest thing I have found. In fact it is the only PS laser near this price I have found. It isn't something I would recommend to do testing on because I have not seen the output just the specs and because I now think 2400x2400 is really closer to what you would want for type tests but:

Oki: B6200 ( PS3 & true 1200x1200dpi ) Also come in a Network version for ethernet
B6200 aprox price - $524.99

The thing is the printer is hard to find. For me in Alaska anyway.


See also


Notice they say : " Text quality is a touch below par for monochrome lasers."

Of course maybe they were not invoking the PS and ere using some kind of emulator because of the source application - word or excel most likeley. But maybe it's not all that wonderful. I can't tell yet. I'd like to see a sample. Most likeley I will save up for a Xante or find another route.

dezcom's picture

The big thing I notice with my HP2300 is the overshoot is overshot--meaning it is greater than it should be (I have compared this to imagesetter output.). This may also be partially the hinting or how HP interprets it.
I do notice that my results are better with pro design apps like the Adobe CS Suite than they are with stuff like Microsoft puts out. I guess Microsoft figures people who use their apps are not going to care about the nuances of type.


Thomas Phinney's picture

That's funny. The problem other folks were having with the 1300 was that the overshoot was being suppressed to much too high a ppem size, resulting in the appearance of rounded characters "floating" up above the baseline.

Of course, with any device there is going to be a range of sizes (possibly a couple) at which overshoot is having a slightly larger than expected effect, where it is first kicking in. This will be more apparent with lower res devices. However, intelligent usage of setting the blue zones (or their equivalent in TT hinting) should suppress overshoot at sizes where it would be particularly troublesome.

It's also possible that HP's solution for the "overshoot is being suppressed to too large a size" was to turn off overshoot suppression altogether, or that their attempt to fix it went too far in the opposite direction. (I don't know why either of these things would occur, seeing as this stuff is pretty clear in the Type 1 spec, but what do I know.)

I'd be curious to hear from all the people having these various problems exactly how old their HP devices are. I expect things are better with more recent versions of their PS emulation.




dezcom's picture

The problem occurs at text sizes, 12pt and below. Larger sizes look fine. I have a 2300, not a 1300 if that matters. I bought it new about a year and a half ago. I use it at its highest resolution which is a fudged 1200dpi (They somehow use software to fake 1200 out of 600.) The fonts bundled with Microsoft apps don't suffer as badly as others and TrueType looks better than type 1.
I mostly use it to test fonts that I am designing so hinting is either minimal or nonexisting in the early stages. It makes it tough to decide on proper overshot if your printer is imposing errors. Wish I could afford a 2400dpi Xanté.


raph's picture

Again, sorry for coming to this thread late.

After some careful printing, I believe I can recommend a good inkjet printer and glossy paper for type proofing.

I found a delicate type specimen in the latest New Yorker, matched it on my trusty Epson 2200, and scanned the results. Of course I used Ghostscript for the RIPping, because I have total control over resolution, ink drop size, and color management (I carefully tuned all these parameters to avoid any dither patterns). Here are the results.

I think the Epson compares favorably with other tests posted upthread, including the Minolta QMS PagePro 1250E, the Minolta 9100, and the Xante AW1200.

The 2200 is not the current model. I'm sure the R2400 is just as good or better, and likely the R1800 as well. Indeed, I wouldn't be surprised if the R800 (at $400 list) would give excellent results at the A4 size. It certainly has the specs for it.

If people are serious about using Epson inkjets for proofing type, I can make a GS software kit available. I can't vouch for the quality using the standard Windows driver.

hrant's picture

Raph, I think you should give a presentation at TypeCon about this.


ebensorkin's picture

What is Ghost Script? And does it work on OSX?

And... Even though it is impressive I see some anomallies... The cross bar on the lc e, the feeling of height in the lc r, the gap between the lc 't'&'y', the right side of the Cap U.

Still, it pretty good!

The main downside is ( still) cost per page.

levonk's picture

Eben, ghostscript is a postscript RIP. It is available for OS X but it is a unix or X11 app, and setting up needs some work. I think there are also installers that will work in OS X without X11 or terminal.

There are certainly better RIPs for the job, but ghostscript is free.

ebensorkin's picture

I take your point.

Thanks for that! I'll look into it.

1985's picture

Where are we at with this now? I am looking to buy a printer, and don't want to start a new thread as the basics are laid out so well here. Has anyone had different experiences since? Are laser printers still adding weight? I hope not, after all, this is the future!

ebensorkin's picture

The idea that you are going to have a single source as a reference is something I can't take too seriously now. Particularly if you are talking about testing your font at the intended size. This is because the environments your font will find itself in are to varied for any one printer to be useful. The best things to do is to try to access as many different environments as your budget time will allow ( be realistic here ), and certainly the relevant ones if you are targeting a specific for of printing - but don't kill yourself.

There is a simpler solution than massive testing that will be good for MUCH of what you need. Probably 90% of what you need to know. Get a decent laser printer, ideally that prints to large format paper and print the fonts tests big enough that differences between laser printers start to be irrelevant - so... fairly big. And then take your print out, put it on a wall, and step back until the letters become the optical size is correct for your intended use.

I hope this helps.

Syndicate content Syndicate content