RGB CMYK LAB and Dan Margulis !

[mquest]

I probably wont get any sleep tonight... thanks Mark!

[MatthewMarshall]

i dont know why you guys all fuse so much about color conversions, when you take it to your print shop or give it to your clients to print, they are just going to hit the print button on there hp printers and let the software do what ever it wants with the file. lol lol lol

[pixel_monkey]

Originally Posted by MatthewMarshall

i dont know why you guys all fuse so much about color conversions, when you take it to your print shop or give it to your clients to print, they are just going to hit the print button on there hp printers and let the software do what ever it wants with the file. lol lol lol

But, this is not just color conversions. It's also a really great approach to nondestructive retouching. You have the ability to retouch at any level within the smart object layer that's within another smart object and use whatever amount of layers you want within each level and you still end up with a single layer for your master file. It sure helps my workflow enormously. I used to have to flatten all my retouching layers to just the original and the retouched version in my files and back up the masters in my hard drive before I pass them over the quality control team...it was just such a pain. Now I don't have to do that anymore.

This is not something new, though. I just realized that Deke McCelland mentioned briefly about this approach before CS2 came out. I didn't pay much attention to it back then and never thought it can be taken to this extent.

[cricket1961]

It was new for CS2, just not talked about evidently.
Be careful with the amount of smart objects you use. I had a empty base CMYK
with a RGB smart object that contained a Lab smart object. It went from 250 mbs to
1.7 gigs. No alpha channels either.
It could eat up your hard drive space rather quickly. And although it is really great, having to save a smart object each time will add an enormous amount of time to your workflow. Of course it all depends on the resolution and file size of the image.
To me, the benefits outweigh everything but the time factor. So it just means to use it when it is most beneficial to me and not any rush jobs.

Chris

[pixel_monkey]

I do notice the increase in file size, but it's ok for what I do here. What I meant by that Deke mentioned it before CS2 came out was that he was showing it on the CS2 space monkey beta version before the final retail release.

[pixel_monkey]

Chris, you brought up a good point about the increase in file size, which led me thinking about smart object instances. I haven't used them, yet. I'm assuming they should behave similar to the instances in Flash or other 3D programs out there that let you edit the look of each instance in whichever way you want as long as you don't mess with the master object and still maintain a considerably small file size . This is going to be wild if this were true.

[KR1156]

funny thing is i use the smart object for embeded illustrator files when i do production and mechanicals...helps when i need a clients logo or something, and need to apply ps effects, if i needed to modify the illustrator file, i would quickly do it and save, instead of having to copy and paste and resize and register all over again.

but i never even knew you can turn a ps layer into a smart object, i vaguely remember glancing over it somewhere, but never knew how to use it.

i'm in RGB 95% of time, and don't do the cmyk conversions, so i really love the bringing in the K plate. that's where i get most use out of it.

[pixel_monkey]

It's amazing how unforeseen possibilities start to surface when we look at the same thing from several different perspectives. I've been possessing my raw files into smart objects so I can reopen them in camera raw whenever I wish. They're basically tiff files with just a smart object layer inside, but it never occurred to me that it can go a lot further from there.

[CaptainHook]

Originally Posted by KR1156

but i never even knew you can turn a ps layer into a smart object, i vaguely remember glancing over it somewhere, but never knew how to use it.

Funny that i thought it was ONLY for ps layers.

[Markzebra]

mquest - try counting sheep, its more effective at getting you to sleep

I had a empty base CMYK with a RGB smart object that contained a Lab smart object. It went from 250 mbs to 1.7 gigs. No alpha channels either. yes this is true - i think its because Photoshop has to generate a full resolution preview of the embedded object in the master document, if you have a CMYK master with an RGB Smart Object there are now three copies of the base pixels: the CMYK base pixels, the RGB smart Object itself ...AND a preview of the RGB smart object converted back to CMYK which sits in the original file. Chris in your 3 level embedded version there is alot of extra data being generated to create the previews. Its also worth mentioning that your LAB object by the time it is reflected in you CMYK master has been converted twice.

Instances - yes I suppose the equivalent in Illustrator is what they call Symbols. This idea is actually more powerful in photoshop than in other software for this reason - now with smart filtering you can apply distortion filtering as well as warp, you can group multiple adjustments/curves, apply opacity, blending - all this individual to each instance. Its capable of enormous subtlety. As an example I have used them to coat a car bodywork and recently as a pitch to do skin where a potentially variable tattoo design had to be imposed on her. In practice each instance can have different perspective and distortion blending and opacity masking. Its magic when you see your entire image change before your eyes on updating the Smart object, or replacing it completely.

[Hello_taipan]

Markzebra is God.
thx for the tip man !

[pixelzombie]

Originally Posted by KR1156

i'm in RGB 95% of time, and don't do the cmyk conversions, so i really love the bringing in the K plate. that's where i get most use out of it.

i'm trying to figure this out but can't seem to get a handle on it, how would i use the smart object to add K only to a RGB file?

[cricket1961]

Originally Posted by pixelzombie

i'm trying to figure this out but can't seem to get a handle on it, how would i use the smart object to add K only to a RGB file?

You convert the smart object to CMYK and adjust with curves.

Chris

[mr.aj]

Tai,

Thanks for starting this thread...I learned a ton

[pixelzombie]

Originally Posted by cricket1961

You convert the smart object to CMYK and adjust with curves.

Chris

ok, i got that far but i can get a similar result by just using selective color on the RGB file itself...

[KR1156]

Originally Posted by pixelzombie

i'm trying to figure this out but can't seem to get a handle on it, how would i use the smart object to add K only to a RGB file?

what i do is double click on the base layer of the CMYK smart object and turn all chanels off except the K, then save. when i get back in RGB, i do the necessary blending of that K plate.

[cricket1961]

Originally Posted by pixelzombie

ok, i got that far but i can get a similar result by just using selective color on the RGB file itself...

True, but you are not pushing in just black. You are pushing in an interpretation of black. At least you can dial in exactly the amount of black you want.

[pixelzombie]

i'm not sure what you mean by interpretation as i'm getting the same numbers when i read the cmyk values...

[edgework]

Originally Posted by pixelzombie

i'm not sure what you mean by interpretation as i'm getting the same numbers when i read the cmyk values...

Those are the numbers you'd get if you converted your RGB file to CMYK. But as long as it's an RGB file, your shadow detail is blended throughout the three color channels. The difference is that when you manipulate a K curve in CMYK you are dealing with only shadow information in a way that RGB and LAB cannot match. I've used the K plate as a layer mask to refine my shadow moves in RGB, but it's not quite the same. That's why Dan always says that CMYK is the best space to work with shadow detail. And nothing beats the effect of sharpening the K plate only, not possible in RGB.

[Markzebra]

i'm not sure what you mean by interpretation as i'm getting the same numbers when i read the cmyk values... - This is a useful technique - what Chris means by interpretation is a re-separation of the K channel back into RGB. Its the same as if you had duplicated your RGB document, converted the duplicate to CMYK, copied the K channel, and pasted this copy from the clipboard back into your original RGB document. If you now look at your three RGB channels you will see the K channel dropped into the R, G and B in equal amounts - because photoshop has made your copy into a new desaturated layer. Each of these channels is identical to the K channel in CMYK.

Glad everyone finds this stuff interesting and useful, Smart objects after the release of CS2 took me a year or so too for me to get my head around, because they were so new. But they are a pretty important usability breakthrough I think

[Markzebra]

Oops sorry edgework,I duplicated your answer there. Yes Its about shadow detail density. In CS3 You can of course use a Smart filter on your new K layer to sharpen it - you could for example apply a blend mode to the K layer, maybe multiply? sort of like the opposite of a shadow highlight, bottom end density control with a scrubby slider.

[pixelzombie]

definitely something to master but i can't justify doubling a file size everytime i need to add black to an RGB image, at least not at my current gig...

[transoptic]

I had two questions:

first, I like your tip edgework on sharpening the black plate. How would you do this? I tried to copy it on a new layer, run high pass, and change to overlay blending. It looks interesting that way. Is there a superior method you use?


Second question is, I'm fully aware of how each mode has its own seperate gamut, but I've never come across color shifts unless printing or uploading to web, etc. Except when switching modes with lots of layers and masks. So can someone give me reasons, or better yet examples where switching modes under normal circumstances are bad?

[edgework]

Originally Posted by transoptic

I had two questions:

first, I like your tip edgework on sharpening the black plate. How would you do this? I tried to copy it on a new layer, run high pass, and change to overlay blending. It looks interesting that way. Is there a superior method you use?

You're thinking too hard. I meant sharpen the K channel. Literally. Go to your channel's palette, activate only the k channel and sharpen it. Works particularly well for faces since there is virtually none of the textural detail in the black plate to get grunged up. Eyes and hair can take a heavy hit and always look better for it. Do this on a duped layer to blend as desired.

Second question is, I'm fully aware of how each mode has its own seperate gamut, but I've never come across color shifts unless printing or uploading to web, etc. Except when switching modes with lots of layers and masks. So can someone give me reasons, or better yet examples where switching modes under normal circumstances are bad?

Not bad, necessarily, but the reality is that CMYK is incapable of producing many of the colors you take for granted in RGB. Find a picture of a vivid blue dress with lots of variations in blues and greens for detail. You'll find that with the default conversion in Photoshop, which is Relative Colormetric, all those subtle shifts in tone will pretty much turn to mud. Blues and greens suffer particularly, as well as vivid purples, turquoise and oranges and virtually all pastels; they are simply out of the normal printing gamut for CMYK. (Reds fare better and Yellow is actually purer in CMYK.) Photoshop converts colors that are in gamut precisely, and the stuff that's outside the range gets crunched at the edges. Hence the blue mud. The other useful conversion mode is Perceptual which recreates the relationship between all the colors but which will necessarily alter all of them in order to do so. It's useful when you're not going to get the original RGB color anyway and at least want to maintain detail and a reasonable facsimile of the original intent. Other options would be to pull out the lightness channel from LAB, or one of the RGB channels that captures the needed detail, copy it before converting, then layering it in on top of the image in luminosity mode. You'll need to play around with moves like this, and you'll send your colors way south, but, again, the colors weren't there to begin with. If it's a toss-up betwen bad color and bad detail, or color that's off but which has clear detail, go with the second.

Of course, not everything goes to a printing press these days, but if you need 1000 copies of something, you'll need to navigate the mysteries of conversion. (Just remember that every beautiful image you've ever seen in a book, magazine or poster has managed to get around these limits, so all is not lost.).

[Markzebra]

Yes -everything that edgework has said is 100%. Having some sort of production experience helps a lot in knowing what a good CMYK file is, and that just converting an Adobe RGB file to PhotoshopSWOP default is a no-no. The key is the black plate. Each press varies immensely in how it handles the black ink: whats the ink limit? what kind of GCR do you need? ie how HEAVY is the black plate as it stretches into the lighter tones of the image. Images vary too, more monochromatic images can benefit from a stronger K plate which helps their detail and highly saturated images a lighter K.

By the way Margulis is a good teacher in these areas because thats his background I reckon. Edgework has given some good advice there about moves on the L channel (which can now be a smart L) and a lum blend on this, thats great technique which I use. Luminosity blend uses LAB calculations fairly non destructively without any conversions actually you can just use adjustments with a Luminosity blend with tremendous power.

[amica999]

Hi guys and girls,
I am reading my way through the LAB book and found some difficulty following the calculation instructions....it states that in the calculation screen choose green channel and it should be applied to the whole picture (all other channels if none is selected). In CS3 I do only have the option the choose red, blue, gray etc. Would that in conclusion mean that I have to dup the layer twice and apply the green channel in the first dup to the red channel, and in the second copy to the blue channel? Wouldn't a channel mixer be more appropriate? Please let me have your thoughts, as I am lost here. Thanks

[katheo]

I have been reading Margulis' color correction books and have a few questions. I hope this an appropriate place to post them; if not, please direct me to a better forum.

Q1-- CMYK corrections sent to Epson RGB printer
I am learning to color correct in CMYK. I would like to print to an Epson 1280. Epson's materials say if you edit in CMYK you should convert back to RGB before printing. Is this necessary or advisable for every print out?!? I am just learning, so I tend to make a few curve adjustments, print, adjust, repeat... How much am I losing by staying in CMYK when I send the print command from Photoshop?

Q2--Appropriate image resolution.
Margulis recommends making sure that the image has normal resolution (1.5-2 times the screen ruling at 100% magnification). My Epson 1280 has a spec of max 2880 dpi x 720 dpi. My Sony A100 camera produces images 3872 x 2592 pixels @ 240 pixels/inch, Adobe RGB 1998 @ 8 bits/channel.

I have read that file pixels/inch should equal 1.5-2 times the printer LPI (is this 720 LPI for the Epson 1280? or 320 LPI?). What should be my 'normal resolution' to make sure that I am viewing USM corrections on the screen at the right resolution?

If I want to print at 5.35 inches x 8 inches, do I need to resample my image before doing the corrections? It seems like I wouldn't want to lose any information if I don't have to.

2592 pixels x 3872 pixels to be printed at 5.35 inches x 8 inches = 484 pixels / inch resolution without any resampling. Is this advisable? What would be better to view USM adjustments ideally?

Can anyone help?

[MacBurg]

Thanks to all above for the interesting info, always put off learning and using Smart Object layers, and from what I have read here its well worth my time to put them into use. I created an action I use at the very start of my workflow, converting the image to CMYK, and then a double USM sharpen directly on the black plate. Image is then duplicated and sent back to my original RGB file and set to "Luminosity" and masked away from out of focus areas.

I find it a great little trick to sharpening an image shot on film and then high res scanned, as they come back to me with a desperate need of selective sharpening and colour correction. So I will put this Smart Object into use...

[mistermonday]

Katheo, Welcome to RetouchPro. I will try to answer some of your questions.

1. The Epson 1280 is an inkjet printer not a printing press. So firstly Line Screen is not applicable to it. Secondly, while it uses Cyan, Magenta, Yellow, and Black inks, its print drivers and assoc S/W are meant to handle a RGB image. All of the conversions factor in that it is being fed an RGB image and one of the RGB color spaces. It does not do very well if you feed it a CMYK file - sorry but that's just the way it is.

2. For most printing 1440 dpi should produce very good results on the 1280. If you print at 2880 (Best Quality) the printer is incredibly slow. While the printer resolution is 2880 your image does not and should not be 2880. That's because each dot printed is really made up of multiple sub dots. If you read your manual or check Epson's site you will see more info and explanation. Ideally 240 dpi or 300 is max you need to use. If you want to keep all of the orig pixel info and print at 484 dpi, that will work but the printer driver will just interpolate / resize anyway when it gets the data. You are better off having a number that divides evenly into 1440 or 2880 or 720 (like 240) so that you get the sharpest image (less round off).
Regards, Murray

[doctorontop]

When Chris Cox did the webinar he hinted that more things will be happening with smart objects in CS5 it will be interesting to see what Adobe come up with.