Step 3: let's color the damn thing already!
Get a chair pillow. This is gonna be long.
First thing's first, make sure your image is in RGB Color mode! Most manga scans are saved in Grayscale mode since it doesn't take up as much space, file size-wise (and manga scans are b&w to begin with, anyway). If you try to color in Grayscale mode, all colors will show up as gray hues instead. To switch to RGB Color mode, go to Image -> Mode and select RGB Color.
After that, you'll need to zoom in quite a bit when coloring. As I color mostly icons (and very rarely bigger scans), I usually zoom in to about 5-600%; depending on the size of your image, it's recommended that you zoom in to at least 300%, especially if your image has lots of details (jewelry, accessories etc.).
3.1 Picking the colors
And I'm bringing up the knowledge part again: you need to know basic stuff like skin tones or hair colors in order to get great images. This depends on your personal taste but there are also certain conventions you need to take into account; some examples:
Dark skinned characters are often a challenge because you have to figure out how "dark" you can go (so that the features are still visible) and how light you can go (so that people can tell they are supposed to be dark-skinned).
Neon colors are trendy, but you do run certain risks if you want to use them: use too many and it'll just look lazy and clashing, use too few and they'll clash even more. There're also quality concerns: one of the most pretentious colors is red because you have to save the image at a higher quality to avoid pixelation; this in turn implies a bigger file size, which can hinder submissions to sites with a certain file limit.
e.g. a completely pure red .JPG file of 1920 x 1200 px takes up 31 KB; adding five small, gray discs with soft edges almost triples its file size (80 KB).
That's not to say you have to try and guess the right colors from the begining. Pick colors that are in the general vicinity of what you want - you will most probably end up modifying the hues along the way, anyway (see Step 4 for correcting hues). The best way to become familiar with "normal" hues is simply looking at images and finding a common denominator between them.
Example of choosing the colors of a skintone (from here).
Tip: depending on the hues, you can use the base coloring for the shading as well, by setting the shading layer's BM to Multiply and, if need be, playing with the Hue, Saturation, Contrast and so on.
3.2 The base coloring
I normally start with coloring the skin, then the hair and, finally, the clothes. The skin comes first because it's in relatively easy to fill shapes (face = circle, arms = rectangles and so on). Then I color the hair, which is more difficult because you also have strands that are more irregular in shape (most can resemble triangles, but what about the wispy bits?). By doing this, I go crazy and just disregard the parts where the hair meets the skin, in order to avoid color leaks (that only works if you place the skin layer above the hair one, though). Finally, I leave the clothes and background last because these aren't that important to me (I like to focus on the features and hair, but that's just me ^_^).
Alternatively, you can color the image by using the Pen Tool, with shape or raster layers (see Step 2.2), but I only recommend this for larger images or those where you find yourself needing to change the brush size often. For icons and smaller images, just painting with a brush is enough.
Tip: never use "pure" colors like black and white, they look too extreme. Just move the color picker a bit to the left or below and pick shades of gray or whatever color you want to use, instead.
There are two main methods mangakas use when shading images: screentones and solid colors*. Screentones are used more often in paper-based manga since they're cheaper to print, while solid colors are used predominantly in digital manga. Most readers, however, associate manga with screentones, which are considered to be the "traditional" form of shading.
* gradients are also used, but very rarely and mostly for digital manga.
3.3.1 Images that already have shading
Coloring the scan depends heavily on what type of shading the mangaka uses. As we've seen in step 2, screentones greatly affect the colors you use and should be removed. How? Well, you can use Levels and Curves for lighter screentones, as those are usually enough to get rid of them. For more complex screentones and darker solid colors, those aren't enough - they have to be deleted manually. To do this, just "paint" over the surface with a normal, round brush or use the Pen Tool to create selections and fill them with white.
Tip: when working on the base image, try not to actually delete things - cover them up with white instead. If you go back to the image later on, you can view it properly without having to place a white layer underneath it. Also, using white helps avoid jagged lines; when you place colored layers under the lineart, it will look softer and more natural.
Before removing the screentones, though, you should make sure you know how to shade the image. This can be done two ways:
1) duplicate your image layer and work on the original (the copy will serve as a backup and reference; you can either set the layer's opacity to something like 30% or turn its visibility on and off when need be);
2) delete the screentones and add your own shading.
Since I make mostly icons, I remove screentones from faces and hair all the time; since the resolution is small (100 x 100 px), I work with brushes about 2-4 px big and modify the Hardness to about 50-60% (this is to ensure a smooth erasure). Personally, I prefer using the original image as shading reference whenever I can because I don't have to think too much about the lighting ^_~
3.3.2 Images that do not have shading a.k.a. creating it from scratch
This is more often the case of simple manga images, as opposed to, say, chapter covers - which are supposed to be eye-catching and, therefore, are more worked on. Simple images don't normally have screentones or solid colors for shading (this depends on the mangaka, of course). So what do we do when we have to guess?
Do your research by observing other images on the Internet and also by seeing how light affects objects nearby. If possible, try to draw objects and add shading to them. Play with size, angles, sources of light etc.
Think about the lighting and location of the scene:
- What moment of the day is the scene happening in? Aside from the shading, this will have an effect on the hues you use:
- dawn: lighter yellows, a bit of blue and pink
- daylight: normal tones
- afternoon: stronger shades of yellow and orange, maybe even a bit of red
- sunset: deep shades of orange and red, a bit of purple [example]
- evening: light blues, dark blues, maybe even a bit of dark purple
- night: dark blues, colors tend more towards grays and aren't as strong [example] [example]
- Where the source of light is: above, below, on the side, in front, behind etc.
- What the source of light is:
- natural: fireflies, fire
- artificial:fireworks, street lights, lamps, candles, spotlights (bright), spotlights (normal), spotlights (concert - mono), spotlights (concert - colorful)
- magical*: example, example, example, example
* Unfortunately, most artists that do official art for series nowadays don't bother to incorporate the lighting from the "magic" part in the background (see first example).
- How the light is: dim, bright, colored, reflected, filtered (passes through something** before hitting the subject) etc.
- What the weather's like: sunny, overcast, full moon, no moon etc.
- Is the scene taking place in- or outdoors?
** What is in the way of the light (window, other objects, other lights etc.)? Also consider the materials the objects are made of.
Think about the angle of the objects/characters. Where/how does the light fall? Even if the light falls normally (the light source is in front), there will be other shadows if, say, the character has their head tilted, than if they were looking "directly at the camera". There are certain areas to consider in particular when handling such images, such as the eye sockets, forehead, cheekbones and neck - those will be the first affected by shadows.
Tip: sometimes, clothes that feature lots of details look better without shading - see the guy on the bottom's kimono, here.
Related tutorial: Basic Lighting by Alenas. I really recommend reading it, it's excellently written, illustrated and put together in general.
Optional step: complex shading
Complex shading means adding more layers of color to deepen the shadows and highlights. Depending on the number of layers, the image may look significantly different from what you started out with: the more layers you add, the more 3D the image will look.
Like in the tip example above, you don't always have to add complex shading; adding it or leaving it out depends on both you and the image: if the image is small and with lots of details, it's better to only add one layer of shading in order not to hide them (sometimes you don't need to add shadows at all); if the image is big and with lots of details, you may add shading in order to avoid it looking too simple.
Only add highlights when you think base and shade coloring don't emphasize your image enough and ONLY to areas that would logically be highlighted, else you risk making everything overly shiny and taking away from the impact of said highlights. Of course, you can add highlights to virtually anything, but don't lose track of why you're adding them in the first place! Areas that normally require highlights are:
- eyes (may look lifeless otherwise): ;
- hair (may look flat otherwise);
- objects made of certain materials (metal, rubber, plastic, satin etc.).
Just like the shadows, highlights can be applied via Brush or Pen Tool and go by the same rules as the shadows, as far as placement goes (see step 3.3.2). You can either use just solid colors or - my personal favorite! - play with layer blending modes (again). I highly recommend using the latter, since BMs actually help, y'know, with the blending *insert mind-bending revelation here* No, really, solid colors end up looking blocky most of the time, like they've been slapped there. So, what BMs can you use for highlights?
Screen is useful for both highlights and other "glowy" effects; it is the opposite of Multiply in that it gets rid of darker colors and keeps the lighter ones. I normally use it on hair highlights; however, there are times when Screen makes the colors look lifeless and should be substituted for another BM, or the color should be changed.
Color Dodge and Linear Dodge are, like Screen, useful for highlights. Color Dodge blends the colors more (not much, though), while Linear Dodge "burns" them; you'll end up with highly contrasting colors so I'd recommend using these BMs for hair highlights if the base hair color is dark and details are less visible (black or dark brown).
Lighten is a BM I don't use for coloring since it replaces the darker hues of the image/coloring with the color of the Lighten layer. This is best used when the image has a moderate amount of dark colors (those will be replaced first); it also tends to obscure details on images with too many dark hues and isn't even noticeable on those with little black. It is best used for effects rather than just coloring, since it has no significant impact (if any) on the base coloring.
(shamelessly (mostly) taken from here...but changed a bit).
Note: putting the highlights on Lighten in the example above makes them virtually invisible.
Tip: in order to better add shading to fabrics, you can pick a single hue of gray for all the shadows that apply to a material - this will prevent the clothes from looking "fragmented"; if you have lots of colorful details, try to add a single shade to the folds, instead of shading each color individually. Be wary of other materials (beads, metal etc.), however, as they DO need special, separate shading!
3.5 Touching up the lineart (again)
After finishing the coloring, you may find that the lineart doesn't look good, even though you took care of it in the beginning. It might still look blurry, even though you sharpened it, or the lines might be too thin and the lineart not too visible. To fix this, I either:
Use more filters: if the lineart looks blurry, another Smart Sharpen or two usually do the job.
Play with layers and BMs by
- duplicating the lineart layer and leaving both with a single BM OR
- duplicating the lineart layer and giving each their own BM (e.g. one has Color Burn, the other - Linear Burn).
Of course, you can play with both filters and BMs, though one method is usually enough.
Optional step: touching up the entire image
Sometimes I feel that the colored image still looks lifeless, even though I've theoretically picked vibrant colors and played with the lineart's colors and BMs. To give that image some more contrast without having to rely on tools, I place two layers on top of all the layers of the image (yes, including the lineart):
- cyan/teal: either #00ffe4 or #00ff96, depending on what you want to do; Color Burn, 100% Opacity, 10% Fill, placed above the red one
- red: the purest hue, #ff0000; Overlay, 30% Opacity, 30% Fill, placed below the cyan one
Of course, you can modify the values to suit your needs: increasing the Opacity/Fill of the red layer (and decreasing those of the cyan one) will make your image look warmer, while the reverse will make it look colder. This is a great way to quickly fix "lifelessness" without going for neon colors for basic coloring, since it affects all colors.
But it's still not done yet! We've still got step 4 - we do have to check for mistakes, don't we?