Pixels are not square.

[SKELTON S. SKELETON]

Kevin G (@applesaucers) on Twitter pointed out this astutely argued article (AAA, truly), that explains why modern ideas of "pixel art" are so wrong-headed. http://alvyray.com/Memos/CG/Microsoft/6_pixel.pdf22

In sum, what we're seeing is not what we're intended to see. Pixels aren't square; they're overlapping fuzzy points, that are fed into a display system. And the pixels designed for a CRT are designed for the artifacts and properties of a CRT display. For instance, the strawberries in Pac-Man

There are no white dots in this sprite...or are there? 


We're not meant to see the raw pixel data any more than we're meant to read raw code unless we're researching how a thing was programmed, or listen to the raw takes of a band's recording before they're engineered into an album. I mean, you can -- but if you do that you know what you're getting into. And for older games, I don't think many people do know.

To that end, post your favorite examples of things that look just fine or otherwise great on a CR, then then dissolve into bland puddles of color if you look at an array of the raw pixel data.

Samus Aran


Ninja Gaiden 1

The green is meant to melt away and turn into cracks, accents, shading. The orange is supposed to bloom and become a highlight. Look at the texture and details in her helmet, in the cook of her knees. Look at the awesome shading on her gun arm. Look how you can actually distinguish her shoulder, and the parts of her gun arm. And then look at the undifferentiated mosaic that the purists will call perfect and clean. Missing so much of the intended signal.
Just, the bricks. Man, look at the difference. How many colors, how much texture, are we missing out on?




Also, this illustrates how solid outlines are meant to work for NES games. In an emulator, Ryu looks like a Colorform19being shifted around on the backdrop. On an actual NES, on an actual TV, he has shading and detail. the outline serves both to differentiate his anatomy and to blend him into the backdrop -- as opposed to set him apart, as it does on your LCD monitor.


[

---

I am including some other examples:
Teenage Mutant Ninja Turtles (arcade)



Without scanline and CRT filtering, it looks like the turtles themselves are odd, blotches of color with shading that makes parts of them look bizarre if not grotesque. The foot clan fight in a sewer over an abyss of empty purple space. The sewer walls and extra items look too new and unworn and stand out oddly. With filtering, the foot are now clearly standing in a mess of filthy sewer water. The pipes and walls behind the turtle have become appropriately worn. The turtle himself looks smooth and like a cartoon character, the shading and details blending together.



Golden Axe (arcade)


By all accounts, most of the pixelart from games of this era doesn't jive spectacularly well with the aesthetic focus of modern pixelart. They look bright, un-contrasted and fuzzy with a softness that recalls childlike enjoyment and pumping quarters into a machine eternally. The fire stands out in the first image: it is a collection of square textures of different colors that give the appearance of fire but the hardness of the graphics stand out and make it look like a mess.


But like this, contrast that previously didn't exist is there: the scales of the creature our bikini-lady is standing on become more prominent and the grooves darker. The fire is an spectacular side to behold, brilliant and textures and filling up the whole screen.

[Filler]

The Ninja Turtles one really shows how nice it looks on original screen, rather than a crisp screenshot.

[Chris2Balls [:B]]

CRTs are amazing. I'll have to dig my CRT monitor out sometime.
As LCDs have become the most common kind of display, I think pixel artists have been trying to work out what works best on it, as they can't use CRT tricks on them anymore (apart from if they're emulating a CRT display).

[Raccoon Sam]

This is an interesting subject (and also guaranteed replies at /vr/)

I play my games exclusively on a CRT

as they can't use CRT tricks on them anymore (apart from if they're emulating a CRT display).

Not yet! CRT tricks are coming back with a vengeance as soon as High-DPI displays are mainstream, because for the first time ever, game and emulator developers have the chance to use non-linear interpolation modes.

My phone is a Samsung Galaxy Note 4 and its display resolution is 2560 x 1440. Its individual pixels are imperceptible, and if I were to emulate Game Boy (160 x 144) on it, instead of filling the 1440 horizontal pixels simply with 9 x 9 clusters of the same color, I could write an intricate algorithm to produce three bands of varying dimness of red, green and blue on a black background, something like you'd see if you stuck your face up to a CRT television, for each pixel.
The concept is still in its infancy and 9 pixels aren't much to work with, but think about ultrawide 4K monitors - their horizontal resolution is 5120 pixels. That would give you 32 pixels of horizontal resolution to work with for a single pixel.

The future looks good imo

See also: what could've been!

[SKELTON S. SKELETON]

As this thread makes us reconsider attitudes about pixelart, I'd like to vote to have this thread stickied so it could have some semblance of permanence

[CyberneticCupcake]

It's a reasonable theory, and the comparisons show your point most eloquently, but I would argue that it's up to the artist to decide what appearances are most suitable to their work.

True, CRT displays show a brilliance in sprites not present on LED, as you've shown in your CRT captures, but your signature image, Skelton, looks beautiful on my LED monitor even with its square pixels. While I agree that square pixels to show "retro" designs is an overdone cliche, I think that what many computer artists are calling "binary art", without anti-aliasing, is a visually interesting technique, if not always suitable for many subjects.

http://fjolabjorn.deviantart.com/art/Bin...-607352182

[Gors]

While pixelart was created in a time when CRT was the norm and with "looking good on the final display method" as their main goal, this doesn't mean that pixelart has been done wrongly all this time and that we should reconsider.

It is true that pixelart was intended to look better on a CRT as you brought up (the oddity and innacuracy of CRTs helped improve the apparent lack of care when placing pixels to form a much more pallatable image), our screens have evolved and thus many of those techniques are rendered obsolete by now.

It's also important to note that we, the internet community, "really understood" what pixelart was about when sites like The Spriters Resource became popular. I doubt we really knew how game graphics really worked when we were kids, and I certainly doubt that any of you really knew that sprites were stored in 8x8 tiles with 4 colors (1 alpha) method in the early days of the NES.

When the first sheets were ripped, future pixelartists studied it and after zooming it 8x, it was clear that the stored image was indeed saved in the ROM as squares - so we just ran with it disregarding the natural distortions of the CRT. This coincidentally happened in tandem with the development and advent of LCD and LED displays, which improved pixel quality and mitigated any use the CRT had prior.

Pixelart was done differently way back - but like any kind of art, it's constantly evolving (especially an art which is heavily tied to electronical and technology progress) and different mediums require different techniques. You can't deny that a lot of modern, square pixel pixelarts do look good, and thus I don't think it's right to say it's "wrong-headed".

After all, pixelart is Art, and there are no rights and wrongs in Art if it manages to perfectly convey what you want.

It's very similar to comparing mosaic art to pointillism and "ponto cruz" embroidery:

(an example of mosaic)

(an example of pointillism)

(an example of "ponto cruz" embroidery)

They all work on the very basic principle of rendering bigger images with small unitary pieces (the mosaic being small stones; the pointillism painting made with regular dots of color; and "ponto cruz" made with embroidered squares). However, due to the nature of each media (rocks are never regular enough and thus they bunch up irregularly; paint dots are regular but due to having no grid on the canvas they bunch up like clusters, improving readability; and ponto cruz is literally physical pixelart), each style has its set of techniques, advantages and drawbacks.

Despite this, they're all valid types of art with no real "correct" method - they only work on whatever media the artist was limited to, and now that we are virtually limited to LCD/LED screens with very definite pixel grid (not considering CRT emulation filters), square pixels are simply the logical step on how we deal with pixelart nowadays.

All in all, it's a very interesting talk and pixelart has evolved quite a lot from being physical mosaic to completely digital pieces of art, but to be this adamant about "pixelart not being used as it was meant to be" is kinda like saying "mp3 shouldn't exist because real instruments exist"

[CyberneticCupcake]

[T]hey're all valid types of art with no real "correct" method - they only work on whatever media the artist was limited to, and now that we are virtually limited to LCD/LED screens with very definite pixel grid (not considering CRT emulation filters), square pixels are simply the logical step on how we deal with pixelart nowadays.

All in all, it's a very interesting talk and pixelart has evolved quite a lot from being physical mosaic to completely digital pieces of art, but to be this adamant about "pixelart not being used as it was meant to be" is kinda like saying "mp3 shouldn't exist because real instruments exist"

Exactly. While I'd like to see more experiments in "non-square" pixels, this is more for artistic consideration than for any sense of "superiority".

[Pik]

I'm not sure a sticky is necessary, since it's an opinion thread and shouldn't be presented or implied as a rule.

Some people take advantage of the grid and make their pixel art based around it, others, like the retro games you cited, have the CRT display in mind first and foremost. Pixels aren't squares? LCD pixels are squares - the end result of graphics made for LCD screens take advantage of this.

I think a more relevant point is "well, we shouldn't judge old graphics by displays they weren't meant to be seen on" and I agree.

By the time crisp screens became the norm, the methods of showing detail changed. The evolution of pixel art has taken this into account, and even new things done in very limited colors use modern techniques because blurry, nonsquare displays are very rare. It would be nice if more people made stuff for old displays, but to claim modern pixel art is "missing something" because of this doesn't really go anywhere because the techniques serve two different eras. Modern pixel art isn't necessarily a revival of a retro artform, it's very much a new thing now.

[jumbods64]

Yeah. Although a pixel isn't a little square, sometimes it is, like in enlarged pixel art or on aliased low-res textures. And sometimes it's a fuzzy, bloomy bit like on a CRT. And sometimes it's red, green, and blue elements in that order, like in 1:1 scale LCD pixel art. A pixel is just a picture element: an element of what ever quantized picture format you're using.

[Chris2Balls [:B]]

Relevant article about CRT simulation, by Ian Bogost