XPS Document with PNG from CMYK

[shueardm]

Hi all,
Looks like a great community here.

My question relates to the Windows XPS print driver.
I set up the driver to use png lossless compression and I printed to XPS from a TIFF file. The results on my test print were as expected, I am very happy with it.
Then I realized something, the TIFF file is a CMYK image but PNG does not support CMYK.
Firstly, why did the print driver allow this to be created without warning and secondly, is the image in the XPS now RGB or is there some funny business going on?
Thanks very much.

 

[curveto]

I expect (since you didn't indicate knowledge of CM or any interaction with ICC profiles) a DeviceCMYK to sRGB transform occurred. Go read this...

https://en.wikipedia.org/wiki/Color_management

Windows (like OS X) has support for color management built in. Much of the time it is not configured correctly but it's there nonetheless. After you read that article, go check out the "Color Management" control panel (Start menu -> Control Panel -> Color Management)

You may also take a peek at the following. It may help you wrap your head around color management some more:

http://www.gballard.net/psd/go_live_page_profile/embeddedJPEGprofiles.html

 

[shueardm]

Thank you,

So if I wish to keep CMYK without correction I should use PDF printer instead.

 

[curveto]

Thank you,

So if I wish to keep CMYK without correction I should use PDF printer instead.

I suppose, yes. Technically, the answer is, "maybe."

...but now you have me wondering what you're trying to accomplish.

Why not just leave it as a CMYK TIFF? (if that's what you have) Or, if you have access to one of the various Adobe products -- I suppose Photoshop would be the most useful (here) -- you may find it easier to package your materials up into a PDF via import/export (there). Of course, Photoshop (like many tools) is *still* going to pump the material through a 4->3->4 (technically N) transform process. Most wouldn't consider that to be a "bad thing" (per se).

Why do/did you think RGB (effectively compressed to a CMYK gamut ... as with your PNG workflow) is a bad thing?

I'm sure there are waxy/watery minds (lurking about) that would like to know...

 

[shueardm]

Hi.
Well, the more I read and try to understand the color profile settings the more I get confused. I do have Photoshop, Illustrator etc, no issue there.
What I was trying to accomplish is to save file size, the TIFF is 16MB where the XPS with PNG is only 505KB. It may stil be practical for my purpose but would require a total understanding of color spaces and ICC profiles that I am still not sure about.

Here's what my purpose is.
The only proof method I am concerned about is a TEAC Dye Sublimation CD printer with CMYK ribbons. That's what I want to match to. There is a color package installed by the printer but it is not a select-able ICC profile (like I know the difference).

I want to accept a print ready file format (prefer a low file size as possible) that I can just print to the TEAC printer without worrying about if the color is what the customer expected.
So , I wish to write a document to explain how they should setup an image in Photoshop and save to (XPS, PDF). Problem is I even doubt I know for sure myself. For instance, do all images must start with working CMYK in Photoshop or RGB and then convert them.
Images are in RGB, printing is to CMYK, how that manifests is still confusing.

 

[curveto]

Ahh...

Three things...

1) You probably should just open the original image in Photoshop and print directly to the TEAC from there. ...treating it as a GDI printer. More to the point... using whatever print driver TEAC gave you for properly feeding their device. You can configure Photoshop device profiling later (let's not complicate things just yet).

TIP: Color management is actually a VERY simple concept. Trouble is it has been mucked up by big words and "science talk" for far too long so people thing there is some big giant mystery to it. It's one of those topics that -- like the heater / cooling controls in an old German car (I'm picturing the controls of an old Porsche 914/911 for those following along) -- will have you scratching your head for far to long and when you finally "get it" you'll wonder what the big deal was.

2) You're thinking to hard. Have your customers give you RGB material (if that's what they have). You want to keep the material in as large a color space as possible prior to your final transfer to the TEAC. Don't get caught up in the misguided impression that you need to transform things to CMYK yourself before hand / upstream.

Folks that don't know color management from Page 8 will invariably build and pass around material made in sRGB. Some that think they understand color management (but don't) will "help you" by giving you Adobe RGB (bad user, bad user). Go with those. Just make sure the customer tells you which (if they don't know the answer to the question assume they gave you sRGB). Let Photoshop transform it from sRGB (or Adobe RGB) to your TEAC's CMYK (or just uncalibrated/uncharacterized CMYK) during the (direct) print process.

You may want to verify that you are correctly telling Photoshop that any unprofiled source material is sRGB. I *think* you do that in Preferences but someone else (here) will know better than I (it's been years since I cared to look as I develop software these days vs. certify with Adobe's bits).

3) Both TIFF and PNG are raster image formats. PNG is three channel (as you mentioned ... technically three plus an optional fourth alpha channel but...) whereas CMYK TIFF is 4 channels. That said, the difference in image size shouldn't be too much more than 3/4ths the size (if you're following me) plus effects from compression. The TIFF probably had no compression whatsoever whereas the PNG probably did. Nevertheless, 505K vs. 16M is waaaay to big a difference to account for those factors alone. The XPS driver *also* (must have) applied image resolution reduction. It knows how much data / resolution your TEAC really needs to do it's job; which for a dye sub system could be radically less than most people think.

PS - Just then JR hears the image density Jihadists firing up their helicopters and polishing their gattling guns for yet another thread suitable for a remake of a 1989 issue of Audiophile magazine (think analog vs. digital or unobtainium cable A is better than unobtainium cable B).

 

[shueardm]

curveto, I am grateful for your time.

I understand the concept of RGB hold more gaumet so keep as much as you can until it gets proofed. However I am more concerned about the perception from the customer that the image printed to disc is not as vibrant as on the monitor (having to explain this everytime) and getting disappointing reactions. I would much prefer to explain up front to use this method to proof your own color (as best you can).
Furthermore, there is more to a CD label than just an image, there will more often be text, logos and maybe even artwork. I don't want them being created in RGB for the same reasons.

So my thoughts are to explain they should
1) set up CMYK working space.
2) Import/Place images in (convert to working color CMYK space automatically)
3) Proof to minitor
4) Export PDF (as XPS would convert to RGB)

By the way, I have no idea why the TIFF when converted to XPS was such a low file size in comparison but I know it was set to PNG lossless. Regardless, the image ot CD print was great.

 

[curveto]

First things first... is this the printer you refer to?

Teac P55 DVD CD Printer

That printer isn't going to want you to send it (what you think of as) CMYK data (trust me, it isn't ... even if you think it is). What you want to do is feed it sRGB material (technically, using the "Perceptual" rendering intent after you get things setup to a "profiled" configuration).

I did a little TEAC site search and was unable to locate a factory device profile for that machine (as a starting point) so you're going to have to build a custom proofing profile (or, more to the point, find one that is "close enough") and treat the machine as a generic sRGB output device. The nature of the printer is going to make it a little challenging / expensive to characterize the device but here's how I'd recommend you go about it...

What I'm about to describe isn't perfect but it will get you close enough to where you want to be that you can iterate / repeat the process working your way toward an acceptable solution. I don't suppose the machine has any mechanism for linearizing (i.e., calibration) so for the purposes of this thread I'll just assume it will remain "stable" over time. It need not be calibrated per se, but it MUST remain stable / constant in the nature of it's ink deposition (over time) for things to work reliably.

0) Before you start, go read the link below. It's old but will help you understand how to configure Photoshop for proofing in an environment like yours. Keep in mind that you don't (yet) have a known good device profile (you're going to locate a "proxy" for one and use *that*).
1) Open Photoshop and make a 12cm, 400dpi (1890 x 1890 pixel) canvas;
2) Make a layer and paint a 100% black, filled circle (it will look like a black CD, if you're following me);
3) Duplicate that layer (and its content) 9 times;
4) Make each (successive) layer's circle 1cm smaller than the previous one and stack the layers so that each smaller circle lays atop its next largest sibling. If they weren't all black they'd resemble a "cone" when you're done;
5) Save that file (it will become a *template* that you'll want to keep around / re-start things from);

NOTE: The image you are building doesn't look like a rainbow yet but it will.

6) Open a *copy* of the template you made and re-fill each circle with a different color. I would start by picking a Cyan, Magenta, Yellow, a high and low saturation Red, a high and low saturation Blue, and a high and low saturation Green. You're going to want to do this over and over using different colors each time.;
7) Use an output device profile (it should be built into Photoshop) that treats the TEAC as an sRGB device;
8) Print the test file (you should get a nice rainbow colored CD);

9) Use a tool to profile your display (there are lots of pages around the web that will describe how to do this so I won't bother here). You may (or probably should) already have a suitable soft proofing display already set up;

NOTE: Your color critical customers are ALSO going to want to have a proofing display that is properly profiled. If those two things aren't in place you CANNOT reliably tell them what they're output is going to look like without printing a sample for them. NO process you outline for them will help if they don't have a profiled environment.

10) Follow (the intent of) that article (noted above) and select a ICC profile for proofing that best matches the output you got from the TEAC (when viewing it's basis image in Photoshop). You should be able to find something that gets you pretty close.

NOTE: From time to time you may want to make a new sample CD with reference colors taken from customer provided material so you can see how their (various) colors match your proofing profile, etc.

I hope this makes sense. It's kind of hard to describe via this medium.

As for source material from your customers. They should give you sRGB (or properly tagged (i.e., having a correct, embedded ICC profile) RGB material). I would recommend they give you PSD files or (high quality) JPEGs (of the same size as your testing image). TIFFs (and PNGs) are useful too but it won't really matter that much as long as your source material is RGB (or RGB /w Alpha) and properly tagged/described. Some content may best be provided as LAB (stuff with lots of neutrals) but describing that will make this thread too long (so I'll just mention it).

DO NOT do what you thought was correct (i.e., manually convert to some unknown CMYK upstream). Ever...

Whoops: I forgot this link... http://www.computer-darkroom.com/softproof/softproof_1.htm

 

[shueardm]

I went missing, I am so sorry, that was rude.
I really appreciate all the effort you put into me. Just so you know, your advise did help me, the never CMYK think is correct, it started me on the right track.
I am well advanced now in my understanding of color management. The XPS question still has me wondering about it's color profile handling though I am there with Adobe stuff I think.

 

[curveto]

I went missing, I am so sorry, that was rude.
I really appreciate all the effort you put into me. Just so you know, your advise did help me, the never CMYK think is correct, it started me on the right track.
I am well advanced now in my understanding of color management. The XPS question still has me wondering about it's color profile handling though I am there with Adobe stuff I think.

Your XPS output uses the Windows color management system. For more info on that read this

Color management settings: frequently asked questions

 

[shueardm]

thanks. Nothing specific to XPS. My real question is does it support embedded profiles or are they thrown away?

 

[Stephen Marsh]

When faced with such unknowns, I would do as much upstream as possible. So I would then assign a CMYK device profile to the CMYK data and then convert it to sRGB - before going to XPS or the printer.

If you know what you are doing with your Windows CM settings, then perhaps setup assumed CMYK profiles to use for 4C input.


Stephen Marsh

 

[shueardm]

It's ok Stephen, curveto put me straight, now we start with RGB