Preparing Lineart Images for Colouring... Quickly

[WhiteFox]

   This guide has one specific purpose: to explain the most effecient way way to take a grayscale or B/W line-art image, and set up so that it can be coloured. I have yet to find a more efficient or effective method for doing this.
   The guide does not cover techniques for colouring or rendering, but there are already many such guides available. Also: this is something completely different from using an Airbrush set to Multiply to shade a coloured image. It uses Multiply, but in a completely different way and for a completely different purpose.
   It should be noted that cleaning up or editing the line art isn't at all necessary for this process, but messy lineart will still look messy. With this method, however, the only editing required is for aesthetics, rather then technical limitations. Maybe you just want to colour a sketch, or you know how to make mess look good.

The tl:dr Version:

   I discovered this method pretty much by stumbling onto it. I don't know if anyone else uses it, but I've yet to see a guide describing a similar process.
   The screenshots here are taken from PhotoshopCS4 for the Mac, but I have added instructions for doing the same thing in GIMP. I imagine the same can be done in other graphic apps too.
   To start, I assume you have an image  to be coloured. It may be scanned art, or an image you found online. Or maybe you lost the original photoshop file of some old line art.

1) Open the image to be coloured. The lineart doesn't have to be B/W, or even grayscale.

This is the Layer Panel. You'll need it.
The one in GIMP is very similar.

2) Make the lineart a layer instead of a Background.

Photoshop: You can do this by right clicking on the background and selecting "Layer From Background," or from the Menu: Layer>New>Layer From Background
GIMP: Right click on an empty spot of the Layer list, or go to the Layer Menu. Select "New From Visible." This will make a layer that looks exactly the same as the Background.

3) Set the Layer Mode to "Multiply." There wont be any visible difference.

This is the Layer Mode.
It's located in nearly the same place in GIMP: Just above the list of layers, and it's labeled "Mode."

This is the drop down list for the Layer Mode, with "Multiply" highlighted.

In GIMP, "Multiply" is in exactly the same place as it is in Phothoshop.

4) Create a new layer underneath the Ink layer to put colour on.

In GIMP, make sure to select "Transparency" for the "Layer Fill Type" when making a new layer.

That's it, you're ready to go. Colour away.

If it's necessary to edit the line art, you can apply white and black with an airbrush to remove or add lines, or use an eraser to remove them. If you're using GIMP, remember that the Background looks exactly the same as the lineart; if the BG is visible, you wont see any differenc if you erase any lines.



Just What the Heck is Multiply?

   To a computer, colours are all numbers and math. Colour values (AKA: Colour Channels) are how a program keeps track of what colour a given pixel is. In RGB, this is how much red, blue, and green is in a colour. (CMYK has cyan, magenta, yellow, and black. K is for black, so it's not confused with Blue, which is not to be confused with Cyan. Just don't ask. Duotone has two tones: black, and any other single colour.)
   Transparency will take the colour vaues of two pixels, and average them together (adjusted by the amount of opacity of that pixel and/or layer); this mixes the colours. Areas of colour will get washed out with white, and the black lines will be tinted by the colours underneath them.
   Multiply, on the other hand, works by multiplying the colour values instead of averaging them. This can make some funky results when mixing colours, but in grayscale the results are pretty straightforward.
   If a pixel is white, its colour values are all 100%. 100% equals 1, and anything times one is itself, so the colour underneath a white pixel wont change at all when multiplied.
   If a pixel is black, it's colour values are all 0%. Anything times 0 is 0, so anything under a black pixel (IE; the lines) will be black.
   If a pixel is gray, the values of the colour underneath it will be reduced depending on the value of the gray; A 75%, 50%, or 25% gray will make a colour 75%, 50%, or 25% darker. This preserves the anti-aliasing of grayscale lineart.
   For example: A middle gray is on top of a bluish green. All the values for the Gray are 50%, and the blue green has 0% red, 80% blue, and 20% green. The resulting colour is 0% red, 40% blue, and 20% green.
   (When multiplying one colour on top of another, it multiplies each individual colour value separately. Since grays have the same amount of all colours, however, it affects all the channels equally.)



How does this compare to other methods?

   The one advantage to this process is that it doesn't require any editing of the lineart. It works with anything: Sketches, pencils or inks; B/W or grayscale images, and even colour scans of lineart.
   The major downside is that it's not possible to flood-fill areas of colour. However, in my experience, the time saved by not having to heavily edit the lineart more then makes up for the time required to add flat colours. Your mileage may vary, depending on your colouring methods and personal preference.

   There are three basic methods (that I've seen) that people will use to prepare lineart for colouring:

Creating An Alpha Map Using a Grayscale Image:
   The Multiply Mode uses different math to get EXACTLY THE SAME RESULTS, and takes a fraction of the time as creating an alpha map does.

Tracing The Lineart On A New Layer:
(Or drawing it digitally to begin with)
   First of all, this may not appeal to artists who simply prefer traditional media.
   While tracing and drawing can be done with a mouse, it generally requires a tablet or vector graphics to get the best results.
   Tablets are expensive. If one has a tablet, there's no reason not to use it to draw and ink, but the majority of artists don't have that option.
   Working with vectors may not appeal to artists who prefer the direct simplicity of a pen or pencil, or a digital airbrush. Vectors graphics also have a steep learning curve, and some things require exponentially more effort to do in vectors then on paper or bitmaps. Vectors are very well suited for flat colours and cel-shading, though, and their scalability makes them handy for production.

Deleting the White From a Black and White Image
   When lineart is converted to B/W, or scanned as B/W, it has a "corrosive" effect on the art. This method is the simplest, but requires the most manual labour.
   While most smudges and semi-erased lines are treated as white, others are not. The ones treated as black have to be taken out manually. Some fine lines disappear entirely, and others may have the wrong line weight. Adjusting the contrast settings will only minimize this problem.
   B/W images have no Anti-aliasing, which is the use of gray pixels to smooth out a black line. This can make a very ragged edge, which usually has to be cleaned up manually. Anti-aliasing can be added  by scaling down the final image size by 25 or 50%. However, the same process works on grayscale images. An anti-aliased grayscale image has a better initial quality, and this carries through to the final version.
   Using B/W, one can use flood-filling to quickly add colour to large areas. This can't be done with anti-aliased images, because the flood fill doesn't colour the gray pixels, which makes that accursed white halo. While flood-filling makes adding flat colours quick and easy, it has limitations. It doesn't make adding shadows any faster or easier, and it all the shapes to be coloured have to be closed shapes (IE: no gaps in the lines). The more open the shapes are, the more work it is to colour them. Not that much more work, mind you, but at some point it takes as much time to flat colour an image using flood filling as without it.



At any rate; this is a first draft of the guide. If there are any suggestions for improving it, I'd like to hear them.

I don't know how well this method will work for other people, either. If you do try it, let me know how it goes.

EDIT: Originally, I'd commented that I hadn't seen this method mentioned anywhere else. Upon reading Turnskys "So you want to Draw," which was posted before this was, I noticed he breifly mentioned using Multiply for line art. That's that, then.