What Anti-Aliasing do you use?

[Gregg]

I'm curious if there is one Anti-Aliasing method that is better than the rest for files that will be printed on an offset press.

I recently had an issue with a cover file that had rasterized text in the flattened TIF file, the Anti-Aliasing was set to Crisp. Now, when we printed the jacket everything looked fine (printed about 1 year ago). The printer did not need to trap the elements within Photoshop, since there was Anti-Aliasing present. The problem arose when the paperback was sent to another printer, and there was a noticeable outline on the rasterized text. They felt the image needed to have Photoshop trapping applied. So I was curious if there is a preferred Anti-Aliasing method that would have solved this issue.

For the record, I know it is not best practice to have rasterized type in a photoshop file (believe me I KNOW THIS). This was just one of those situations...

 

[rich apollo]

If you save the file as a Photoshop PDF you can maintain the type in vector form. Then it's not an issue.

 

[Gregg]

Thanks, Rich. I do know that and maybe I should have been more clear in my post. Basically, I am wondering if there is a preferred Anti-Aliasing method when you absolutely have to rasterize type. Perhaps it is somewhat of an outdated question, but I'm sure there are some out there that still prefer to not receive live type in their P'shop files.

 

[apepp]

If you save the file as a Photoshop PDF you can maintain the type in vector form. Then it's not an issue.

...only if the type isn't using photoshop faux styles like bold, italic etc, etc...

 

[Lukas Engqvist]

I would say the style depends on the effect you are after (and to some degree the image). If you must rasterise text i would suggest considering increasing Image ppi.

 

[claude72]

Basically, I am wondering if there is a preferred Anti-Aliasing method when you absolutely have to rasterize type

AFAIK, (but perhaps I have a lack of english vocabulary) anti-aliasing works only for screen-display...

but I'm sure there are some out there that still prefer to not receive live type in their P'shop files.

I overall prefer not having type in Photoshop files!
Live type is a real nightmare to succeed to keep them in vector mode, with black N100 only and with correct overprinting/knock-out...
... and rasterized type is crap job, as you already "KNOW THIS"!

But when there is really no other choice than Photoshop, I prefer to receive a PSD file with live type (and I spend the time needed to keep the type in vector-mode to print quality job) rather than to rasterize the type and print crap job.

The printer did not need to trap the elements within Photoshop, since there was Anti-Aliasing present.

??? again, AFAIK and if I good understand the sentence, trapping and anti-aliasing have no relationship: trapping is for printing, anti-aliasing is for screen-display ???

The problem arose when the paperback was sent to another printer, and there was a noticeable outline on the rasterized text. They felt the image needed to have Photoshop trapping applied.

Is it a mis-registration issue with printing of a CMYK black?
(it's the most common problem when printing rasterized text... but no trapping will be able to correct this issue!)



**************

If you must rasterise text i would suggest considering increasing Image ppi.

That's correct for a text in 1-bit picture : the normal resolution is 1200 ppi, and at this resolution a rasterized text is perfectly printed.
So, rasterizing a text as a 1-bit picture at 1200 ppi instead of the common used 300ppi resolution will greatly improve the printed result.

But it's absolutely no use for a text in a contone mode picture: increasing the resolution of a contone picture beyond the normal 2x quality factor (300 ppi for 150 lpi screen for exemple) will only increase the weight of the file and increase the time of working on the file (transferts, ripping, etc.) without changing the printed result: when you print a contone picture, the pixels of the picture cannot be printed "as is" and must be transformed (by the RIP) in screen-dots... so, after printing, on the paper you can no more see any pixel, but you can only see screen dots... and increasing the resolution of non-visible pixels changes nothing !!!
In order to lower the haziness of a rasterized contone text, you have to change something that it visible after printing: the only way is to reduce the size of the visible screen-dots, by increasing the printing screen-ruling... which will "improve" the result (or more exactly lower the flaw!!!): normaly, with a 175 lpi screen the haziness is no more noticable by a normal customer...

 

[Lukas Engqvist]

But it's absolutely no use for a text in a contone mode picture: increasing the resolution of a contone picture beyond the normal 2x quality factor (300 ppi for 150 lpi screen for exemple) will only increase the weight of the file and increase the time of working on the file (transferts, ripping, etc.) without changing the printed result: when you print a contone picture, the pixels of the picture cannot be printed "as is" and must be transformed (by the RIP) in screen-dots... so, after printing, on the paper you can no more see any pixel, but you can only see screen dots... and increasing the resolution of non-visible pixels changes nothing !!!

Have you tried it? Or are you extrapolating from theory?

If you have a blue image with white text you will have a difference in the edge of the characters if it is 600ppi or 300 ppi, provided you do not downsample in pre-rip conversion to PDF. (I know the DALIM workflow tends to do that in configurations I have seen).

2x quality factor is not for sharp CG edges. It is for apparent photographic detail.

 

[claude72]

Have you tried it? Or are you extrapolating from theory?

Both!

The theory is that in contone mode no pixel can be printed and only screen-dots are visible on the film/plate/paper...

... and I doubly confirmed the theory:

1- with my daily pre-press job: today, most designers are too lazy to search for vector-logos or to rebuilt pixels-logos in vector mode (or don't understand why they should use vector logos!), and they do the jobs using the jpeg-office logos provided by the customer (or other crappy pixels-logos)*...

... and it's then easy to compare the printing results of:
• either the same JPEG logo used at two different sizes (meaning at 2 different output resolutions) in 2 different NATIVE files, one for a poster and one for a flyer,
• or the same PDF file imaged twice at two different sizes to print both a poster and a flyer
(I did such a job 2 days ago with one 200 ppi Photoshop PDF file, imaged twice: once at 100% for the poster, and once at 33.82% for the flyer, giving a 591.3 ppi resolution for the flyer)...
... commonly, the size ratio between the poster and the flyer is about 3 times, meaning that in a native file the resolution of the logo in the flyer native file is about 3 times greater than the résolution of the logo in the poster native file, but with the same screen-ruling the printed result is as crappy in the poster and in the flyer!
(and sometimes, the text in the little logo of the flyer is so tiny compared with the screen-dots that it becomes unreadable, althought it is readable- hazy, but readable - in the logo of the poster)

* (In these cases, if I get native files, I take some time freely to replace (by vector-logos) as much crappy JPEG logos than I can replace, but I cannot have or find all the logos, and I cannot spend too much time (without being payed for this job) to rebuilt all the logos, so some JPEG logos remain untouched...
... and when I work with ready-to-print PDF, I leave the PDF as it is given, with its crappy JPEG logos!)


2- with a test: I printed at 2400 dpi / 150 lpi some Times 8 pt text-images in grayscale mode with different resolutions (2400-1200-300-225) using an Agfa Viper3 RIP set to output the 1-bit raster to a file (instead of outputting to the imagesetter). Then I gave a blue color to the pixels of the 1-bit raster and add a red outline to show the real shape of the glyphs and allow the comparison...

As you can see, there are some changes in the screens dots positions, due to different placements of the picture on the screen grid, but whatever the resolution is, the result remains as crappy!

... provided you do not downsample in pre-rip conversion to PDF. (I know the DALIM workflow tends to do that in configurations I have seen)

Yes, some workflows systematically resample contone pictures at a given résolution... and you're right, the DALIM workflow is set by defaut by its manufacturer to resample all contone pictures at 300 ppi (and this setting is often left untouched by installers and users)...

... but I'm sure that my Viper 3 RIP does no resampling, and I'm sure that my contone text picture have been imaged with their real resolutions!!!

 

[kandod]

I don't know what theory says, but when I rasterize text at 600 ppi no AA the result with my HP 5000 inkjet continuous tone equals that of printing vector text at 600 dpi print res (not that I am actually doing that in practice, due to the size of the image produced I rarely go beyond 150 ppi with anti-aliasing on).

 

[claude72]

I don't know what theory says, but when I rasterize text at 600 ppi no AA the result with my HP 5000 inkjet continuous tone equals that of printing vector text at 600 dpi print res...

First, theory says that your HP 5000 ink-jet printer uses a stochastic screening technology (aka "FM screening") that lower the hazyness enough to make it invisible for normal eyes: that's why you don't see any difference... and that's why you believe that there is no difference... but you're wrong: the difference exists! and a sharp look, with a good magnification shows that the flaw is the same.


Second, in this topic, we are talking about offset printing (not ink-jet printing), with AM screening (not FM screening)... and offset 150 lpi AM screening doesn't lower the hazyness as does the FM screening of your ink-jet printer: that's a big difference!
(experienced people with expert well-trained eyes say that the FM screening of an ink-jet printer does the same job that a 175-200 lpi AM screening)

(And that's one reason why many people use Photoshop to make their layout, or rasterize their files before giving them to a print-shop: just because the result seems correct on their ink-jet printer, and it makes them believe that it will be equal (or better) with offset printing...
... as they don't understand that a 2 millions dollars offset press will not give an as good result than an ink-jet printer bought 50 dollars in the 2 blocks away supermaket.)


Third, an imagesetter (film or plate) prints commonly at 2400 dpi... and the vectors output much more crisp from an imagesetter than from your 600 dpi HP5000 ink-jet printer...

 

[Lukas Engqvist]

Well this is what it looks like in my rip

Well this is what it looks like in my rip

The places in practice I have found a difference is as you say in logos, and in negative text on dark backgrounds. Screendumps from Apogee, raster preview 1:1

 

[claude72]

Screendumps from Apogee, raster preview 1:1

Apogee is not very different than a Viper, but your 600 and 1200 ppi tests are obviously very different of mine... ???