Seperating Artwork via Photoshop Channels

[ericheartsu]

Since we added an I-Image to our shop, we need to learn how to separate art via channels, as opposed to layers.

Since i Know nothing about art seps, and my wife handles it all is usually to busy to be a chatty cathy on the message board, is there anywhere we can read about or lean how to do it? Or would any of you captains of industry be able to explain the best way to do it?

thanks in advance!

[Frog]

I was first taught this by using the Color Range tool. I think this first article covers that, but I included a couple of others as well.

http://layersmagazine.com/spot-color-separations-in-photoshop-cs5.html

http://planetphotoshop.com/working-with-spot-color-channels.html

http://murakamiscreen.com/wp-content/uploads/2012/07/Creating-Spot-Color-Channels.pdf

[tpitman]

For more complex images, there's Mitch Different's book recently mentioned elsewhere on these boards:
http://mitchdifferent.com/book.htm

I purchased the .pdf version and find it very informative.

[Rockers]

The problem I always have with channel seps  from Photoshop is that if I have to trap stuff it will be at least by 1 point as that is the minimum given by photoshop. Unless there is a way around it, I`m just not very familiar with PS at all to be honest.

[Full-SpectrumSeparator]

     I have some new videos on my youtube channel that show manual methods to creating color-separations in Photoshop through layers or channels, halftone RIP and even creating your own actions to repeat the manual methods that are shown in the videos.

    Photoshop channels cannot accurately preview "blending" color-separations no matter how "good" someone is at it, that is just a fact.   Channels can help make it easier to output to a print device, and they can work good for spot-colors and when you halftone the channels they can sometimes preview a little more accurately but still have major flaws behind how the program interprets the math of blending the colors in a channel.. it is just the nature of the problems with Photoshop in Greyscale, CMYK, Duotone, or Multi-channel mode, it cannot be made to work correctly within those modes - but that is a deeper subject on its own.

    Layers can be made to accurately preview the color-separations but it can be more technically complex to work with the separations that way, and most output-methods won't take all the separations that way at once, the way it works when they are in channels.  I usually just proceed direct from separation to halftone RIP with automated methods, perhaps choke or trap, merge some colors, and just output to film and use quality-control and other techniques built into the sep/halftone/print method that gives me confidence and repeatability with automated color-separation and halftone conversion.  The output is dependent on quality-control between the separation/rip/stencil-making and all of it has to work together.   Previewing in a color workspace that literally makes things appear the opposite to what you think you will get, is the last thing you would want in a quality-control color-calibrated printing environment.   

    Color-range tool cannot accurately extract the gradients needed for full-blending work, but can be somewhat effective at times for basic spot-color separation, although it has too many handicaps compared to other techniques when it comes to custom-color separation gradients, so I just stay away from it altogether.

     I will continue to share new videos on my channel that show the manual techniques behind my methods in color-separation and halftone rip, film output and testing, print calibration, screenmaking and printing as I get into the various steps and show examples on my youtube channel.   For now here is an action and step-by-step tutorial I shared recently elsewhere in the forum that helps go directly from image to halftone RIP color-separation ready to output to film or to your computer-to-screen device.  The free action and text-tutorial (Photoshop 7 compatible) is meant for photographic or complex blending artwork typically called "simulated process" - where you print with halftones blending of just a few screens to make all the colors in the image.   This converts the image directly from RGB original to Halftone RIP channels you can just output to your film device the ones that have dots in them for the image.    It contains the process CMYK mode, and the basic Hexachromatic (6-hues) Simulated Process, which has Red, Yellow, Green, Light Blue (Cyan), Blue, Magenta, and a black, white and Underbase white.   

   

   

     

[Full-SpectrumSeparator]

The problem I always have with channel seps  from Photoshop is that if I have to trap stuff it will be at least by 1 point as that is the minimum given by photoshop. Unless there is a way around it, I`m just not very familiar with PS at all to be honest.

Rockers if you increase the resolution you can increase how to choke or trap by other amounts.

The basic rule is 300 DPI = 1 Pixel = 1 point

But if you really want a 0.5 point choke or trap (what happens usually an object in illustrator - it chokes or traps by only half of the stroke so 0.5 point for a 1-point centerline stroke) --  so with Photoshop or raster images you would increase the main resolution with (Nearest Neighbor) upsampling - and this will change what 1-pixel is equal to.

So if you are at 600 DPI then 1 Pixel = 0.5 point

1200 DPI - 1-pixel = 0.25 point

Often I have to use this rule when I am working on high-resolution 1200 DPI  halftone RIP channels, choking or trapping the dots I will just use 1-pixel for fine-tuning,  but for shapes where I need to choke an underbase perhaps by 1-point - then I will need to choke by 4 pixels at 1200 dpi to equal 1-point total.       

So does that make sense,  resolution DPI makes a pixel equal different sizes in terms of a point-measurement.   300 DPI / 1-pixel = 1 point,  -- increasing resolution makes the pixels smaller - so if you double the resolution then you make a single pixel only equal to half what it was before... a 1-pixel line will be two-pixels,  still the same thickness in actual dimensions, but if it was 1-point before and you wanted to choke it 0.5 a point, you could not do it at 300 dpi (it would just cut in by the whole width of the line, 1 pixel) -- but if you increase to 600 dpi, then you have 2 pixels for that line, and you can choke in by 1-pixel and it actually equals the 0.5 point you might want.     

So there is in fact no real "minimum" that photoshop can trap or choke, it depends on the resolution you are at with the image/channels. 

[ericheartsu]

woah! this is so helpful! thanks everyone!

[starchild]

Photoshop channels cannot accurately preview "blending" color-separations no matter how "good" someone is at it, that is just a fact.   Channels can help make it easier to output to a print device, and they can work good for spot-colors and when you halftone the channels they can sometimes preview a little more accurately but still have major flaws behind how the program interprets the math of blending the colors in a channel.. it is just the nature of the problems with Photoshop in Greyscale, CMYK, Duotone, or Multi-channel mode, it cannot be made to work correctly within those modes - but that is a deeper subject on its own.

I actually work around the channel preview situation using Mark Coudray color settings preset and I get almost accurate previews.

Now about the Hexochromatic seps.. I've been using certain steps to do a 12 hues or custom color pull and will appreciate your opinion on it..

So using your color wheel pattern, lets say I want only the orange on a screen..

1. I use the selective color tool and remove all the blacks, neutrals and whites.

2. Next I use the Hue/Saturation slider to target the orange and push the lightness to -100% so the orange values is 100% black..

3. I then duplicate the layer- one to get rid of the remaining hues of the orange screen (by pushing each of the six hues black sliders to -100% black) back inside the selective color tool.. And the other duplicated layer to continue pulling the other remaining hues I push the black slider -100% to get rid of the orange values that is now black.

3. Now that I've got both my orange pull and the duplicated layer with my orange values subtracted I can continue duplicating this channel for the six hues using the selective color tool.

So the thing is if I have more than one custom color to pull or I want to pull 12 hues rather than 6 hues, I will have to do steps 1-3 for every color.. Writing an action for this will take some thinking plus I'll have to launch the Hue/Saturation tool for every secondary or custom hue pull because I cannot come up with another way to target those hues.. say with an I dropper..

Sent using Tapatalk

[Sbrem]

The problem I always have with channel seps  from Photoshop is that if I have to trap stuff it will be at least by 1 point as that is the minimum given by photoshop. Unless there is a way around it, I`m just not very familiar with PS at all to be honest.

Rockers if you increase the resolution you can increase how to choke or trap by other amounts.

The basic rule is 300 DPI = 1 Pixel = 1 point

Maybe I'm missing something, here, but I always thought that one point is 1/72 of an inch, so therefore, at 300 ppi, 1 pixel equals 1 point would not add up; one would need just under 3 pixels to equal a point. Great post as always...
Increasing the resolution to have better control of the traps is how we handle it too. If you're posting vids, how about the HSB one you took down? ?

Steve

Steve

[Full-SpectrumSeparator]

Yeah you're right about that,  perhaps I was carrying over the idea that 1-pixel = 1-point for a resolution of 72 DPI, and that is exactly what it is.

So for correcting the math,  it is really the 72 dpi resolution (point being 1/72 of an inch) - where 1 pixel is 1 point.

I checked it out and indeed drawing a point line in photoshop at 300 dpi = 4 pixels basically.  (72 x 4 = 288)   

Perhaps if we are trying to choke in by 0.5 a point, then at 300 dpi it would really require 2 pixels,  4 for a whole point.       

But this shows you can actually get access even at 300 dpi, to using chokes or traps of 0.25 points per pixel increments.

So really it is like this...   

1 point @ 72 DPI resolution = 1 Pixel
1 point @ 144 DPI resolution = 2 Pixels (also for 150 DPI)
1 point @ 288 DPI resolution = 4 Pixels (also for 300 DPI)
1 point @ 576 DPI resolution = 8 Pixels (also for 600 DPI)
1 point @ 1152 DPI resolution = 16 Pixels (also for 1200 DPI)


Thanks for catching that, with the 1 point standard of 1/72 of an inch, then the 1-pixel rule starts at 72 dpi, and doubling from there.   
Good to know.... been using 1-pixel 300 dpi choking like on underbases forever, and really those are about 1/4 of a point, still helps and works great, it is tighter registration than using a whole point choke, but this could be difficult for some, so it would really need about 4-pixels choking at roughly 300 dpi to get a full 1-point choke (most 1-point chokes are strokes on vector objects and are centerlined, so only a half-point ends up cutting into the object, that would be 2 pixels at 300 dpi.)

--- as for the videos,  the old HSB videos are outdated and were incorrect in some ways as well.   It is HWB - Hue, Whiteness, Blackness, and I have only recently finished up years of research into this and realized I've been separating and building the automations that are actually HWB the whole time and was calling it HSB because of using the HSB/HSL tools to get to the results.    The Highlight white is the whiteness, not saturation which is a different gradient entirely, and the brightness is converted to blackness... so it is full HWB color separation, and HWB is simply the best color model on the planet.

I will do a quick new video about the manual separation with the HSL tool, while it is more versatile than selective color for pulling hues, it is more difficult to understand how to set the levels for the hues appropriately and extract each one.    The Black & White effect and the Selective Color effect are a bit easier to use in getting the separations so that is why I have shown those in my newer videos.    But I will do one for the HSL method, it helps show how you would get into more custom pulls of the hues like Starchild asked about. 

One thing Starchild said:  "I actually work around the channel preview situation using Mark Coudray color settings preset and I get almost accurate previews."

It doesn't fix it at all...there is no way to work around it.  It is not easy to explain but very easy to show, so I will get to that in a video as well.   

Also:  "1. I use the selective color tool and remove all the blacks, neutrals and whites." --- This does not remove them right.   It leaves in the shaded parts of the hues and grey-mixtures... so it is not the correct way to remove the blacks/neutrals/whites to prepare for a hue-extraction using HSL.   You have to remove the blacks with the special blend-mode layer or channel-calculation combinations I show in the videos.    Unfortunately this is one of the things photoshop doesn't have a quick way with an effect tool to do properly.  Corel you can do channel-mixer with Simulated Color Management off and you can remove the blacks easily.. but Photoshop's channel-mixer uses the color mode you're in, either RGB or CMYK, and you can't do CMYK ideal conversion or preview CMYK correctly in that situation either so it doesn't work.     

I just tried out some methods, I will do a new vid for the HSL method and show:

1) How to pull the blackness and the whiteness,(black and white tool, or HSL tool) and how to pull the blacks out of the image to get to the standard hue gradient (Duplicate the original layer, do black and white mode - max white, to get the black sep extraction - then invert the layer, and set the blend mode to Linear dodge (add)

2) How to use the HSL sliders to separate out the hue -- you need to do it in two steps... 1 shifting the hue and ramp you want to -100 lightness (black) and then you repeat where you push all hues to white to remove the left-over colors mixing with the original, leaves it just neutral greyscale.    Through this you can get to orange or any hue and any gradient ramp you want of the hue.     It should make for a rather quick video hopefully under 10 minutes that shows all those basics and just how they are different from the selective-color method but still get the same results.    Color-range tool has built-in 6-hues and white/black, but it does not pull them correctly at all, the test-pattern shows why so I might just compare the various methods and such.

Attached are some newer color-wheel test patterns I've set up using the HWB color model, with the hue-blocks split in 10% increments of White-ness/Blackness, the Hue groups are in roughly 7.5 degree increments (15 degrees on the outside, 15 on the inside, shifted 7 degrees).     

 

[Sbrem]

Wicked pissah post as we say here Mass. Great info, and yes, a bit over my head, but I know enough that if I read it a few times, it starts to get clear. I'll keep an eye open for the vids. As for Corel, I'm a Mac guy, and they really don't like Mac much for whatever reason, probably the same one I have for not liking the Windows environment.  Thanks, man.

Steve

[starchild]

Jeff you are correct that tones of neutrals and hues are left over when using the selective color tool.. I call it photoshop half ass effort because when I hit ok for the first "hue wash" I run it again and pushing the whites takes care of the left over hues and finessing (if you push to 100% then some info in the hues is lost) the neutral removes the remaining neutrals..

I would like to hear your explanation on the color settings for channel preview.. The settings Coudray recommends- for each setting he explains how it pertains to and what affect it has on the actual print production. How it's viewed in the channel was a plus.. In your recent video you double checked if the blue channel was in spot mode, I figure you won't satisfied with how it looked.. But I just did not have the same look on my end.. It looked like the original and that is when I said to myself it's because our color settings are set differently.

Hope to see more about interlocking the top colors as well..

[Dottonedan]

Eric,


If I don't do a step by step posted here, remind me later and I'll get to it.


How you get there, (what program you start in, Corel, Illy, Photoshop or what sep program, color spectrum method) is not important. That all can be hashed out by others. You want to know (the process) of saving channels so that they will come out of the I-Image. Thats more specific. I can get you there.