A very opinionated guide by icrawler.
This guide is very opinionated and you don't need to follow the exact steps I outline here.
This has been written for desktop but most of the concepts here apply to many editors as well.
For the programs, you can use pretty much any well-featured photo-editing/digital art program. Those that include:
- Layer blend modes (essential). Multiply, Addition, Overlay, Color, and Lighten are the most commonly used in my stuff.
- Sane selection tools such as Lasso and Magic Wand (sometimes called Contiguous Area Selector.)
- If you're primarily drawing stuff instead of selecting, a fill (non-transparent "inking") brush and and airbrush.
- Basic selection modifying commands such as "grow/shrink selection" and "invert selection".
- Layer masks (not strictly essential but really convenient if you don't want to erase and re-draw stuff all the time.)
- Image adjustments functions/layers (again, not strictly essential but you can use it to make it look better.) Stuff like Levels and Color Balance are pretty useful.
If you don't know if your program has these features, google your program name and that feature.
- Add examples
- Add examples
I put the linework below everything and add the color adjustments above it. Some people would have the page above everything and set to Multiply (which is completely fine!) but I feel this way is more flexible for me.
Value is a term used in art that essentially means "brightness" or "lightness". I'm going to be using that term a lot here.
Masks are layer modifiers that can make some parts of the layer opaque while the others transparent without permanently altering the layer. On PS and similar this appears beside the layer.
Depending on how you work or how you want your pages to come out you can skip some of these steps.
First, I like to make sure all the linework is pure black instead of dark gray (I'll explain why later.) I use the Levels adjustment to do that.
If your linework is already black you can skip this.
Levels changes how "exposed" the colors are. In most programs, in the Levels dialog there's a map of your image's value range (the histogram) and two sliders below it. The most important part is the map and the first slider (usually labled Input Levels or Input.)
Paint.NET has levels but the graphs and bars are vertical so for those who are using that, mentally replace "left" and "right" with "bottom" and "top".
Since we're dealing with grayscale manga pages, the histogram maps how much of a particular "gray" is in that image. So if your page is overall light you might see a big area on the right half while if it's overall dark you'd see a big area on the left.
In Photoshop and other programs you can see handles below the graph that you can adjust (in Paint.NET that's on a separate grayscale bar that's labeled Input.)
Input (usually) has three controllers (or points), a black one, a gray, and a white. Moving these three alters the exposure in some way but the one we'll be using is the black point (usually the one in the left side or bottom end.)
So anyway, that's enough exposition. Create a new Levels Adjustment Layer or (if you'd like it to be permanent, go to Image > Adjustments > Levels [PS] or Adjustments > Levels [Paint.NET])
If your linework is not black but dark grey (which is the case of most official scans like Crunchyroll), you'll see a small "hump" near the left side of the graph instead of a line at the very edge. Move the black point to the right until it's just past the bump. Now all your linework is black! Simple right?
Also, if your page has a light gray instead of pure white, you can apply the same thing above but with the white point and moving ot the left.
I don't like having my colors overlay the SFX and text bubbles so I copy them before starting the coloring.
With the main page layer active, take your Magic Wand tool and click on the white area of the text bubbles. Do the same for the rest while holding down the Shift key to add them to your current selection.
If you're trying that and it your selection spills outside the text bubble, that means the Contiguous option is disabled. Enable that by clicking on the checkbox and try again.
Create a new empty layer and fill your selections with a visible (for your convenience) color like bright red or magenta . In PS and other programs you can do Shift+Backspace to do it quickly.
Now you should see a couple of bubbles with the text looking like it was cut out. With the layer active, use the Magic Wand tool and click on the area outside the bubbles.
If you did that and it selects only a small space instead of the entire area outside the bubbles, that means the Sample all layers option is enabled. Uncheck that option and try again.
Invert your selection. In PS and similar programs that should be in Select > Invert selection, in Paint.NET the shortcut is Ctrl+I. Now (hopefully) you finally have the selection required to copy the text bubble.
With the original page active, copy it using your new selection and paste to a new layer.
Read this whole section first before doing anything.
These are a bit harder to select but to get a suitable headstart, use the Magic Wand to select their innards. Sometimes you might spill your selection so be careful. Adjust the selection by using the polygonal lasso tool (or lasso tool if you're using a tablet) while holding down Shift to add to the selection if necessary.
If you have tiny holes when selecting the SFX using Magic Wand, increase the Tolerance value. Don't go too far though or your selection will spill further.
After you have your selection, copy and paste from the original page onto a new layer...
...is what I would say but because SFX is less straightforward to select than text bubbles (you will get holes if you're not meticulous,) it's better to do something different.
Duplicate/clone your page layer (in PS and similar it's Ctrl+J; in other cases you can just select everything and copy onto an empty layer), then use your selection to create a Mask.
You can then make a temporary layer containing a bright color to clean up your selection if necessary.
Now for the meaty part. Coloring the pages.
Because I like being organized, I create a group layer per every related panel. Every group layer is masked so it appears only in that panel.
There are two ways I color stuff in a panel:
- Coloring with Multiply blend mode and
- Coloring with Color blend mode
Most people have used Multiply at least once. It's the most basic shading blend mode that's really simple to use and makes great colored shadows.
This is super basic and achieves decent results very fast. Because manga is predominately based on pure white (the skin is white with various tones, items are white, sometimes the background is white), the Multiply blend mode achieves great results.
However, when you add a color on an item that is toned darker in the manga (like handguns), the end result is that it would be even darker than the original. That is because the way Multiply works.
Multiply literally multiplies the top pixel's RGB components (from 0% - 100%) with the bottom pixel's RGB components. Because white the brighest color (r, g, b=100%), a color on Multiply blend mode will be applied as-is (color × 100% = color.)
However, if you multiply a color to a non-white gray, you will get a color which is darker than either of them (e.g. 50% × 90% = 45%.)
In some cases this is not what you want.
I mostly use Multiply to color the backgrounds, to darken items, or on really simple panels with lots of white.
When using Multiply to subtly shade stuff, instead of adding a color to a layer and reducing its opacity, try selecting paler colors first for a subtler effect.
Color is a blend mode that I don't see used often for non-photographic content because of how unintuitive it is at first.
If you try to add skin color using the Color blend mode to a face, you'd probably get this result:
(white with skin-colored tone.)
That's because it only applies the hue and saturation to a pixel, not the value/brightness.
Skin is not pure white like manga pages, it's slightly darker. If you want to color using Color blend mode, you need another layer beneath it to correct the values.
This is where Multiply and Overlay comes in handy. Create a new layer below it with Multiply blend mode and apply various tones of gray (starting from the very light ones) to progressively correct the colors.
For higher-contrast dark/light items (metal, jewelry), you can use a different layer with Overlay and add various shades of dark or light gray. Essentially, if you add dark gray to a high-contrast dark item, the highlights will be left mostly untouched.
Because Color doesn't screw with the lightness, you can use it to color originally dark items (hand-guns, suits, etc.) accurately without it becoming too dark.
I structure my main Color layers as follows
- Various miscellaneous layers for other items @ Color
- A uniform layer with the colors of the clothing @ Color
- A filled skin, hair, and eye tone layer @ Color with a mask so it applies only to the skin/hair/eyes.
- One or two value adjustment layers @ Multiply/Overlay
- A background tinting layer @ Multiply
You don't have to. If you can use Multiply on its own to great effect, by all means do it! You can mix-and-match any of the methods here depending on which is appropriate
This way, in my opinion, allows for finer and more predictable control on the panel's value distribution and composition. Which is pretty useful for really complex panels.
(It's also because I paint stuff starting from grayscale before adding color, so I'm already used to this.)
There is however a very useful property that can make you more efficient:
Since using Color ignores your pixel's value/brightness, while you're coloring, you can select any part that is colored (that isn't black or white) and paint it and it would work. This is very useful if you are lazy or if your program doesn't have a "sample current layer" option for the eyedropper tool.
Sometimes having black lineart looks really plain and boring. What if you want the lineart to be colored?
There's a very simple way of doing this, and that is using a layer with the Lighten blend mode.
This works as long as your new lineart color isn't lighter than the main tones.
Lighten works by applying your color if that color is lighter than the bottom pixel. Darken works the same way but in reverse (apply if top is darker than bottom.) I haven't found much use for Darken though. Perhaps it could be used for coloring highlights without affecting the midtones (such as giving green jewels a nice light blue sheen?.)
There are 4 blend modes that I use most often for lighting tweaks.
Screen, Linear Dodge, Color Dodge, and Addition.
These have their own unique character:
- Screen is like Multiply but for brightening stuff instead of darkening.
- Color Dodge brightens and saturates the colors as well. This is very useful for vivid lighting.
- Addition is like a harsher Color Dodge. Great for explosions.
- Linear Dodge is like the middle-ground between Color Dodge and Screen.
When using Color Dodge, Addition, and Linear Dodge for vivid lighting, use a brighter saturated color instead of a light pale color. You'll get better results from that.