Designing, CMYK or RGB?

[sjm]

You mean the combination of RGB primaries using additive colour?

Like say in a projector (monitors and TV don't use real additive color) or a Lightjet printer that uses RGB lasers to image photographic medias?

Or are you being a smarty pants?

I thought you were speaking about CMYK printers (using the subtractive model) that receive RGB data (using the additive model). That's what I was interested in learning more about.

Can you help?

 

[Rooster]

Sure try these resources if you're unclear on what's being discussed.

http://en.wikipedia.org/wiki/Additive_color
http://en.wikipedia.org/wiki/Subtractive_color
http://en.wikipedia.org/wiki/CIE_1931_color_space
http://en.wikipedia.org/wiki/Lab_color_space
http://en.wikipedia.org/wiki/CMYK_color_model
http://en.wikipedia.org/wiki/Color_theory
http://en.wikipedia.org/wiki/Grey_component_replacement
http://en.wikipedia.org/wiki/Under_color_removal
http://en.wikipedia.org/wiki/Under_color_addition
http://en.wikipedia.org/wiki/Rich_black
http://en.wikipedia.org/wiki/Raster_Image_Processor
http://en.wikipedia.org/wiki/Color_management
http://en.wikipedia.org/wiki/Color_profile
http://en.wikipedia.org/wiki/Gamut

Once you understand the basics you're able to ask more specific questions about the process. Check out the links above and it will at the least help you get more out of discussions like this.

Personally I learned color management principals and theory by reading the manual that came with my first color management application (candela color synergy) from cover to cover as many times as I needed to to "get" it. Then it was a matter of putting it all into practice and playing around a bit. It can be a very dry and boring subject at times, but the rewards for learning it are well worth the effort. At least today the information is more readily available to learn this stuff. When I taught myself we weren't even using icc profiles. None of the OS's would recognize them. Apple had colorsync, but our RIP used a precursor to icc profiles called a Candela Color Circuit that did essentially the same thing.

The above links will just scratch the surface, but in reality most people today don't need to be an expert to use it as the tools have become so easy to use. You do need to understand the basics for it all to make sense eventually though. If you can use the software and understand the basic work flow of ink-limiting, linearization and profiling and the different types of rendering intents and how they'll effect color then it all becomes automatic in the RIP. You just need enough knowledge to know what options to select for the job or work flow.

 

[sjm]

Thank you for that. I understand the theory behind it all, but was hoping to get some practical tips.

Someone else recommended me to this site

http://www.digitaldog.net/

 

[Rooster]

What do you mean by practical tips? You're very vague about what you're after.

 

[ProColorGraphics]

If you use VersaWorks it says to use RGB for raster and CMYK for Vector. The rip actually converts the image to the rgb to cmyk, so if you have it in cmyk, it converts it to rgb then back to cmyk, so you can have really messed up colors from the multiple conversions. That is what I was reading the manual anyways. So I do RGB for rasters and CMYK for vectors.

 

[Rustyrussell]

Rgb

I design and print in RGB it almost always works out better.

 

[eye4clr]

If you use VersaWorks it says to use RGB for raster and CMYK for Vector. The rip actually converts the image to the rgb to cmyk, so if you have it in cmyk, it converts it to rgb then back to cmyk, so you can have really messed up colors from the multiple conversions. That is what I was reading the manual anyways. So I do RGB for rasters and CMYK for vectors.

I'm astonished there is so much bizarre misinformation in this day and age; especially from a manual.

If your color management is setup conventionally, all incoming color gets converted from whatever the color is in the file to LAB then to CMYK for the media profile. So, to design in RGB, CMYK, or Pantone doesn't change if your print workflow converts or not - it always converts - and this should be a good thing.

Assuming your output profile does a decent to good job of giving you accurate output, RGB should give you a little bit better color than CMYK. The reality of it is the differences should be very small assuming you're comparing sRGB to US Web Coated SWOP CMYK.

There are times it is helpful to not convert in your RIP. Maybe you just want direct control over the CMYK numbers or to mix a leaner black. No matter the reasoning, you must have CMYK in your file in this case because there is no means to properly convert RGB with the conversion function turned off in the RIP. For most folks, this should be an unusual workflow and color conversions should be turned on most of the time if not all the time.

 

[luggnut]

assuming you're comparing sRGB to US Web Coated SWOP CMYK.

what about abobe rgb 1998 will there be a larger difference in the achievable color?

 

[eye4clr]

what about abobe rgb 1998 will there be a larger difference in the achievable color?

Absolutely. Thing is, unless the file is created in Adobe RGB there is no advantage. IOW, if you have an sRGB or CMYK file and convert it to Adobe RGB, the color should be exactly the same as it was in the originating color space.

So, if you design in Adobe RGB or convert images from your scanner's profile to Adobe RGB then you're ahead of the game. Otherwise, if you just taking in customer files from the random public, sRGB will serve you best since 99.99999% of those files that come in RGB will be sRGB.

 

[luggnut]

i design in house with adobe rgb 1998 but i assume file sent unless embed with adobe are sRGB...

IOW, if you have an sRGB or CMYK file and convert it to Adobe RGB, the color should be exactly the same as it was in the originating color space.

i thought this might be what was happening with insignia's post about trying different spaces (cmyk vs. RGB) and getting the same color or very close across the board. just converting color space should change the color right?

so insignia would need to create the files in sRGB, CMYK and probably adobe rgb to really see how his color is working ?

 

[eye4clr]

i design in house with adobe rgb 1998 but i assume file sent unless embed with adobe are sRGB

Totally viable as long as you have the awareness to juggle the different files.

i thought this might be what was happening with insignia's post about trying different spaces (cmyk vs. RGB) and getting the same color or very close across the board. just converting color space should change the color right?

Yup.

so insignia would need to create the files in sRGB, CMYK and probably adobe rgb to really see how his color is working ?

yes. Or start with a true Pantone chart that is Pantone named colors, not CMYK values. This vector chart could be rasterized with Photoshop directly into whatever space you want to test.

Or, and this may be a bit out there if you don't fully understand the difference between converting and assigning, have some simple RGB solid colors like 255/0/0, 0/255/0, and 0/0/255. Assign different RGB spaces to it and print them. You'll see the max potential for each space within the potential of your ink/media/setting combination.

When i've done this for classes it has always been surprising how little a difference it really makes. My experience in production leads me to put far more emphasis on consistency in the print systems than fussing over multiple working spaces. In fact, i convert incoming profile mismatches to either sRGB or SWOP to keep things simple in production. With the exception of fine art. Then i fuss.

 

[Letterbox Mike]

Or start with a true Pantone chart that is Pantone named colors, not CMYK values. This vector chart could be rasterized with Photoshop directly into whatever space you want to test.

For what it's worth this is what I did. All PMS spot colors, none were converted to process color.