Gorsal's Pixelart Tutorial 2017

[Gors]

A dot, a passion, a living

Disclaimer: All explanations in here are personal and thus should not be taken as the only way to make Pixelart. This is for didatic purposes only. Do not copy this text or post it as your own. Thank you.

Preface

Pixelart is an art media that is older than you might think. Thriving as mosaic, then as embroidery, many of pixelart's elements can be traced back to those ancient medias. Creating pixelart is enjoyable, as long as you build the stepping stones of what is a pixelart, and how much you can extract out of it. Think about it as a dictionary for pixelart. I want you to browse and look up for the meanings of the terms, and practice them yourself.

Getting Started
I'll be working on this thread during the course of this year. I have learned lots of things since the last time I made the current Pixelart Dictionary, and this is the moment to improve upon it. Nothing better than a pixelartist to talk about pixelart, much like you want a programmer to deal with programming and a musician to make music. I want to have full control of what is being made here, is all.

I'll create each entry in this thread as its own post, so that you can easily browse between them. This also means, I'd appreciate if no one posted here until I'm finished. If you really need to talk about it, send me a PM and I'll work on it. Additionally, since I'm no longer a mod, this means that the forum programming will automatically merge my posts if I attempt to make many posts in succession, so I'll have to wait at least 1 hour until I make the next entry. I apologize for the slow start, but I intend to create something that will be more durable than the current thread.

History
Pixelart is above all, an art of non-complexity. Not in the sense that it is easy to make, far from it. But it is a media that requires the piece to be understood clearly even in a small resolution. A pixel is the lowest graphical element in a pixelart, and thus some detail is lost in the process. Additionally, game cartridges came with very limited graphical capabilities, which further crippled the way you can make pixelart in a game. It's a media about using resources well, in a strategical manner, to represent an element, or a character. This gives Pixelart a very iconic feel - each frame or each character is like a symbol that is read - and being able to quickly discern it is the speciality of this media.

There is no such thing as "half-a-pixel", and while you can mimic it, there are inherent complications on how to represent a given graphic. This in turn gives us a whole slew of terms that are not only art-related, but Pixelart-only terms that needs to be understood. This Tutorial's aim is to provide you, the reader, a better understanding of those techniques, with examples and how to make them.

Glossary
Click on the term to jump to its entry. Note that some of them can't be clicked yet due to reasons stated above. These terms are in alphabetical order for your convenience. The small parenthesis denote the parent topic of the given subject.

8-bit
Antialiasing
Banding (Shading)
Cluster
Contrast
Custom (Sprite)
Dithering
Edit (Sprite)
Format
Frame
GIF (Format)
Gradient Shading (Shading)
Index Painting
Isometric (Perspective)
JPG (Format)
Jaggy (Outline)
Layering
Lightsource (Shading)
Lone Pixel (Cluster)
Modular Sprite
Motion Blur
NES (8-bit)
Outline
Palette
Perspective
Pillowshading (Shading)
Pixelart
Readability
Recolor
Resolution
Saturation
Selout
Shading
Sprite
Subpixeling (Antialiasing)
Splice
Tileset

[Gors]

8-bit

---

Forget about this term, pixelart-wise. When people say "8-bit", usually it refers to low-quality pixelart with reduced palette and apparent pixels. But actually, 8-bit is a much more abrangent term and totally not related to the quality of graphics.


Wait, this isn't 8-bit??

8-bit is a term that refers to the processor of old game consoles and computers. That is to say, one bit is a single binary unit (that is, 0 or 1) while eight bits are, as you expected, eight binary units. This system allows programmers to store variables from 0 to 255 units, which is why this number appears a lot as a cap number, such as RPG stats, maximum money you can carry, or Pacman's kill screen.

Graphically, such a processor doesn't really define the looks of the game, as many devices used 8-bit processors, for example:

NES

NES is able to display sprites and graphics with up to 3 colors and 1 alpha transparency for detail.

Gameboy

Original Gameboy screen. Graphics are rendered in 4 monochromatic tones: White, Light Gray, Gray and Black.

Master System

Sega's Master System is able to produce more colorful sprites than the NES, though it has other types of limitations.

Commodore 64

Commodore 64 is very modest, with only 1 color for each graphic (there are no intermediate shades). Late C64 games could produce more colorful and detailed graphics.

There are so many other devices with 8-bit processors, but as you can see, there are actually many graphical discrepancies between those systems and therefore, the usage of "8-bit" to refer to pixelart or aesthetic should be avoided. Use "(console)-inspired", such as "NES-inspired" or "Gameboy-inspired" instead. Each console has its ups and downs when rendering graphics, so grouping them together into a single umbrella term doesn't really help.

[Gors]

Antialiasing

---

Antialiasing, also known as anti-aliasing, antialias or AA for short, is a technique where you place intermediate pixels on strategical places in order to smoothen it.


Left: A pixel line. Right: The same line, with 3 gray tones of AA.

Usually, lines are anti-aliased on the corners and diagonal lines, where the irregularity is more apparent. Since pixelart deals with squares as the smallest graphical element, the places where "half a pixel" are painted are rendered with a mix of the main color and the background.


Diagram representing zoomed pixels on a screen. Notice that the pixels that are partially painted gets a lighter tone in order to render the line.

To successfully antialiase a pixelart, you need to pick a color that is intermediate between the line and the background. So, if the line were to be red against a white background, the anti-alias lines should be lighter versions of red to properly blend the line. Notice that overdoing anti-aliasing is not a wise idea and the more shades you create to smoothen, the more blurred your sprite will be.

Due to its nature, antialiasing should be limited to details inside the sprite's outline than outside, in order to conserve readability, sharpness and use. Antialiasing outside the object should only be done if it's not supposed to move or be used in any other background.


Left: A crudely pixeled Mario head. Mid: Same head, but with AA dots for a gray BG. Right: Same as Mid, but now against a light red BG. Notice the gray AA dots contrasting with the BG, filling the piece with unwanted noise.

Antialiasing properly is the key of pixelart proficience. Some methods I use are as follows:


Types of antialiasing employed.


There are 4 sets of lines, in 1x, 2x and 4x zoom for your convenience. The leftmost set has no antialiasing, while the rest has different styles of antialiasing.

Figure 2 has the most basic antialiasing method: placing gray dots where the line breaks. Notice that it is not placed in every single corner - only the corners that need extra care should be worked on. Try to picture where the pixels would be "half-painted", and antialiase them accordingly.

Figure 3 is a development of the basic antialiasing. It uses two antialiasing shades, with the darker shade touching the corner and the lighter shade placed tangentially. This makes the line smoother, but also appear thicker. Use wisely as you don't want to make it extra blurry.

Figure 4 is what I call "split-line antialiasing". This is because the antialiasing dots split the line. This requires at least 2 shades to look good, but it's as smooth as Figure 3 and doesn't look as thick.

You can try and come up with different ways to do antialiasing, such as increasing the gray dots' length, or combining split-line with traditional. As long as it looks good and in-style, everything is doable.

Remember that antialiasing is not to be used exaggeratedly, or else you'll make the pixelart very blurry and harder to work on later. For animated characters in games, 1 or 2 shades often work well, but static images can have more shades for it.

[Gors]

Banding (This is a child topic of Shading)

---

Banding is a faulty shading method where the shades stick to the outline, without real consideration to the volume. Banding makes the sprite look like a sticker instead of a real object. This is a mistake commonly made by starters who recently learned to avoid pillowshading, which is another faulty shading method.


Examples of shading. Fig. 1: Pillowshading. Fig. 2: Gradient shading. Fig. 3: Banding. Fig. 4: Properly shaded sphere. They all use the same outline and colors.

When shading an object that is supposed to be 3D, think about how light and shadow would interact with it, and shade it accordingly. Simply adding colors into the pixelart isn't synonymous with quality, after all.

[Gors]

Cluster

---

Despite pixelart being the art with pixels, the part that is actually important when making pixelart is managing the clusters that make up the image. In other words, a cluster is the group of same-colored pixels, or close enough to be read as a single cluster.


A small pixelart of Mario. The only thing between them is a single eye pixel, but even a single pixel causes the image to be read differently.

Making definite clusters is a great way to give your work enough readability even in a small sprite like this. This happens because pixelart is a 2D media, and same colored areas tend to be grouped up by our brain.


Every artwork (not only pixelart) can be interpreted as a sum of equivalent parts. Notice that each color (pink, skin, black) takes its own space, forming strong silhouettes as they're combined.



The same sprite, but with arbitrary lines all over it. Because of them, the clusters become subdivided into smaller sections and thus, unable to form shapes. It's unreadable.


Additionally, even if the sprite has more than three colors, this rule still applies. With correct contrast, you can make two or more shades still be interpreted as one.


I gave Mario some extra colors, but these extra shades add up to the base color and retains readability. However, if I use a very dark color to shade the skin, the shade stops being read as part of the skin tone and becomes part of the facial features, creating another unplanned cluster and ruining readability.

Readability is especially important in smaller pieces like this Mario, but it's a must even for bigger pieces. It makes easier for the brain to process the image, which is helpful in the long run.

[Gors]

Contrast

---

Contrast is the relation between two shades. This applies not only within a sprite, but also to the sprite in relation to the background. A good contrast is the key for readability and enjoyment of the piece.

A high contrast means that the difference between the shades are too great. For example, the reason you can read this text is because the letters are black and the BG is white (or vice-versa), and there is a huge gap between black and white.


High contrast is great to form strong shapes and silhouettes, such as text.

Conversely, low contrast means that there is little difference between the shades. Usually a low contrast shade is used to add in-between details between two high contrast shades.


Same image as above, with low contrasting shades for the text. Notice how harder it is to read the text now.

It's important to note that since "contrast" is the difference between two shades, this means that one color can be both high and low in contrast, depending on the BG it is placed against.


In this example, the color A has low contrast compared to color B, but has high contrast compared to color C. With this in mind, you can create interesting color palettes.

Keep in mind that ideally, you will want to keep contrast in a balanced level so it is not too poppy nor blurry - a high contrast might create unwanted clusters all over the image, while low contrast might be interpreted as one single cluster.