It's simple:

Pixel art likes low color counts. With a finite amount of colors, you have to choose ones that will play well off of each other and be usable in the most situations.

Whereas in regular CG art a thousand shades of blue can be used across the entire piece with only 1 color shifts, you need to minimize these to one. But, let's say you're reducing colors and find the blue and violet you're using aren't too different and not extensively used. Taking a shade between these two reduces the color count without effectively changing the piece.

Color theory is hard to explain in such a small paragraph, as it is basically unique to each situation. If you want more, go Googling for more in-depth answers.

In all media, color is an important aspect for any artist. In pixelart nothing is more intricate and complex than color and is especially important. Regardless of your experience or skill level ALL of us struggle with it all the time on every piece. Even those who garner the respect of others and seem to be experts struggle with color.
It is important to understand that color will always remain mysterious and even frustrating. Achieving a certain level of comfort, even minimally, with colors when creating pixelart frees you to be creative.
You need to know the most basic color theory knowledge...start HERE or HERE...there's plenty of other in-depth sites with more advanced articles for you to read and study. In pixelart where trying to use the least amount of colors then it can seem even more daunting but it really isn't. When it comes down to it...Does it look right is all you need?

Finding and developing your own neutral ramp is a very valuable thing. Once armed with this valuable tool your mind is free to create and express yourself. Staring with your own base freely allows you to color and adjust, shift, add and refine as you work without getting all stressed about color which is often the main reason a piece ends up in the forgotten/dump pile.

fool is highly admired by all so studying his color use should be helpful.
Notice how he basically uses one ramp in all his work (first panel). This ramp is always changing depending on the lighting, mood and the scene. When looking at fool's work, black plays a crucial part even when he uses no black. He tends to start at the darkest color (panel 2) of his ramp and flows to the lightest color in whatever color direction he needs to go. This makes his pieces have a level of contrast that is prominent even when very little darks are used. Whatever color ramp he works on there is always at least one neutralizer from the basic ramp (panel 3) to seam it all together.


skamocore, made an average of all the colors in fool's gallery:


...an updated version of Fool's averages: *mysteriously small*




Edited by jalonso - 25 July 2014 at 6:05am

This was posted by onek on a recent WIP thread and is worth sharing here.

onek: ...a technique for nice, natural color ramps that I'm recently using, is to spice up the colors simply by putting some 10% opacity fill of (full!) blue, magenta, red and yellow, from dark to bright, the result then (because of the full tones) looks very comic-ish, (sometimes thats nice too...)... if I'm not satisfied, I pretty much try to edit the colors as I would do editing a photo, meaning 'color balance', 'hue/ saturation', contrast etc editing...
maybe this makes it more clear



Edited by jalonso - 25 July 2014 at 6:07am

Neutralizers are the key to minimizing color count as one single shade can easily replace many others. Even the ones that don't make sense. Its key in great pixelart.

Adarias has one of the best understanding of colors of anyone I know. He is far too intellectual for me to understand most of the time but in its simplest term, he lives and dies in 2 or 3 neutrals and expands from there always bringing his neutrals in most if not every ramp he uses.


Great insight by him is given in his PJ feature:
Adarias: The important thing about neutralizing colors is that the added color (should) be the exact opposite hue as the color it is meant to kill. If you only use gray to neutralize, you won't ever achieve gray unless the piece is 100% gray. Neutralizing colors must be far enough away that they balance TO gray. If your second color is exactly opposite from the other, it should occupy 50% of the space (equal parts) because neutral gray will be exactly halfway. I made this quick visual to explain this.

Of course, if one or the other is highly saturated, you need far less of it (because it will dominate). Because computers are what they are, the fact that it says colors are opposite hue and same saturation doesn't mean it's telling the truth. From different angles these will turn all sorts of colors and are best seen on a perfectly calibrated CRT monitor. Of course you don't always want flat gray, and I certainly don't measure. Typically I just want to reduce contrast while increasing excitement.  I often go overboard on this :).  A good example of when I kept things under control is the bronze bust piece, which contains only 4 shades that could be considered bronzy, and relies on the other 11 colors to make the piece.  Because the bronzes were left and the other colors generally warm, instead of canceling out the color completely, I felt it created very active piece that is not nearly so saturated as most metal renderings. (Bronze pixel is shown above).
Lately, blue has been my color of choice for taking down while activating, although you can see that this blue is not the only color that may be used.  It's great for midtones or shadows, but sometimes it just doesn't belong. Occasionally, you will see reds in my shadows (to keep the sense of light/heat), or teals (which are calmer than blues), and lavenders mix into my highlights sometimes to make them a little less BAM (this is particularly true of soft objects like cloth, hair, and skin).
There are other applications of this thought too: Triadic blending (where the color wheel is split 3 ways instead of 2). The bronze bust is another good example of this, where green and purple move around orange. This is another example of where practice plays an important role in pixel art.  All of these concepts come straight out of experimenting with paint. My weakness for burning colors comes too from traditional paint too - paint is pigment, not light, and therefore wont 'burn' the way my pixel colors often do. Tetradic blending works theoretically, though often if I use more than 3 of the 6 generalized colors (primaries and secondaries), I'm using too many colors to count.

His simple explanation to me is:

*  a neutralizer is basically any color that would, if blended with another, bring the result closer to 0 saturation. For example, a brown in an orange ramp - is 'more' neutral. Or a green in a magenta ramp, because for that ramp to approach green, it would have to cross neutral (or more neutral) tones.

The real decider is whether the tone used 'behaves' (helm©), meaning it 'stays in the ramp' or doesn't which will create its own distinct shape and stands out. If it doesn't behave, it will add busy-ness and break the form and doesn't neutralize anything -- in fact, it's more likely to polarize the two hues you are trying to neutralize. An example of this is the difference between dithering (a field) and checkerboard (interlocking shapes seen above). Same idea, different scale - different effect.

Edited by jalonso - 25 July 2014 at 6:10am

@ mashimi, they go off like that because it seemed to make sense to show where ramps begin, end and have neutralizers within them. I don't think anyone does that with their palettes its just to illustrate my obvservations.
Most do (or should) be doing their colors like this



Edited by jalonso - 25 July 2014 at 6:11am

This thread will assume you know/understand the very basic language of color and color theories. If not, these are some basic pages with visuals.
color wheel with color schemes, plus tint (addition of white) and shade (addition of black)
'nother color wheel with a more interesting shape :P
more useful explanation of schemes/etc
color chart showing chroma/hue/value as 3d model
explanation of common terms
color temperature (should make a note that context matters, and "warm" colors can sometimes be "cool", and vice versa.
and this, which explains several useful things and is a little interactive.
some interesting things here too (ugly site)

Not specific to choosing colors alone but this Pixelation thread is related enough and very, very informative. Must reading for all!!

Edited by jalonso - 25 July 2014 at 6:13am

This is DawnBringer's take on his palette, which may be helpful.
... This time I followed that old path of making well balanced palette suited for game-gfx. I sacrificed a bit of the dark register for some brighter colors. Tried to optimize for many smooth ramps and mixes while still having useful colors on their own.
Here's a handmade "ramp-scheme", and a Hue-Brightness plot + isometric colorspace-cubes made with scripts in Grafx2. (The brightness is perceptual according to my formula: sqr[(0.26*r)^2, (0.55*g)^2, (0.19*b)^2] , multiplying with 1.56905 normalizes to a scale of 0-255... it's better than Luma Y'601 according to my eyes at least).


And finally a little Mockup color test:

But...I notice that some aspects of the palette is not used to its full potential; esp. how the blue & greens can be used to interpolate the greys and even create new shades. So I made a little example to illustrate my intentions. I took the fredoom the redraw snippets from a few tiles...this is not in any way intended as critique, just examples of some possible AA/mixes/methods.

Working with colors in RGB-colorspace is not all that simple & intuitive. Looking at what's going is a bit clearer when viewing HSL/HSV colorspace, that is HUE, SATURATION and BRIGHTNESS (Lightness, Value, whatever...).
But it's pretty important that brightness is adjusted to the perception of the human eye (phenomena I'm exploring a lot). But  I don't recommend that you tweak individual colors using a HSL/HSV model; good 'ol RGB is more exact and probably more familiar.

Smaller and balanced palettes usually have colors of quite similar saturation, so you can focus on the two more important dimension: Hue & Brightness.

When observing the following example I think it's quite clear what a nice tool Hue-Brightness diagrams are for palette evaluation and optimization. (Note that Hue is cylindrical and wraps around).



Edited by jalonso - 25 July 2014 at 6:15am

More hardcore palette play by Dawnbringer

From Topic:
DB's 16 Color Palette v1.0

Edited by jalonso - 29 July 2014 at 5:45am

Some things you should be aware of is that the most important aspect/dimension of any color is its brightness. And the the color-channels acts vastly different in this respect: Green is roughly twice as bright as Red, and Red is about twice as bright as Blue. This means that adding +5 in Blue may not produce a noticable change but +5 in Green could be a huge leap. So consistently adding X in your ramps might not produce a smooth/uniform look. Also, aestethically, it can sometimes be better to avoid strictly linear ramps.

syosa generally uses fairly desaturated palettes, they're just high in luminosity. if you want an example of how to use high-saturation colors well, check out Delicious' work.

the brown used as aa in the green of the mohawk isn't working, brown is too warm for this purpose and so it burns against the greens. yellow is a slight jump from the greens your using, I probably would have used a light green to highlight the greens (yellow-tinted, but still green), so long as I have room in my palette for that color. as it stands the yellow is a bit warm.

neutralizers are one thing I've never really understood that well, I'm also sort of in the dark as to when and where they're useful. I guess if you needed a grey but don't have one in your palette? maybe when you add a color for visual interest, but don't want it to read as the actual hue of the object, you neutralize it with a color opposite in hue?





Edited by jalonso - 25 July 2014 at 6:21am

This might seem like a silly follow up question, but...

When artists talk about  adding black or white to a color (shades and tints), how exactly do you do that in a program like photoshop?  Do you just change the saturation?  The brightness?  Or both?  Or is there some other method that is easier when transitioning between lighter and darker versions of the same color.

Just curious: can someone put together a brief thing on how to find and properly use neutralizers? I read all the stuff in this thread (and some stuff in other places) about them, but I still can't seem to wrap my head around the concept.

From what I have read, the term "Neutralizer" seems to be used interchangeably (and confusingly) for two different operations:

1. A low saturation color that can work as a shade in multiple ramps of different hues (versatile interpolator).

2. The mixture of two complementary colors (opposite on the hue-wheel) to accomplish a grayscale. For an example study the metal objects in this (DB16) image where I use this technique extensively.

If I understood things correctly, it doesn't seem very practical to use the term "Neutralizer" for both operations, rather only the 2nd one.


Edited by DawnBringer - 18 December 2014 at 12:19pm

Original post by andrae:
When I was starting at pixel art, I tried to emulate how other artist choose their palette. I tried to study and copy the palette used by syosa(pixeljoint) and uruchi(pixiv) but when I need new colors I develop my own color ramp.

I usually start a sketch using a gameboy palette. This helps me in placing the colors. Then I think of what materials will be colored. Based on these, I choose the base colors. Usually, I use the same linear color ramp for two materials, for example metal and cloth. To differentiate them, I choose darker colors for metal while lighter for the cloth.

For the color ramp, It starts with a linear ramp of brown that starts with a violet and ends with a white/cream. Then I add another ramp which also starts and ends with the same colors, but can use different ramp as routes.

an illustration:



Edited by jalonso - 25 July 2014 at 6:29am

Color-theory101, Hue-shifting palettes and stuff Tutorial
http://www.planetminecraft.com/blog/1000-subscriber-special-color-theory-101-hue-shifting-palettes-and-stuff/

A nifty tool to make color changes on pixelarts
Pixel Palette - Tool for color palette


Edited by jalonso - 29 July 2014 at 5:47am

There is no method that will represent every real-life scenario accurately, nor will it be the best solution for every piece of art. A very generalized rule of thumb that I use is that highlights and shadows are less saturated than midtones, but it really depends on what sort of lighting you're trying to portray. Specular reflections usually retain the color of the light source, which tends to desaturate the areas of highlight. I find that the saturation tends to increase from there on until it drops off in the areas of deepest shadow. If there's any fill, bounce, or reflected light (which pretty much always is), that will impart its own characteristics upon the areas of shadow. Also, the hue will shift along with the saturation (although you probably know this since you referred to it as "hue shifting"), and changes in hue correspond to changes in perceived value since some colors are inherently brighter than others.Incidentally, I just published an article about relative color that touches on some of those topics, as well as others that have been discussed here in the past. Hope this was helpful to you at all! :X

Edited by jalonso - 12 June 2015 at 4:56am

I don't know if this belongs here or not, but I had peoples on DA ask me about my palettes often enough that I felt it warranted a tutorial/ramble page about how I go about creating my palettes

http://i.imgur.com/tpz4o10.png


Edited by rhlstudios - 04 May 2015 at 7:26pm

bump

Here is my palette. 1st, 2nd, 3rd and current 4th attempt:


It is hard to organize a palette on a 2D space when you are working with 3 variables (HSL/RGB). My actual palette array is a mess:


However, my palette is getting better once I started using HSL instead of RGB, though I still use RGB sometimes.

So, what I understand is that from your base color, if you go darker, you should go towards colder hues and decrease the saturation and value. If you go lighter, you should go towards warmer hues, decrease the saturation, and increase the value. This will give you more realistic looking colors. It's called the arch method.

For more cartoony colors, you'll increase the saturation if you're going darker.