Photoshop CS4 and Type

[Prepper]

Is there a new way to preserve type as vector in CS4? My designer sent me files with all the type in PS and at 600dpi instead of our normal 300. He says he saw a training video that said if you made it 600dpi it will be smoother and not rasterized, will be vector like Illustrator. I know, I know, but any chance that there's something new on this front?
I normally take his images with fonts and save as PDF and relink to the PDF in INdesign to keep text sharp as possible but that only works if it has no effects applied in PS to start with.
The 600dpi files he sent me, one 15.5" x 10.5" spread is over 800meg!

Thanks

 

[kyle]

Text from Photoshop saved as a PDF isn't exactly vector. Without any effects, it is encoded as an image as large as necessary to fill the text area with all of the pixels the color of the text, then that image is clipped by a text clipping frame. As far as sharpness goes, it's just as good as straight type because the edge will be rendered at device resolution, but it won't trap because there isn't a normal vector edge.

If saved as a PDF, there is no advantage to making the image 600 ppi as far as the text is concerned, as long as the type layers aren't rasterized. Even if you apply effects, the appropriate edges are maintained as clipping text. You will only get rasterized type if a faux bold or italic effect is applied, or if the font is not available. Anti-alias settings make no difference when the edges are preserved as clipping text.

If for some reason the type needs to be rasterized, 600 ppi would be better than 300 ppi, but turning anti-aliasing off will increase the sharpness more than doubling the resolution. A large image filled with rasterized type would probably also compress smaller with ZIP or LZW compression if the type is not anti-aliased.

If files are saved with less compatibility options (like preserving Photoshop editing capabilities for PDF's), the files are often much smaller due to the lack of redundancy, but you probably want all of that extra information if you have to mess with the files.

There shouldn't be any case in any context where increases the resolution past a magical point causes anything to change from raster to vector.

 

[Joe]

If you have text in Photoshop and output a PDF the fill of the text will be an image but if you select the text with Pitstop in Acrobat you will see it selects it as a Clipping textline object. The clipping textline will make the edges high resolution so when you output it, it will look like vector type. Here is a screen shot from Acrobat zoomed in on the type with the clipping textline shown.

There should be no need for 600 dpi raster images from the designer.

 

[hansman]

If you have text in Photoshop save it as an .eps when the eps option box pop's up chose include vector data. Resolution can be 300 for the raster (photo) data. Place eps in In Design and create a pdf as normal.
This will preserve all vector data as well as enable you to lower the resolution of the entire file to 300.

 

[Joe]

If you have text in Photoshop save it as an .eps when the eps option box pop's up chose include vector data. Resolution can be 300 for the raster (photo) data. Place eps in In Design and create a pdf as normal.
This will preserve all vector data as well as enable you to lower the resolution of the entire file to 300.

In CS4 that basically gives you the same thing I described above. You end up with a 300 dpi image with a high res clipping path for the text edges. Either way the type still outputs at high resolution.

 

[Lukas Engqvist]

What colour type on what background.
If antialiasing works really depends on the size, colours (background and text), and fonts.
A small white text on a rich background will allways be a problem in photoshop especially if bodoni or a serif font is used (but you all know that). In those cases a 600ppi will be a definate improvement.

If InDesign is set to downsample at 300ppi in the PDF joboptions, then you will counter all the extra load of handling 600ppi…*so your joboptions also need to have 600ppi.

If it trapps is an issue of trap settings, it can be trapped if the RIP is configured that way.
It is a good Idea to work as you do adding the text in InDesign, it gives more flexibility and better editability. It also means that the photoshop is able to stay in PSD (preferably as an RGB so that InDesign can be the place for Colourmanagement IMO), and you do not have the potential error of wrong PDF settings in the creation of the Photoshop PDF.

 

[Prepper]

Thanks for the clarification on vector vs. outlines in Photoshop, I was just using the vector term as referring to sharp crisp edges, I learned a little on text in PS. He loves the faux bold button, just talked with him again and he just now realized that that doesn't actually use the bold font like picking the bold version in the font dropdown does, although I'm pretty sure we've covered this lots in our 20+ year working together.

Anyway, your responses verify what I thought, and I learned a little, I do just normally take his .psd's and save them out as pdf so that the type that doesn't have faux bold or italics applied can be as crisp as possible.

Lukas, we have been using RGB into Indesign almost exclusively here for several years, this way we can take our legacy files and convert them upon export from INDD to any cmyk format we may need to at the time or for any other purposes. Files we send overseas have to be done with ISO Coated and ours here are done G7. We do have a lot of reversed out type on rich backgrounds (Caslon 10pt. on rich black background) but that type is in INDD, wouldn't the color of the type (% of color) and the screen ruling be the biggest limiting factors to how sharp it will be on paper, and ultimately press registration?

The designer just likes to create many samples in Photoshop and says it's easier to do it all there in one place and doesn't want to take the time, and doesn't have the time, to recreate it in INDD at the end, and I see his point on that. So I just take those kind of pages and convert them to pdf and relink in INDD before I do the proofs and mockups.

I never do any downsampling as a rule, just kind of as a safety that I know I didn't alter something somehow. Do you all? I would really like to sometimes because of the large files placed in INDD and scaled down to small size, and sometimes many of them, make huge files to us to process.

Thanks again,
Terry

 

[Joe]

It's like I said, if he makes his files as 300dpi the text will still output at high res, 2400 dpi or whatever your output device is, because of the clipping path of the text. I do downsample all images to 300 dpi, which if there is any text that is pure raster image at 600 dpi or higher it will down sample but it shouldn't hurt the text he is doing in Photoshop because it will not down sample the clipping path. That is, as long as the text isn't getting rasterized somewhere along the way.

 

[Lukas Engqvist]

Reversed text on rich backgrounds will allways be a problem, even in high end fashion magazenes. If it turns out bad even if you are breaking the advice of experienced people it is called "a bad printer". I would support you in your discussion with the client, you have told him the risk, if he will not listen he has made the choice. There was another thread discussing resolutions, and i posted a screen shot rasterd text of different ppi. Try it and compare, no point arguing thory when it is easy to test.

 

[Joe]

Reversed text on rich backgrounds will allways be a problem, even in high end fashion magazenes. If it turns out bad even if you are breaking the advice of experienced people it is called "a bad printer". I would support you in your discussion with the client, you have told him the risk, if he will not listen he has made the choice. There was another thread discussing resolutions, and i posted a screen shot rasterd text of different ppi. Try it and compare, no point arguing thory when it is easy to test.

While I agree about the white text on rich black background, because not matter how high res it is it will always be a problem with registration, text from Photoshop is not rasterized if you output a PDF directly from Photoshop. It contains a clipping path that will output the text at the resolution of the output device. Do I recommend it? Nope, but it will work.

 

[claude72]

Text from Photoshop saved as a PDF isn't exactly vector. Without any effects, it is encoded as an image as large as necessary to fill the text area with all of the pixels the color of the text, then that image is clipped by a text clipping frame.

Yes, you're right: in a PhotoshopPDF, the text is not real vector text, but is in a (strange) vector mode...

... but in a PSD file, text is normal vector text, handled normally by a normal Adobe text-engine, and the strange structure using clipping pathes over pixels colored images is only for outputting in PDF and EPS formats.

it is encoded as an image as large as necessary to fill the text area with all of the pixels the color of the text

And sometimes the image isn't large enough to fill the text area, and some ascents or descents are missing... (in fact, they exist, but they are not visible because there is no color under them to make them visible)
(I had this problem, long time ago)

If for some reason the type needs to be rasterized, 600 ppi would be better than 300 ppi,

and

In those cases a 600ppi will be a definate improvement.

I already had such a discussion with Lukas Engqvist, and we had slightly different opinions, (sorry Lukas, I didn't yet make further tests of outputting rasterized text)...

... but, theorically, there is (or should be) no difference between 300 and 600 ppi for a contone picture, because the pixels have to be transformed in screen dots to be printed, and what can be seen on the paper are not the pixels, but only the screen dots...

... so, (assuming, of course, that the resolution is high enough to match with the screen-ruling) the aspect of the rasterized text will not depend of the resolution, but will depend only of the printing screen-ruling, and only a finer screen-ruling can give a real improvement.
(assuming that the picture has at least 300 ppi, a 175 lpi screen will be a good improvement compared to a 150 lpi screen, and a stochastic screen will be a better improvement!)

If you have text in Photoshop save it as an .eps when the eps option box pop's up chose include vector data.

Outputting EPS (including vector data) or PDF from Photoshop are 2 suitable ways to output crisp vector text from Photoshop. But none is a better solution than the other, and problems occur with both...
The good news is that problems occurring with PDF can often be fixed using EPS, and problems occurring with EPS can often be fixed with PDF...
... but sometimes both PDF and EPS do not work properly
(it happened to me last week!)...



But, hopefully, there is a third way to output crisp text, it is simply to print/image directly from Photoshop, as Photoshop is able to print his own vector text in vector mode. It works great for posters being printed with a single pose on the plate.
If imposition is needed, it can be done by imposing the rasters outputted by the RIP...
... and if not possible, another solution is to make the imposition in Photoshop, or to output films and make manual stripping, as we did in the previous century...

I never do any downsampling as a rule, just kind of as a safety that I know I didn't alter something somehow. Do you all?

Yes I do... all, no... but when stupid "designers" send me pictures with tremendously too high resolutions, I often downsample to the good resolution, to avoid too large files.

(one common mistake is the use of "pixels per centimeter" instead of "pixels per inch": 300 pixels per centimeter = 762 ppi, giving an unusefully nearly 6.5 times larger file )


But, you must keep in mind that:
1° for each output screen-ruling the RIP needs an optimal resolutions range to match with the screen-ruling...
2° the rasterization made by the RIP is a kind of re-sampling: using a 300 ppi picture to output an 150 lpi screen means that each screen dot will be built from 2 x 2 pixels, so 4 pixels have to be "melted" together in one screen-dot... and that's a real 2-time downsampling!
3° downsampling removes pixels, and removing pixels means removing details: more pixels you remove, more details you loose...

... so:
• if you don't downsample your pictures manually in Photoshop, and send the pictures with all their unuseful pixels to the RIP, then the RIP will do the downsampling-job, but it will do it without any possibility for you to really control the result before the first sheet outputs from the press, when it is too late...
• but if you downsample your pictures manually in Photoshop to set the good resolution matching with the screen-ruling before ripping, then you can control the result on the screen in Photoshop, and you can see whether the picture remains acceptable or if too much details are lost...
... and if needed you can redo the job trying another downsampling algorythm giving a better result (or a less bad result).

The designer just likes to create many samples in Photoshop and says it's easier to do it all there in one place and doesn't want to take the time, and doesn't have the time, to recreate it in INDD at the end

Many "designers" work that way...
But that's not a designer's job, that's simply a butcher's crap-job..
... and these "arguments" are typically what bad "designers" always argue to justify their crappy bad job.
(once, one said to me that "using InDesign + Illustrator + Photoshop limits his creativity, tought using Photoshop alone allows him to free his mind from technic constraints and allows him to be more creative"... that's only bullshit: he was simply uncompetent!)

OK, it's possible to output crisp vector text from Photoshop, but with some extra job (and extra cost), but with many other problems occurring, like problems with the separation/knock-out/surimpression of the black, no trapping possibility, one file for each page, logos and other elements outputting hazy because in pixels instead of being vector-based, etc.

Definitely, Photoshop is not a text layout software, and good designers work with vector tools.

 

[kyle]

If it trapps is an issue of trap settings, it can be trapped if the RIP is configured that way.

Are you referring to trapping of rasterized text, or vector-based clipping text? Our RIP uses Adobe's APPE, and I am unable to trap clipping text edges. Since the clipping text is not really the edge of anything but instead a mask that reveals an image, I wouldn't expect any trapping process to trap it unless everything is rasterized and flattened before a raster-based trapping is applied. I would be interested to know if anyone has a RIP that can trap clipping text.

I never do any downsampling as a rule, just kind of as a safety that I know I didn't alter something somehow. Do you all? I would really like to sometimes because of the large files placed in INDD and scaled down to small size, and sometimes many of them, make huge files to us to process.

I always preserve every file customers supply. If I have to modify anything, I create a new file with a suffix indicating wether the change was made before the first proof, or on which round of corrections. I have never reduced the resolution of any file to speed up processing, because that would just be moving the process upstream, and based on my practices would create a new file taking up more space on our file server. The extra time an image takes to process because it is unnecessarily large may as well occur automatically when Indesign exports a PDF instead of when an operator opens the image and resamples it. The resampling algorithm used by the PDF export process is also superior to any method available in Photoshop for any image that is less than twice the final ppi (cf. http://printplanet.com/forums/digital-printing-discussion/20739-new-resizing-images-pdf#5 ). If you have a single image that is linked into many documents, then resampling the link would probably make sense. If a single link is on many pages in the same document, there may be no advantage to resampling the link if they are all at the same scaling percentage because I think Indesign encodes the image once and references the same resource on multiple pages.

...but in a PSD file, text is normal vector text, handled normally by a normal Adobe text-engine, and the strange structure using clipping pathes over pixels colored images is only for outputting in PDF and EPS formats....But, hopefully, there is a third way to output crisp text, it is simply to print/image directly from Photoshop, as Photoshop is able to print his own vector text in vector mode. It works great for posters being printed with a single pose on the plate.

Unless the behavior has changed with CS4, when Indesign interprets a PSD file, it uses the text layer's pixel data instead of rendering the text in the font referenced by the file, yielding completely rasterized type. When Photoshop prints, it uses the same clipping-text method in the Postscript as it does when it saves as a PDF. I guess there's a chance it would perform otherwise when printing to a non-Postscript printer, but I would bet against it.

...but, theorically, there is (or should be) no difference between 300 and 600 ppi for a contone picture, because the pixels have to be transformed in screen dots to be printed, and what can be seen on the paper are not the pixels, but only the screen dots...

―

1° for each output screen-ruling the RIP needs an optimal resolutions range to match with the screen-ruling...
2° the rasterization made by the RIP is a kind of re-sampling: using a 300 ppi picture to output an 150 lpi screen means that each screen dot will be built from 2 x 2 pixels, so 4 pixels have to be "melted" together in one screen-dot... and that's a real 2-time downsampling!
3° downsampling removes pixels, and removing pixels means removing details: more pixels you remove, more details you loose...

... so:
• if you don't downsample your pictures manually in Photoshop, and send the pictures with all their unuseful pixels to the RIP, then the RIP will do the downsampling-job, but it will do it without any possibility for you to really control the result before the first sheet outputs from the press, when it is too late...
• but if you downsample your pictures manually in Photoshop to set the good resolution matching with the screen-ruling before ripping, then you can control the result on the screen in Photoshop, and you can see whether the picture remains acceptable or if too much details are lost...
... and if needed you can redo the job trying another downsampling algorythm giving a better result (or a less bad result).

A normal RIP does not resample images to fit a screen ruling - there is no need. The screening process does NOT work this way:

Every halftone cell covers an area of pixels. All of the pixels in that area are averaged out to determine the weight of that cell. The average value is then used to select which of a number of possible tiles is used to fill that cell, with each possibility being a strictly binary black-and-white dot of varying size. This is NOT how it works.

The screening process DOES work this way (or by a similar algorithm with the same effect):

Before screening, all of the various abstract elements are trapped, rasterized and flattened (not necessarily in that order). For each ink channel, the RIP then works from a flat raster image that is contone (typically 256 levels, like a grayscale image). It then uses a screen tile that is also contone and for traditional AM screening would be visualized as sort of a blurry checkerboard. This image usually has many halftone dot positions in it so that more precise angles and more levels can be achieved, and is continuous in the sense that you can tile it repetitively and not be able to see the seams. The screen tile is then repeatedly tiled over the flat contone image (think of it as a separate layer sitting on top). The RIP then compares at every pixel the value of the image layer with the value of the screen layer. If the screen layer value is less, the final 1-bit screened image pixel becomes black/inked. If the screen layer value is greater, the final 1-bit screened image pixel becomes white/paper. By this method, most of the available detail is preserved. If you have a 2400 ppi input image that is white with a thin single-pixel black diagonal line rendered at 2400 dpi (even if the lpi were only 85), the final screened image will match the source image regardless of the screening used, because every black pixel in the thin line will "win" every test against the screen, and every white pixel won't. If the line is a 90% tint instead of solid, it will "win" about 90% of the time, revealing a thin slice of a 90% flood. This is very analogous to an analog process where a photographic image was projected onto narrow-threshold, high-contrast plates/film through a contone film screen with a blurry checkerboard appearance. The screen caused a repetitive modulation in the amount of light necessary to hit the burn threshold and cause the plate/film to show ink versus paper.

This is a common misconception that is probably responsible for the disparate formulas I have seen for determining necessary resolution based on lpi. There is not a strong correlation between lpi and sharpness/detail. Look closely at your raster data after ripping a high-resolution image with lots of contrast and detail - the dots that don't look anything like circles or squares and the little bits of black and white detritus floating about is the image detail surviving through the screening process. If there was a significant dependent relationship between 300 ppi and 150 lpi, consider that this would only hold for a zero degree line screen (typically yellow), because the dot centers of 150 lpi at 45 degrees would repeat in either dimension every (square root of 2)/150 inches. For CMYK images, you would have a different ideal for yellow, black and cyan/magenta.

Definitely, Photoshop is not a text layout software, and good designers work with vector tools.

You got an amen on that one, brother. Photoshop is in my opinion simultaneously one of the best software programs ever developed - and one that a whole lot of people should use a hell of a lot less often. I particularly enjoy perfect-bound covers with 13/32″ spines that are set up as Photoshop documents at some random size that bears no correlation to the quoted dimensions. You can't even make a perfect eighth of an inch at 300 ppi, and increasing the resolution by a multiple of 300 to make a perfect thirty-second requires 2400 ppi!


I hope I won the longest post prize.

 

[Joe]

I hope I won the longest post prize.

Ya got my vote and it was spot on too.

 

[Lukas Engqvist]

Because of the large amounts of transparency effects, we sometime need to trap image edges, and image to image. Setting trapping to treat image as vector. Yes APPE and InRIP trapping, I'll post a screendump of a test sample if you like.

 

[claude72]

Unless the behavior has changed with CS4, when Indesign interprets a PSD file, it uses the text layer's pixel data instead of rendering the text in the font referenced by the file, yielding completely rasterized type.

I don't have a CS4 to do the test...

... but effectively with CS3, and lower, InDesign is not able to use the vector text datas of a PSD and has to use only the pixels.
(really, what a shame... Adobe softwares are even not able to use Adobe's datas )

When Photoshop prints, it uses the same clipping-text method in the Postscript as it does when it saves as a PDF.

I never examined how a PostScript file (made to be sent to a RIP) exactly handles the vector text datas, but I often image PSD layout containing vector text directly from Photoshop to my RIP (because it's often easier and safier than using an intermediate file format like EPS or PDF) and I surely got vector text.

Concerning the trapping possibilities, infortunaly my RIPs are old RIPs and don't have any in-RIP trapping feature...

... but when I image a Photoshop layout, I focus on outputting vector text with real black with overprinting, adding bleeds, replacing JPEG hazy logos by vector logos and correcting manually all the spelling and typography errors...

... and trapping becomes then a secondary issue!!!

I guess there's a chance it would perform otherwise when printing to a non-Postscript printer, but I would bet against it.

I cannot confirm or not... I don't have non-PostScript printer... Sorry, in my previous post I forgot that some printers are not PostScript... and when I said "print/image directly from Photoshop", I was only talking about printing on a PostScript printer...
(imaging being of course always done with a PostScript RIP)

A normal RIP does not resample images to fit a screen ruling - there is no need
(...)

Thanks very much for the explanation... sincerely, I believed that it worked the way you say it doesn't work... sorry for my mistake.

In fact, it works like making a grayscales negative film from a paper-photo with an old camera and a "gray" screen?

By this method, most of the available detail is preserved. If you have a 2400 ppi input image that is white with a thin single-pixel black diagonal line rendered at 2400 dpi (even if the lpi were only 85), the final screened image will match the source image regardless of the screening used, because every black pixel in the thin line will "win" every test against the screen

OK, I'll do the test.

But if you have a letter using 16 by 16 pixels at 2400 dpi it can be read... hazy, but readable...
... but if you image this picture at 150 lpi, the 16 x 16 pixels of the letter will become ONE screen dot, and all the detials of the letter are lost!!!

You can't even make a perfect eighth of an inch at 300 ppi, and increasing the resolution by a multiple of 300 to make a perfect thirty-second requires 2400 ppi!

Althought you are using the same units!!! OK, an eighth of an inch doesn't match at 300 ppi, but some other dimensions can match, like tenth (which is exactly 30 pixels) or third (exactly 100 pixels)...
... but imagine what happens when using millimeters!!!

 

[almaink]

I got a new one here. Text in Photoshop CS4 psd file is a "Smart" object and edits in Illustrator? Seems like a dumb workflow. I couldn't figure out how to image the file and keep the text vector from Photoshop. No matter what I did it rasterized. It wasn't a lot of text so I saved each "edit" in Illustrator as an ai and placed it back together, along with the images in Indesign. If this is the direction Adobe is heading, I'm gonna bail...

 

[Lukas Engqvist]

Slightly OT i guess...

Well photoshop seems to be developping the wannabe skills, 3D, animation etc… there is no glamour in having better "live-GCR" a propper view of TIC/TAC and true vector text.
If I had to choose Vector text is the least important as that can be done in InDesign. I guess GCR and TAC can be analysed in InDesign but would be good to be able to analyse in the same program that you edit, so you know what you are passing on

 

[hansman]

I got a new one here. Text in Photoshop CS4 psd file is a "Smart" object and edits in Illustrator? Seems like a dumb workflow. I couldn't figure out how to image the file and keep the text vector from Photoshop. No matter what I did it rasterized. It wasn't a lot of text so I saved each "edit" in Illustrator as an ai and placed it back together, along with the images in Indesign. If this is the direction Adobe is heading, I'm gonna bail...

did you try & save the .psd as a photoshop .eps with vector checked off?

 

[almaink]

I tried saving it as a Photoshop PDF. I thought eps's were a dead format? I haven't saved as an eps in years now...

 

[pacart]

No EPS isn't dead, it is a good format to use, i prefer it over pdf. I don't prefer photoshop eps files though.

The problem with pdf is that so many programs can make a pdf and they all do it differently, i just worked on a pdf that was a photoshop pdf file, it was only text, but when i put it into pitstop and viewed it in the wireframe mode, the entire page when black, totally uneditable of course you say open it up in photoshop, well i didn't have the fonts to do the edits i needed as they were embedded subsets. So i had to have the designer send a new file and make the change on their end....

The way i deal with text in photoshop is to make it a 600dpi bitmap file, then place it into the document in indesign. Works everytime, looks smooth like vector. I should say prints smooth like vector, looks like crap on your screen sometimes, FYI.