Pixel art (dotto-e in Japanese) is a form of visual representation. The term “pixel art” implies “retro game graphics” as it was the mainstream of video game graphics from the 1970s to 1990s. On the other hand, in recent years, the status as a graphic style of “old and new” is being established. In this article series, we introduce the features and attractiveness of pixel art, including these trends. The first part considers what pixel art is.
eBoy, Rio (2011).
From http://hello.eboy.com/eboy/2011/12/01/eby-rio-poster-34k-png/.
Video games and pixel arts
Pixel art (dotto-e in Japanese) is a form of visual representation. Most of early video games used pixel arts for their graphics due to their technological constraints (*1). From the 1970s when Atari’s arcade games flourished, to the mid-1990s when 3D computer graphics really came in the realm of console games, it may be called the age of pixel arts in digital culture (*2).
It was also the time when video game as a medium was more and more sophisticated and established as a mass culture. Thus, we now see pixel art as a visual form that is deeply connected with video game culture. There are a number of pixel arts used for book covers or event flyers related to video games.
In fact, pixel arts have been out of the mainstream in video game graphics for the last two decades. As a result, they imply something “nostalgic” as well as “related to games.” Pixel arts may bring retrospective feelings such as “those good old days.”
It is often the case that nostalgia yields a new style, however. Just as chiptune derived from old video game music with square waves, a new visual style departing from old pixel arts has been growing for the last decade. One of the main stage for this new pixel style is indie games, where designers and artists explore the potential of pixel art as a form of representation.
Pixel art as a visual culture
Despite its strong connection to video games, pixel art as a style of representation is no longer limited to video game culture. To create and appreciate pixel arts is now a culture in its own right, where people enjoy pixel arts in different respects and interests: new visual style, nostalgia, cuteness, and so on.
A number of art books devoted to pixel art has recently been published in Japan. FF Dot: The Art of Final Fantasy (Square Enix, 2018) showcases pixel art works from the Final Fantasy series in the age of the Nintendo Entertainment System and the Super Nintendo, especially works by the company’s graphic designer Kazuko Shibuya, also known as “the master of pixel art” (dotto-e no takumi in Japanese). Pixel Vistas: A Collection of Contemporary Pixel Art (Pikuseru Hyakkei, Graphic-sha, 2019) collects works by the “front-line” pixel artists from inside and outside the country, showing various trends of pixel arts in recent years.
From Square Enix, FF Dot: The Art of Final Fantasy (2018).
hermippe, Shiro [A Castle] from Pikuseru Hyakkei (Graphicsha, 2019).
Exhibitions that display and possibly sell works of pixel art are also regularly held. Pixel Art Park, an annual exhibition and sale exclusively for pixel art goods, started in 2015 and has been growing year by year (*3). Shibuya Pixel Art is an annual contest and festival for pixel artists since 2017, in which prize winners are exhibited in various parts Tokyo’s Shibuya Ward (*4).
Besides these recent domestic cases, there has been a lot of graphic artists specialized in pixel art for a long time. For example, eBoy, an artist group based in Germany and Canada that focuses on creating pixel arts, has published their works since around 2000. They are well-known for the “Pixorama” series that represents various cities in the form of isometric pixel art. In addition, there were and are many household or stationary goods that borrow pixel graphics from video game classics. These examples all show that pixel art has been appreciated in its own right, that is, as a distinctive kind of pictures rather than graphics just attached to games.
“Hobonichi Techo 2018: Koko [wa] Majikanto No Kuni [This Is the Realm of Magicant].” A diary cover using pixel arts from EarthBound .
The definition of pixel art?
As you can see from figures above, the term “pixel art” covers a great variety of images. But I will leave the question of what subclasses pixel art has for another time (*5). Instead, the rest of this essay examines what pixel art is in the first place: the definition of pixel art.
What is pixel art? In other words, what is the feature( s) that all and only things called “pixel art” have in common? There has never been any sophisticated definition of pixel art as far as I know, but some observations on pixel art can be a starting point to define it. Illustrator Masakazu Sukura (a.k.a. scrama_sax) describes a peculiar feature of pixel art on Twitter as follows:
I think that a pixel art is an image such that changing any one pixel of it results in changing the information that it conveys. If you change vertical two pixels into horizontal two pixels, open eyes turn into closed, for example (*6).
As art critic gnck agrees (*7), Sukura’s view exactly focuses on one of the features that pixel arts share. Whether it is a pixel art or not, any raster image consists of a set of pixels. But not all raster images are pixel arts: Being consists of a set of pixels is not the sufficient condition for an image to be a pixel art. Pixel art is distinctive in that each single pixel constituting it is important. As Sukura says, this importance means that any slight change in it leads to the change of its content. In contrast, in raster images that are not pixel arts (e.g. photo images saved as JPEG format), each single pixel is not so important. I think Sukura’s observation is right.
Another helpful account is from game journalist Imai Shin’s essay:
A pixel art is a raster image that uses a set of pixels to depict something and is a representation designed to be such that each pixel or each bunch of pixels is distinguishable with the naked eye (*8).
The point is that pixel are “distinguishable with the naked eye.” Not to mention, we can generally recognize a pixel art as pixel art at a glance (at least just with a little zooming in) and that is because we can distinguish each pixel as constituents of a pixel art with our own eyes. Conversely, in a raster image that is not a pixel art, its constituents are generally not distinguishable with the naked eye. It is difficult to identify individual pixels of a digital photo unless it has an extremely low resolution.
If I just want to roughly understand general features of pixel art, these two insights would be satisfying. But I, as a philosopher who has a bad habit to stick to seeking exact definitions of things, will try to propose a well-formed definition of pixel art, invoking an existing systematic theory of depiction.
Pictures in general
Pixel art is a kind of picture. Put differently, it is a thing depicting something. Not all graphics that consist of a set of pixels are pixel arts, even if each pixel can be distinguished with the naked eye. For example, a pattern generated by cellular automaton whose constituents are pixels is not a pixel art in the sense that it is an abstract design but not a picture (*9). Mere pixel pattern is not pixel art. For a set of pixels to be a pixel art, it is necessary to be a picture.
/div>
“Glider Gun,” a pixel pattern generated by Conway’s Game of Life.
From https://commons.wikimedia.org/wiki/ File:Gospers_glider_gun.gif.
In order to examine the features of pixel art, we have to understand the features of its superclass. What is a picture? This question has been discussed in the philosophy of depiction, a branch of modern philosophical aesthetics. There are several approaches to the nature of pictures (*10). The approach I will consult here is one called a “structural account,” which attempts to understand pictures as a kind of symbol (*11).
According to the structural account, words, musical scores, maps, diagrams, and pictograms are kinds of symbol in that they all represent things (persons, objects, events, states of affairs, relations, thoughts, and so on). Pictures such as paintings, drawings, and photos are also a kind of symbol exactly in the same way. Then what makes pictures differ from other sorts of symbols? John Kulvicki, the proponent of structural account, lists four features as distinctive of pictures (*12):
1. A little change of property in a pictorial symbol leads to a change in its identity and content. Other kinds of symbol do not have this feature relatively. In the alphabet, for example, it is usually not the case that an instance of the letter A becomes that of other letters just because its shape changes slightly. In contrast, in drawing or painting, any minor change in the shape and color of a mark on the surface can affect what it depicts.
2. In pictures, relatively many types of properties are relevant to a symbol’s identity and its change. In a line graph that represents monthly average temperature, for example, only the shape of line is relevant for it to be a certain symbol. In contrast, relevant properties for a drawing that represents a mountain ridge includes not only the shape of line but its color, thickness, brushstroke, and so on.
3. In pictures, each symbol is assigned different contents. As a result, there is little redundancy such as synonymy, in which different symbols have the same meaning. This feature becomes clearer by comparing pictures with pictograms. As far as we see it as a picture, the desktop icon of trash bin depicts a trash bin as having rich and specific features, e.g. gray, metallic, wire meshing, cylindrical, etc. But if we take it just as a pictogram, a part of the graphical user interface, then it stands for only a poor content, the function of deleting files, regardless its unique color and shape. So are pictograms for restroom. Unlike pictures, in pictograms, the variation of symbols do not correspond to the variation of meanings.
4. In pictures, properties of a symbol basically straightforwardly reflects in properties of a thing it represents: Briefly, a symbol generally resembles its content (*13). For example, if you draw a circular figure with blue ink on paper, you depict a round blue thing (perhaps placed in three-dimensional space). On the other hand, even if you use a blue ink pen to write a sentence with round letters, you do not represents something round or blue.
Kulvicki calls the first feature “syntactic sensitivity”. It is obvious that this feature is very like what Sukura points out as a feature of pixel art. Taken literally, Sukura describes a feature of pictures in general rather than that of pixel art. I do not think that is not what Sukura really intended to say, of course. Accurately restating Sukura’s intent using Kulvicki’s framework, it seems as follows: A pixel art is more syntactically sensitive than a raster image that is not a pixel art.
Pay attention to the fourth feature, too, which is called “transparency”. Though neither Sukura nor Imai mentions, pixel art inherits this feature from its superclass: In pixel art, like pictures in general, if a color is put in a pixel, the same color is put in the corresponding part of a thing it depicts. This may seem to be trivial, but it is an important feature of pixel art as a kind of picture. Pixel font, a kind of font composed of a set of pixels distinguishable with the naked eye, does not have this feature because it is a kind of letter, not of picture.
Defining pixel art
In the previous section, I described the definitive features of pictures based on Kulvicki’s theory. Then what makes pixel art differ from pictures in general? In a philosophical jargon, what is the “differentia” of pixel art? I think both Imai’s and Sukura’s observations are helpful here: to consist of a set of pixels distinguishable with the naked eye, and to be relatively syntactically sensitive. But some additional supplements are also required for a well-formed definition.
First, the “pixels” here are not necessarily identical with pixels as constituents of a raster image. As Imai suggests (*14), pixels as elements of a raster image should be conceptually distinguished from pixels as elements of a pixel art, even though both can be the same in some cases. The former are minimum units to each of which a color value is assigned in raster data (*15). The latter are minimum units a set of which constitutes a special kind of symbol, a pixel art. Therefore, a constituent of pixel art, a pixel in the latter sense, can be each pixel as an element of a raster image (i.e. a pixel in the former sense), each set of them (e.g. 4×4 pixels), or each shape in a vector image (e.g. a square as a polygon), if each is distinguishable in the naked eye. Pixels in this sense even do not have to be in any digital image: Hand-drawn squares or colorful beads can be elements of which a pixel art is made.
Second, pixels as constituents of pixel art generally have the features as follows:
a. square, or at least shaped like a square; b. equal to each other in size, and arranged along (usually hidden) horizontal and vertical gridlines; c. each having one color.
The condition (a) allows for flexibility. For example, for a pixel art that is fine-grained and thereby is hard to distinguish its individual constituents, whether each pixel is exactly square does not matter. There is also a pixel art whose elements are rectangles rather than squares, such as Atari 2600’s graphics that may feel slack due to a part of its minimum unit being wide (*16). Moreover, elements of some kinds of pixel art are not clearly articulated as squares, rectangles, or even any polygonal shape: A pixel art may be made of cylindrical beads as well as of blurred or scratchy marks (*17).
The condition (b) is more essential to pixel art: You could make a picture by arranging triangles or hexagons without any gap, but it should be given a name other than “pixel art.” That is also why a mosaic is not a pixel art. In addition, as gnck points out (*18), a “pixel-art-like thing,” which consists of pixels but is not along any gridline or whose elements are irregular in size, is so awkward that the eye familiar to genuine pixel arts will refuse to count it as a pixel art. The equality of constituents in size and the horizontal and vertical arrangement are the necessary features for pixel art.
The condition (c) is with a reservation. When contemporary video games use a pixel art, it is common to add a lighting effect to it, letting individual pixels have gradation or vein. But even in such a case, it is natural to say that each pixel should have had only one color in the base image, that is, before the effect is given over it. This fact can be explained in terms of perceptual consistency. No matter how modified the actual color is, the base color is perceptually stable.
Summing up these considerations, I propose a formal definition of pixel art: A pixel art is a picture that consists of constituents distinguishable by the naked eye and is relatively syntactically sensitive, where the constituents satisfy the above three conditions (a?c); where “picture” refers to a kind of symbol that have the four features Kulvicki specifies (*19).
French linguist Andre Martinet said that all languages have “double articulation” as their general feature. Double articulation refers to that a symbol system has not only units in the level of words, that is, meaningful minimum elements, but also the smaller units that constitute words together but are meaningless by themselves. In English, for example, meaningful minimum elements are words such as “cat” while words consist of smaller meaningless elements such as “c,” “a,” “t” (*20).
Unlike language, standard pictures do not have the second level of articulation at least explicitly. In normal paintings or drawings, most parts in the surface that can be individuated as a unit will be meaningful: a part depicting a cat, a part depicting the face of a cat, a part depicting the nose in the face of a cat, and so on. In contrast, pixel arts explicitly do have the second articulation. Each pixel, which is meaningless by itself, is to pixel art what each alphabetical letter is to English. In this sense, pixel art is like language, though it is still a kind of picture.
The notion of double articulation has been frequently invoked in order to explain that language embodies two virtues: economy (made of a relatively small number of elements) and creativity (making up an infinite number of meaning) (*21).
Probably the same goes for pixel art. Pixel art is a kind of visual representation that looks simple on the surface but has a great potential due to the infinite number of meaningful combinations of the finite number of meaningless elements.
(notes)
*1
These constraints are practical ones due to the limit of video game hardware’s performance. The technological basis for pixel art is raster graphics as opposed to vector graphics and 3D computer graphics. Both vector graphics and 3D graphics had already been available in the 1970s, at least theoretically. Indeed, some Atari’s arcade games such as Asteroid (1979) and Tempest (1981) use beautiful vector graphics while Battlezone (1980) and Star Wars (1983) adopt a kind of three dimensional representation, though it consists of wire-frame models rather than polygon meshes.
*2
In arcade games, which were usually superior to console games in performance, video games using 3D models began to come out in the 1980s: Atari’s I, Robot (1984) is an early example. But it was the so-called fifth generation of home console that arose in the mid-1990s (e.g. PlayStation and Nintendo 64) that drastically broadened the use of 3D graphics for video games. Classics in this period such as Super Mario 64 (1996) and Final Fantasy VII (1997) symbolize the shift from pixel arts to 3D graphics.
*3
https://pixelartpark.com/ (in Japanese).
*4
https://pixel-art.jp/ (in Japanese).
*5
For an excellent overview of the diversity of contemporary pixel arts and their classification, see Shin Imai, “Gendai No Geemu To Pikuseru Hyogen [Modern Games and Pixel Arts],” in Pikuseru Hyakkei (Tokyo: Graphic-sha, 2019), 192-195.
*6
Masakazu Sukura (@scrama_sax), Twitter, January 17, 2012,
https://twitter.com/scrama_sax/status/159244346590371840 (in Japanese)..
*7
gnck, “Gijutsuteki Seiyaku Kara Bigakuteki Joken E [From Technological Constraints to Aesthetic Conditions],” in Pikuseru Hyakkei, 11.
*8
Imai, “Gendai No Geemu To Pikuseru Hyogen,” 192.
*9
Given that it is named “glider gun”, we may be inclined to say that this pattern depicts gliders and guns that shoot them. But that is like saying that a pattern on the moon or a particular arrangement of stars “depicts” animals. A part of the moon surface or a constellation is not a picture (at least not what I want to call picture here) but a thing in which we might happen to see something. For a related discussion, see Richard Wollheim, Painting as an Art (Princeton: Princeton University Press, 1987), 46-51.
*10
For a classification of different approaches of philosophy of depiction, see Catharine Abell and Katerina Bantinaki, “Introduction,” in Philosophical Perspectives on Depiction, eds. Catharine Abell and Katerina Bantinaki (Oxford: Oxford University Press, 2010), 2-6.
*11
The founder of the structural account is Nelson Goodman. See Nelson Goodman, Languages of Art, 2nd ed. (Indianapolis: Hackett, 1976), esp. chap. 1 and 6. John Kulvicki, who I will invoke, inherits and updates Goodman’s theory. Note that strictly speaking, structuralists focuses not on the nature of individual symbols but on that of “symbol systems” individual symbols belong to.
*12
John Kulvicki, “Image Structure,” Journal of Aesthetics and Art Criticism 61, no. 4 (2003): 323-40; John Kulvicki, On Images: Their Structure and Content (Oxford: Oxford University Press, 2006), chap. 2 and 3.
*13
In fact, Kulvicki makes so technical arguments on the fourth feature (transparency), according to which to equate this feature with similarity is highly problematic, if not wrong. But for our purpose, the crude simplification works well.
*14
In the above quote, Imai carefully writes that a constituent of pixel art is “each pixel or each bunch of pixels” (my italic). As Imai himself explains, the latter disjunct is added in order to distinguish pixels as constituents of pixel art from pixels as particles of digital image or display device.
*15
Strictly speaking, besides the level of rasterization as digital data format (raster image), there is also the level of rasterization as display format in monitors or printing devices (raster scan). Even image data stored in vector format are eventually rasterized by the operating system or the print driver when they are displayed on monitor or paper. Therefore, most images that we encounter today may be rasterized, or “pixelized,” except hand-made pictures or film photos. Whether an image is rasterized in this sense has nothing to do with whether it is a pixel art, of course. For a related point, see Alistair M. C. Isaac, “Digital Images: Content and Compositionality,” Journal of the American Philosophical Association 3, no. 1 (2017): 106-26.
*16
Aside from its technological background, many parts of images in Atari 2600 graphics seem to have a low resolution in horizontal direction. For example, see screenshots posted on a review site for Atari 2600, https:// videogamecritic.com/2600aa.htm.
*17
For a related topic, it is often discussed that in the early video game consoles such as the Nintendo Entertainment System, graphic designers deliberately created pixel arts taking into account their “blurring,” which results from their technological conditions, that is, a CRT monitor and/or an analog video transmission. See a series of conversations around this topic on Twitter (in Japanese), https://togetter.com/li/1131267. If that is correct, the notion that constituents of pixel art are supposed to be distinctly square would been a latecomer in the history of video games. But it could be said that even if old pixel images were all blurred, viewers understood that a pixel art comprises of a collection of squares and they imaginatively saw clear-cut squares beyond vague shapes.
*18
gnck, “Gijutsuteki Seiyaku Kara Bigakuteki Joken E,” 12.
*19
In addition to this base definition, I also offer some additional options. First, the option how strictly we take the condition of distinguishability of elements: It is possible that individual pixels in an image are too small or blurred to be hard to distinguish with naked eye. Whether we should still count such a case as a pixel art or not is the matter of choice. If we decide it is a pixel art, the condition of distinguishability will require to be looser. Second, the option whether we adopt how to create as an additional condition: On the one hand, we have the choice whether or not we should count as “genuine” pixel arts the pixel-art-like images that are procedurally produced by algorithm processing hand-made pictures, photos, or 3D graphic images. On the other hand, we have the choice whether or not we should count as pixel arts mere low-resolution bitmap images that are crudely drawn using pen tools or brush tools of a raster graphic editor such as Microsoft Paint. In short, it is the option how heavily we weigh the way to manually put pixel by pixel, the traditional, and perhaps artisanal, way to create pixel arts. Both options should be decided not in terms of what is a more correct definition but of what is more useful. In other words, it is the matter of how wide or narrow meaning you want to give to the term “pixel art” for your own purpose. Though I am not committed to any option that I offered here, I think that it is important to add optional extensions to the theory to be useful for different purposes.
*20
For convenience, here I take the alphabet in written English as an example, but the same is true for spoken English. The word “cat” in spoken English, for example, consists of phonemes /k/, /a/, and /t/, each of which means nothing by itself.
*21
See Daniel Chandler, “Articulation,” Semiotics for Beginners, last modified July 5, 2017,
http://visual-memory.co.uk/daniel/Documents/S4B/sem08a.html.
*URL links were confirmed on May 11, 2020.
/div>“Glider Gun,” a pixel pattern generated by Conway’s Game of Life.
From https://commons.wikimedia.org/wiki/ File:Gospers_glider_gun.gif.
In order to examine the features of pixel art, we have to understand the features of its superclass. What is a picture? This question has been discussed in the philosophy of depiction, a branch of modern philosophical aesthetics. There are several approaches to the nature of pictures (*10). The approach I will consult here is one called a “structural account,” which attempts to understand pictures as a kind of symbol (*11).
According to the structural account, words, musical scores, maps, diagrams, and pictograms are kinds of symbol in that they all represent things (persons, objects, events, states of affairs, relations, thoughts, and so on). Pictures such as paintings, drawings, and photos are also a kind of symbol exactly in the same way. Then what makes pictures differ from other sorts of symbols? John Kulvicki, the proponent of structural account, lists four features as distinctive of pictures (*12):
1. A little change of property in a pictorial symbol leads to a change in its identity and content. Other kinds of symbol do not have this feature relatively. In the alphabet, for example, it is usually not the case that an instance of the letter A becomes that of other letters just because its shape changes slightly. In contrast, in drawing or painting, any minor change in the shape and color of a mark on the surface can affect what it depicts.
2. In pictures, relatively many types of properties are relevant to a symbol’s identity and its change. In a line graph that represents monthly average temperature, for example, only the shape of line is relevant for it to be a certain symbol. In contrast, relevant properties for a drawing that represents a mountain ridge includes not only the shape of line but its color, thickness, brushstroke, and so on.
3. In pictures, each symbol is assigned different contents. As a result, there is little redundancy such as synonymy, in which different symbols have the same meaning. This feature becomes clearer by comparing pictures with pictograms. As far as we see it as a picture, the desktop icon of trash bin depicts a trash bin as having rich and specific features, e.g. gray, metallic, wire meshing, cylindrical, etc. But if we take it just as a pictogram, a part of the graphical user interface, then it stands for only a poor content, the function of deleting files, regardless its unique color and shape. So are pictograms for restroom. Unlike pictures, in pictograms, the variation of symbols do not correspond to the variation of meanings.
4. In pictures, properties of a symbol basically straightforwardly reflects in properties of a thing it represents: Briefly, a symbol generally resembles its content (*13). For example, if you draw a circular figure with blue ink on paper, you depict a round blue thing (perhaps placed in three-dimensional space). On the other hand, even if you use a blue ink pen to write a sentence with round letters, you do not represents something round or blue.
Kulvicki calls the first feature “syntactic sensitivity”. It is obvious that this feature is very like what Sukura points out as a feature of pixel art. Taken literally, Sukura describes a feature of pictures in general rather than that of pixel art. I do not think that is not what Sukura really intended to say, of course. Accurately restating Sukura’s intent using Kulvicki’s framework, it seems as follows: A pixel art is more syntactically sensitive than a raster image that is not a pixel art.
Pay attention to the fourth feature, too, which is called “transparency”. Though neither Sukura nor Imai mentions, pixel art inherits this feature from its superclass: In pixel art, like pictures in general, if a color is put in a pixel, the same color is put in the corresponding part of a thing it depicts. This may seem to be trivial, but it is an important feature of pixel art as a kind of picture. Pixel font, a kind of font composed of a set of pixels distinguishable with the naked eye, does not have this feature because it is a kind of letter, not of picture.
Defining pixel art
In the previous section, I described the definitive features of pictures based on Kulvicki’s theory. Then what makes pixel art differ from pictures in general? In a philosophical jargon, what is the “differentia” of pixel art? I think both Imai’s and Sukura’s observations are helpful here: to consist of a set of pixels distinguishable with the naked eye, and to be relatively syntactically sensitive. But some additional supplements are also required for a well-formed definition.
First, the “pixels” here are not necessarily identical with pixels as constituents of a raster image. As Imai suggests (*14), pixels as elements of a raster image should be conceptually distinguished from pixels as elements of a pixel art, even though both can be the same in some cases. The former are minimum units to each of which a color value is assigned in raster data (*15). The latter are minimum units a set of which constitutes a special kind of symbol, a pixel art. Therefore, a constituent of pixel art, a pixel in the latter sense, can be each pixel as an element of a raster image (i.e. a pixel in the former sense), each set of them (e.g. 4×4 pixels), or each shape in a vector image (e.g. a square as a polygon), if each is distinguishable in the naked eye. Pixels in this sense even do not have to be in any digital image: Hand-drawn squares or colorful beads can be elements of which a pixel art is made.
Second, pixels as constituents of pixel art generally have the features as follows:
a. square, or at least shaped like a square; b. equal to each other in size, and arranged along (usually hidden) horizontal and vertical gridlines; c. each having one color.
The condition (a) allows for flexibility. For example, for a pixel art that is fine-grained and thereby is hard to distinguish its individual constituents, whether each pixel is exactly square does not matter. There is also a pixel art whose elements are rectangles rather than squares, such as Atari 2600’s graphics that may feel slack due to a part of its minimum unit being wide (*16). Moreover, elements of some kinds of pixel art are not clearly articulated as squares, rectangles, or even any polygonal shape: A pixel art may be made of cylindrical beads as well as of blurred or scratchy marks (*17).
The condition (b) is more essential to pixel art: You could make a picture by arranging triangles or hexagons without any gap, but it should be given a name other than “pixel art.” That is also why a mosaic is not a pixel art. In addition, as gnck points out (*18), a “pixel-art-like thing,” which consists of pixels but is not along any gridline or whose elements are irregular in size, is so awkward that the eye familiar to genuine pixel arts will refuse to count it as a pixel art. The equality of constituents in size and the horizontal and vertical arrangement are the necessary features for pixel art.
The condition (c) is with a reservation. When contemporary video games use a pixel art, it is common to add a lighting effect to it, letting individual pixels have gradation or vein. But even in such a case, it is natural to say that each pixel should have had only one color in the base image, that is, before the effect is given over it. This fact can be explained in terms of perceptual consistency. No matter how modified the actual color is, the base color is perceptually stable.
Summing up these considerations, I propose a formal definition of pixel art: A pixel art is a picture that consists of constituents distinguishable by the naked eye and is relatively syntactically sensitive, where the constituents satisfy the above three conditions (a?c); where “picture” refers to a kind of symbol that have the four features Kulvicki specifies (*19).
French linguist Andre Martinet said that all languages have “double articulation” as their general feature. Double articulation refers to that a symbol system has not only units in the level of words, that is, meaningful minimum elements, but also the smaller units that constitute words together but are meaningless by themselves. In English, for example, meaningful minimum elements are words such as “cat” while words consist of smaller meaningless elements such as “c,” “a,” “t” (*20).
Unlike language, standard pictures do not have the second level of articulation at least explicitly. In normal paintings or drawings, most parts in the surface that can be individuated as a unit will be meaningful: a part depicting a cat, a part depicting the face of a cat, a part depicting the nose in the face of a cat, and so on. In contrast, pixel arts explicitly do have the second articulation. Each pixel, which is meaningless by itself, is to pixel art what each alphabetical letter is to English. In this sense, pixel art is like language, though it is still a kind of picture.
The notion of double articulation has been frequently invoked in order to explain that language embodies two virtues: economy (made of a relatively small number of elements) and creativity (making up an infinite number of meaning) (*21).
Probably the same goes for pixel art. Pixel art is a kind of visual representation that looks simple on the surface but has a great potential due to the infinite number of meaningful combinations of the finite number of meaningless elements.
(notes)
*1
These constraints are practical ones due to the limit of video game hardware’s performance. The technological basis for pixel art is raster graphics as opposed to vector graphics and 3D computer graphics. Both vector graphics and 3D graphics had already been available in the 1970s, at least theoretically. Indeed, some Atari’s arcade games such as Asteroid (1979) and Tempest (1981) use beautiful vector graphics while Battlezone (1980) and Star Wars (1983) adopt a kind of three dimensional representation, though it consists of wire-frame models rather than polygon meshes.
*2
In arcade games, which were usually superior to console games in performance, video games using 3D models began to come out in the 1980s: Atari’s I, Robot (1984) is an early example. But it was the so-called fifth generation of home console that arose in the mid-1990s (e.g. PlayStation and Nintendo 64) that drastically broadened the use of 3D graphics for video games. Classics in this period such as Super Mario 64 (1996) and Final Fantasy VII (1997) symbolize the shift from pixel arts to 3D graphics.
*3
https://pixelartpark.com/ (in Japanese).
*4
https://pixel-art.jp/ (in Japanese).
*5
For an excellent overview of the diversity of contemporary pixel arts and their classification, see Shin Imai, “Gendai No Geemu To Pikuseru Hyogen [Modern Games and Pixel Arts],” in Pikuseru Hyakkei (Tokyo: Graphic-sha, 2019), 192-195.
*6
Masakazu Sukura (@scrama_sax), Twitter, January 17, 2012,
https://twitter.com/scrama_sax/status/159244346590371840 (in Japanese)..
*7
gnck, “Gijutsuteki Seiyaku Kara Bigakuteki Joken E [From Technological Constraints to Aesthetic Conditions],” in Pikuseru Hyakkei, 11.
*8
Imai, “Gendai No Geemu To Pikuseru Hyogen,” 192.
*9
Given that it is named “glider gun”, we may be inclined to say that this pattern depicts gliders and guns that shoot them. But that is like saying that a pattern on the moon or a particular arrangement of stars “depicts” animals. A part of the moon surface or a constellation is not a picture (at least not what I want to call picture here) but a thing in which we might happen to see something. For a related discussion, see Richard Wollheim, Painting as an Art (Princeton: Princeton University Press, 1987), 46-51.
*10
For a classification of different approaches of philosophy of depiction, see Catharine Abell and Katerina Bantinaki, “Introduction,” in Philosophical Perspectives on Depiction, eds. Catharine Abell and Katerina Bantinaki (Oxford: Oxford University Press, 2010), 2-6.
*11
The founder of the structural account is Nelson Goodman. See Nelson Goodman, Languages of Art, 2nd ed. (Indianapolis: Hackett, 1976), esp. chap. 1 and 6. John Kulvicki, who I will invoke, inherits and updates Goodman’s theory. Note that strictly speaking, structuralists focuses not on the nature of individual symbols but on that of “symbol systems” individual symbols belong to.
*12
John Kulvicki, “Image Structure,” Journal of Aesthetics and Art Criticism 61, no. 4 (2003): 323-40; John Kulvicki, On Images: Their Structure and Content (Oxford: Oxford University Press, 2006), chap. 2 and 3.
*13
In fact, Kulvicki makes so technical arguments on the fourth feature (transparency), according to which to equate this feature with similarity is highly problematic, if not wrong. But for our purpose, the crude simplification works well.
*14
In the above quote, Imai carefully writes that a constituent of pixel art is “each pixel or each bunch of pixels” (my italic). As Imai himself explains, the latter disjunct is added in order to distinguish pixels as constituents of pixel art from pixels as particles of digital image or display device.
*15
Strictly speaking, besides the level of rasterization as digital data format (raster image), there is also the level of rasterization as display format in monitors or printing devices (raster scan). Even image data stored in vector format are eventually rasterized by the operating system or the print driver when they are displayed on monitor or paper. Therefore, most images that we encounter today may be rasterized, or “pixelized,” except hand-made pictures or film photos. Whether an image is rasterized in this sense has nothing to do with whether it is a pixel art, of course. For a related point, see Alistair M. C. Isaac, “Digital Images: Content and Compositionality,” Journal of the American Philosophical Association 3, no. 1 (2017): 106-26.
*16
Aside from its technological background, many parts of images in Atari 2600 graphics seem to have a low resolution in horizontal direction. For example, see screenshots posted on a review site for Atari 2600, https:// videogamecritic.com/2600aa.htm.
*17
For a related topic, it is often discussed that in the early video game consoles such as the Nintendo Entertainment System, graphic designers deliberately created pixel arts taking into account their “blurring,” which results from their technological conditions, that is, a CRT monitor and/or an analog video transmission. See a series of conversations around this topic on Twitter (in Japanese), https://togetter.com/li/1131267. If that is correct, the notion that constituents of pixel art are supposed to be distinctly square would been a latecomer in the history of video games. But it could be said that even if old pixel images were all blurred, viewers understood that a pixel art comprises of a collection of squares and they imaginatively saw clear-cut squares beyond vague shapes.
*18
gnck, “Gijutsuteki Seiyaku Kara Bigakuteki Joken E,” 12.
*19
In addition to this base definition, I also offer some additional options. First, the option how strictly we take the condition of distinguishability of elements: It is possible that individual pixels in an image are too small or blurred to be hard to distinguish with naked eye. Whether we should still count such a case as a pixel art or not is the matter of choice. If we decide it is a pixel art, the condition of distinguishability will require to be looser. Second, the option whether we adopt how to create as an additional condition: On the one hand, we have the choice whether or not we should count as “genuine” pixel arts the pixel-art-like images that are procedurally produced by algorithm processing hand-made pictures, photos, or 3D graphic images. On the other hand, we have the choice whether or not we should count as pixel arts mere low-resolution bitmap images that are crudely drawn using pen tools or brush tools of a raster graphic editor such as Microsoft Paint. In short, it is the option how heavily we weigh the way to manually put pixel by pixel, the traditional, and perhaps artisanal, way to create pixel arts. Both options should be decided not in terms of what is a more correct definition but of what is more useful. In other words, it is the matter of how wide or narrow meaning you want to give to the term “pixel art” for your own purpose. Though I am not committed to any option that I offered here, I think that it is important to add optional extensions to the theory to be useful for different purposes.
*20
For convenience, here I take the alphabet in written English as an example, but the same is true for spoken English. The word “cat” in spoken English, for example, consists of phonemes /k/, /a/, and /t/, each of which means nothing by itself.
*21
See Daniel Chandler, “Articulation,” Semiotics for Beginners, last modified July 5, 2017,
http://visual-memory.co.uk/daniel/Documents/S4B/sem08a.html.
*URL links were confirmed on May 11, 2020.