GR#189 - Monster Girl - Anatomy, Shading

[Jesia]

Hello everyone ^^

So i just started pixel art. I have been doing digital and traditional drawing for a while, and decided that i would try out pixel art as well. I quickly realized that it was something befitting me, so i have decided to develop it further. Yesterday and today i created two characters, after having looked up a bunch of tutorials. I am still very unclear as to what is right and wrong in the pixel world. So a little guidance would be nice.

#Note that these two characters aren't meant to be used in any game. Their purpose is to be "eye candy" haha ^^


They are both based on some of my old designs, from a few years back

[Pix3M]

Eye candy?

If that was your intent, I think we can be more artistically brave and go with a saucier pose 

Otherwise, they come off as stiff pictures meant only for character reference.

[astraldata]

I agree with Pix3M on the pose. In addition, I would suggest to limit or totally eliminate the dithering on pixel art that size unless you are going for a technical limitation of some sort on purpose.

Since you stated your intent was "eye candy", you should add more colors to buffer the transitions between shades, preferably implementing hue-shifting when you do. This would increase its visual appeal depending on how well you implement it.

Another issue I have is that there are small highlights in odd places that don't seem to add anything to the image. It's best to either leave these out completely or increase their area to include a larger cluster of pixels instead of dotting them here and there (see your chars' knees, elbows, and wrist joints). On pieces this large, a pixel or two has a much smaller effect in providing contrast you seem to be trying to apply.

One last thing is that stray single pixels on the outline silhouette is something you ought to avoid where possible. You did mention this wasn't for game art, but if you wanted this to work on darker backgrounds, it simply won't without double or triple pixels around the outline of very thin parts of your characters to allow it to 'blend' into black.

This technique isn't talked about too much in tutorials, but looking at old SNES sprites, particularly RPG character sprites, you will see that they all tend to look fine on different backgrounds of any shade or color, including the exact shade of their outline color.

To illustrate this, if you picked the color of your pink character's horn tips (I'm assuming that's a horn of a goat or something), and filled the background in with this exact color, chances are, your image will look really bad in the sense that it has no real form anymore aside from the inside colors, which should be used to create a sort of "reverse outline".  This technique will cover scenarios like that. Unless you know exactly what BG color your art will be placed on, you should keep this in mind -- even for non-game-related pixel artwork.

[Jesia]

Thanks to both of you for replying. Some of the small odd things that you have pointed out, like the small highlights, come from the fact that i usually paint more realistically, and i guess those sorta small highlights just don't read well in pixel haha (same goes for the small ones on the knees and such, cause i wanna convey the natural texture of skin, guess i shouldn't) I'll try and stay away from that.

I got a few questions, regarding some of the things you pointed out.

- Is dithering really that bad in this size? Cause i find it aesthetically appealing. What size is it then ideal for?

- I also thought that too much color was a no go when it came to pixel art? I purposely kept myself restrained not to use too many colors. Is there are ratio between pixel character size and the recommended number of colors?

- I'm at the moment working on creating an actual pose, but i'm not sure i can lose the stiffness, as i feel pixels in generally makes it still. I can't seem to hold on to the gesture of the original sketch. Any tips or ideas as to how to proceed with that? 

Thank you again for taking your time to comment ^^

[Probo]

It is unusual to have lots of dithering on large sprites. although dithering big sprites isnt intrinsically bad, it could be a stylistic choice. but you should know exactly why youre doing it.

what looks off to me, is that the dithering on the cloth of the purple lady for example, seems at odds with the shading elsewhere on her.

[coffee]

Dithering is basically just used when you have a color restriction, or like Probo said, for a personal challenge or style. However, like people here said, it tends to look pretty bad for a trained eye
when you don't really know what it is used for. It's basically for blending two colours together to eliminate some size to the image file, by creating a third colour for example by blending two colours together, or make a nice gradient with minimal amount of colours used.

Here's a good tutorial:
http://www.spriteland.com/tutorials/pixel-art-dithering-tutorial-what-is-dithering.html]]http://www.spriteland.com/tutorials/pixel-art-dithering-tutorial-what-is-dithering.html

[Jesia]

Kay so i took everything you guys had to say and made a new version of the first character.
- saucier pose
- Less stiff (tried to make it less stiff atleast)
- No dithering
- more colors
- no small highlights
- no small pixels in outline

I hope it is any better

[32]

Absolutely fantastic improvement 

I think if you just keep on working on stuff and studying some pixel art you'll be fine. In terms of pixel art specific stuff you just really need to focus on keeping the edges of your colour clusters clean and descriptive. What I mean is to make sure your lines are as smooth as possible and making sure your shading describes real topography.

I would suggest seeing what happens when you try to remove outlines, they are far from necessary and pixel art can just be another form of painting if you want. I think in particular that the line along the bicep looks quite out of place, mostly due to how straight it is though. Definitely look into anti-aliasing, it will help immensely in adding fine detail to your lines and shapes.

Read Helm's post on banding and do a forum search if you don't understand it for a wealth of information on the subject. You're not offending too badly but there are some places, especially the face, which are suffering from it.

Lastly, it's pretty clear that you aren't going for completely realistic anatomy but your characters legs confuse me. Are they supposed to have really tiny feet? Do they have two leg joints or just bulgy ankles? I think you would be best off when inventing anatomy to really think about how it functions in terms of joints, bones and muscles. Also her crotch looks misplaced.

Can't wait to see the next update .

[Jesia]

Thank you for the nice comment. I'm happy to hear that it has improved. About the anatomy... It was supposed to be semi realistic, but i am having troubles making curves with pixels, which makes especially my arms and legs look strange. I will work on it the best i can. balancing a natural feel and flow and still keeping an eye on jaggies, is harder than i imagined xD

I will make sure to research more about banding and anti-aliasing. ^^


So i decided to try making a simpler character. Smaller and without an outline.

Original size

Resized x3

and i thought, what the heck, why not try and animate it xD It's most likely a bit stiff, but it is also my first 

Orignal

Resized x3

wasn't really sure what to do with her other leg :/

[r4c7]

Edit:
Mainly tweaked the colors. Heightened contrast and saturation. Some other color and design ideas and tweaks. I don't know if you were going for pastel colors, but it looked washed out. Unless they are supposed to be a ballerina, I'm guessing they should be more solid on the ground. I also messed around with the lighting a bit, but that isn't necessarily the best, just something to think more about.

[astraldata]

versus



As you can probably tell, not a lot of visual difference at first glance, however I reduced a TON of colors on your image -- I took it from 86 colors down to 54, though I'm not sure if the alpha channel was included in that color count or not. Either way, this is a pretty huge difference for not much lost in visibility or appearance.

As you can probably tell, all I did was take some colors with similar value and replace them with other colors of a similar value. I didn't go through the entire sprite because it's getting late here, but I went through most of it. You should look very closely at the differences between yours and mine to see which colors I replaced and which colors I left alone. I did no edits whatsoever to the image or coloring itself outside of color replacement (except for that bottom leg shading, which was bothering my OCD pretty badly).

After messing with this sprite, it must have taken you a significantly longer time creating it than I spent on editing it (which must have been a long time ), but, in reality, a sprite with this style should really not take so long if you simplify the number of colors you use to the bare minimum necessary (i.e. simply avoid having to use dithering by creating a palette with 3-6 shades depending on how much you use a color).

A good rule of thumb is to keep your sprites between 16-20 colors maximum (including transparency) to ensure maximum readability, and blend the colors between one another to ensure maximum unity of the palette. There's nothing inherently wrong with using more colors, it's just that you should keep in mind that the more colors you add, the more difficult it gets when you want to change those colors (especially a shade or value) because you can quickly end up with a huge number of colors just by adding or slightly modifying a shade/hue here and there and forgetting you already have one pretty close to that color/value (thus throwing off the unity of the entire piece). This is what makes the number of colors blow up exponentially, such as what happened in your image.

I admit, my color choice sucks in which colors I kept (i.e. I kept the more grayed-out monotonous/dull colors over the vibrant ones), but I wasn't thinking about that at the time of the edit. Instead, I was simplifying the groups of colors and eliminating banding to show you how to do the same thing with minimal anti-aliasing (AA). I could have simply tweaked the shades I kept to be a halfway blend of the dull color and the most vibrant one (example: the 2 colors that gave the hint of yellow lighting on the grey fur in the original image, for example).

And just fyi, I think your 'flat' color choice for your animation was actually very well made (you obviously have an amazing eye for color) -- it provided much more unity in the colors than r4c7's edit. Once you perfected your motion, THEN give it more form/contrast via lighting/shadow (no need for any contrast, etc. until that time). I'd suggest, as far as the animation itself goes, roll the ball a little more and give it a slight bounce or two to indicate she isn't kicking a beanbag/hackey-sack (unless that's what you were going for, then by all means, you've got it correct).

Hopefully that helps.

[Probo]

here it is with 17 colours, save you a bit of work reducing it down.

using transparencies like that isnt a great idea, the colour count goes through the roof! theyre fine to play around with and experiment, ive often looked at how colours blend. but you should choose and add the colours yourself for the final piece.

an important thing to bear in mind to keep the colour count down is to think how shades can be reused within the picture, the flat colour from one area could be used as a shadow elsewhere, and so-on. then you tweak colours en masse as you go along. you might actually change the colour of a piece of clothing or something so that shades can be recycled elsewhere for example.

edit: i should go into more detail about what i changed. i changed the shadow at the crotch a bit, so that light wasnt getting right to the bottom of it, and used one of the lighter shades from the tights as a bit of AA where the flat skin colour meets the shadow. I lost most of the browns of the hair completely but if you choose carefully you could put one or two back and have it look brown again. I used the colours from the tights and pants to add shadow for the skin, and theres only one skin highlight colour.

[Jesia]

Thanks Astral for taking the time to change my sprite. Perhaps i took your suggestion about using more colors a bit too far xD I will try to find a balance in my use of colors, and stick with those 16-20, which might be a really wise idea. I really like the challenge of working with a limited pallet. Also many thanks for the compliment about color choices. Choosing colors is my favorite part ^^
I will see if i can give the ball in the animation a bit more life, thanks for the suggestion ^^

The same goes to you Probo Many thanks for taking the time to modify my sprite. It was purely me going way overboard with colors. I will restrain myself some more xD
A question though, what exactly do you mean with transparencies? OAO

I will study both of your modifications closely, before starting my new character ^^

r4c4 Thank you for the suggestion. The pastel colors were completely intentional, since i thought they would fit more to a sweet little innocent girl. I also thought not putting on a certain light source, would make animating it a whole lot easier. I do agree though, that there could perhaps be a bit more contrast, since she is such a little sprite. So thank you for reminding me of that ^^

[Probo]

i thought youd achieved the 86 colour count by laying semi-transparent colours over existing colours, as the PNG had some alpha properties. did you not?

[PixelPiledriver]

One place the dithering was working well was the fur boa.
The current shading lacks any sort of descriptive texture where originally even just the slight hint was nice.


I just drew a bunch of lines and it could be done more properly with reference.
https://www.google.com/search?q=fur+boa&rlz=1C1CHFX_enUS553US553&espv=210&es_sm=93&source=lnms&tbm=isch&sa=X&ei=xr7pUqqjJdbloASqmYHwDg&ved=0CAoQ_AUoAg&biw=1920&bih=947

[Jesia]

Probo I did apply a color filter, yes. But it was across the whole sprite. meaning i changed every single pixels nuance slightly. So it shouldn't have added to the color count OAO

I agree with that that Piledriver, i thought it worked well there as well, but as a suggestion, i didn't use any dithering on it. I will try recoloring it with less colors and dither the boa, as others also have mentioned ^^

Edit:

Been a while since i posted. Been busy with other sorts of artwork. But! I finally finished my new version of the lady lying down. I reduced the colors to 17 and tried adding some texture to the boa as Piledriver suggested. I'm not really happy about it, so i might try and change it. Anyone have any ideas as to how to make it more convincing? I also changed the features of her face, as they weren't exactly right, and also tried to smooth out her bicep and leg just a little bit, to make it more convincing, anatomy wise.

[Tapsu]

I have always wondered at how people shade fur according to the general form, but not according to the direction of the individual hairs, which should especially apply to reflected light.
My own sketchy first attempt (in my life) at shading fur would be as follows, in the green box:

Also, assuming light from above, the chest should be in shadow, shouldn't it?

[Jesia]

do you have any real life references that showcases this? I tried to use a reference when i did the furboa, but the problem is, that all pictures are so evenly lit, so i kinda just had to go with my guts xD

The light does indeed come from above, top left, to be more exact. You are right about the chest having to be more dark, i'll see if i can change that, without leaving it too dark and flat (atleast that is what i fear xD)

[Tapsu]

just in case, please don't get me wrong, the fur in the picture looks OK already, and unless you are interested in exploring fur more there might be no pressing reason to try what I wrote about.

The point was that people have trouble drawing furry things, and also human hair, and it seems to me it comes from not noticing how the location of highlights depends on the angle of the individual hairs, not on the angle of the general shape.

If human hair there is this crescent shaped highlight:
http://www.hairstylesguide.org/wp-content/uploads/Hairstyles-For-Your-Face-Shape-34.jpg
which in drawings usually looks like this:
http://th00.deviantart.net/fs70/PRE/f/2012/311/9/f/beyond_birthday_chibi_coloring_by_sima_sama-d5k9h5d.png
I have not taken the time to make it extremely clear to myself why hair highlights have this shape, and how it depends on the light angle, sorry.
Some more examples of "weird" highlight locations:
http://virtuavet.files.wordpress.com/2012/08/healthy-scared-dumbo-rat.jpg
http://resources.waza.org/files/images/w(415)h(252)c(1)q(90)/cccf5852eaca0fa9161ad53f87baae6e.jpg
Same should apply to shiny cloth which is sometimes hard to shade because of more complicated highlight locations.

Colored hair becomes more saturated as light is passing through a semi-transparent colored material the same way light becomes saturated green when passing through green leaves.

Looking in the same direction as the hair, some darkness deeper between the hairs should be more visible:
http://kaufmanfurs.com/gallery/albums/userpics/10001/thumb_dyed_purple_silver_fox_fur_boa_scarf_furs_5%20(296x500)~0.jpg

[Jesia]

thank you a lot Tapsu for taking the time to post these examples! I see what you mean, and i can't really explain it either haha, but it is definitely true. I'll try and see if i can make it look better, but it seems like a very hard thing to replicate in pixels, at least for someone my level, but i will surely do my best. Perhaps i will try and do some furry animal, to really force myself. Other than that, you have definitely given me food for thought, both when it comes to pixels and other types of art, thank you, yet again ^^

[HezaKey]

it seems to me it comes from not noticing how the location of highlights depends on the angle of the individual hairs, not on the angle of the general shape

This is confusing to me.  Did you mean to say the opposite?


I made up a quick image on how highlights tend to work on hair. 

First you have to take into account the shape of what the hair is resting on, in this case the spherical shape of the head. 

Then I showed some various hairstyles and how highlights might show up on them. (I did not take into account any other lighting effects such as shadows, reflective light, ect.)

The first one curves over the head and really only presents a couple places where a highlight can be seen.  The rest of the hair curves with the head and doesn't have another surface in which to reflect the light source.

The second has longer hair, and as the hair rests on the shoulder it is directly under the light source.  It starts to reveal clumps and strands as it reflects light.

The third one has shorter hair and a lot of clumps.  Highlights will reflect off each clump revealing it's shape and volume.

And then I did the mouse too.  If you see the three dimensional shape of the mouse, then the glossy highlights on it's fur make perfect sense.  They're all curving over the shape of the body, shoulder, and face. 

Short version: Highlights follow the shape of the three dimensional form.

Also that boa example doesn't seem like a good choice for lighting reference though.  It's so dark in the center because it's a three color dye job (the bottom of the hairs are dyed dark, while the ends are dyed light)  So it makes it hard to see exactly what is happening with the light versus the local color of the object itself.

As for the artwork itself, maybe your boa just needs some variation in the clumps.  Not a whole lot, but if a few of the clumps where facing different directions it might look more natural

[Tapsu]

Perhaps i will try and do some furry animal, to really force myself.

That task might be intimidating, considering http://wayofthepixel.net/index.php?topic=8925.0, though I have thought of trying it myself some day. Wondering if it would be polite to resurrect that old thread in that case.

I am still learning to draw, but challenging tasks like this make me think of diffuse highlight planes and specular highlight planes to simplify things for myself (and to exercise imagining those planes while shading). Like if some plane in the drawing is not parallel to the specular highlight plane, then I try not to draw a specular highlight there. Sometimes I even hold my palm near the picture for reference. There are more competent people here, though, maybe that method is suboptimal, but, for example, if the light source is from left and above, I doubt there would be reflection on the mid-body like your picture seems to have. The reflection curve continuing there looks nice, though.

it seems to me it comes from not noticing how the location of highlights depends on the angle of the individual hairs, not on the angle of the general shape

This is confusing to me.  Did you mean to say the opposite?
[...]
First you have to take into account the shape of what the hair is resting on, in this case the spherical shape of the head. 
[...]
The second has longer hair, and as the hair rests on the shoulder it is directly under the light source.  It starts to reveal clumps and strands as it reflects light.
[...]

In the context of your picture, what I was trying to say is: if the general shape of the hair was made from some other material that is homogeneous, say, plastic, or dough, the highlight would not have the same shape as, say, the second hair in your picture.

Short version: Highlights follow the shape of the three dimensional form.

In this case, assuming the light source is not behind the head, the highlight would be round, not crescent-shaped. I think this kind of highlights as we see on hair must come from the direction of the hairs.

[Facet]

This is a pretty interesting topic of material/reflective qualities, just hopefully clarifying what Tapsu was noting with some examples:

The collective highlights on hair as a mass are extended linearly either side of it's position on any underlying smooth form (and so forming the crescent) by the highlight position of many separate strands (essentially cylinders); the same effect is more apparent in the elongated reflections created by choppy water with the crest of each wave adding a separate component reflection. This is pretty much what underpins how diffuse any surface is only usually it's on a smaller scale (and without the directionality) but it's more appreciable (and interesting) here.

I'll have a quick doodle on the sprite in a minute, just wanted to chime in on the material stuff . 

[astraldata]

Gah. Lol, what program are you using to paint this in? The alpha channel is horrible. Makes it very difficult to provide color edits. XD

As an aside, I use Graphics Gale to do my pixel art and animation. Maybe you should check it out. It lets you right-click to select colors, count your colors, has onion-skinning, layers, custom-grids, custom interface, keyboard/mouse shortcuts, etc., and it's free as long as you don't need the animation feature (but you can get around that by exporting as a combined image PNG file), though it's only ~$20 if you do.



Anyway, this is what I've come up with. I didn't realize that was a feather boa at first. I thought it was some extra type of hair or something.

Regardless, my edit reflects a removal of 2 extra colors on the top leg and the hair (which added much needed contrast in those two areas, although you could still add another color to the hair to get that). In addition, I added in some texture to the boa that makes it have more of a feather/fur look that it was lacking with the stringy lines of pixels in the original. I also lightened the next-to-darkest color on the forearms (and everywhere else it was) and helped with the face shape a little bit (hopefully). I also played with the lighting/anatomy a little on the chest and arm, and lighting on the (ears?).

One of the biggest issues with your original and this one as well was the use of colors that didn't add anything to the image but a slight subtle shift from one color to the next. It's best to either remove these colors completely or use them to add a bit more 3d shape to the image with value gradients instead of using them to make the colors appear to blend. The blending should be done when you select the color, not when you put the color to the image (as is the opposite of digital/traditional painting of course).

Hopefully this edit helps you see what I was going for.

[Facet]

I didn't actually modify the palette properly but I did straight chuck it on something non forum-white which helps.

The lightsource wasn’t crystal clear but it seemed more frontal and focused so it’s hard to say it was wrong period, I changed it up to a stronger top just because it provides lots of nice mysterious shadow which is very useful in creating intrigue; teasing but not stating stuff which is the idea with pin-up or burlesque cabaret or whatever (assuming that’s what you’re going for with the feather boa and lingerie).

…and I just now realise I drew a bloomin’ feather boa after a conversation on fur, is that foxfur or something? I need to take an early night.

[Jesia]

It is true that i am having problems using the colors in the right way Astra, it's cause i try using the methods i use when painting regularly... which... of course do not work at all! haha. It's a habit i just have to get rid of. To answer your question i use photoshop. I'm gonna look into that program you talked about, it sounds very interesting, and might make it a bit easier keeping track of things. Also. i like what you did to her chest, lot more voluminous teehee.

The hair discussion has been interesting to read. Thanks Milo and Tapsu, for trying to explain it. It has made me think about that topic some more haha, i hope it will help me render better fur and hair in pixels ^^

The lady is based on an old sci-fi sketch i did a long time ago. I wasn't exactly going for cabaret, but someone said it should be more "eye candy" so i kinda just played of on that. Otherwise i really liked what you did facet the volume and depth is great!. I have tried approaching things in a more painterish way, like you did. But whenever i clean it up.. it just... i don't know. loses a lot of initial feeling. Any hints on that? Also, what is the best way to do pixel art? Paint and then clean, or do linework?

Sadly. i haven't had time to do edits myself, and i won't be able to for the next few weeks, as i am busy making a portfolio for an animation school. As soon as i get that out of the world, then i will be back studying and learning pixel art with the help of you guys. Thanks a lot, for spending your time helping me improve. I hope you feel i am taking to heart all that you have to say ^^

[Facet]

I hope you feel i am taking to heart all that you have to say ^^

Absolutely! You've addressed everything brought up even if it wasn't what you might have set out to discuss .

Losing feeling & the magic of suggestion when you’re cleaning up is a perennial problem but my working preference is to clean up only when absolutely necessary . I just typed out a bunch on linework and method versus insight in a neighbouring thread, so that may be sorta the same stuff you're thinking about.

[Jesia]

So i finally finished with my whole portfolio and school thing (took a long time haha), and so i am back at practicing Pixel art again.  After looking through the posts it seemed one of my main issues was using too many colors, so i have tried limiting myself to a smaller pallet (12 colors in this one)
It is a work in progress, not made for anything particular other than serve as a practice. Im still new to the whole "banding thing" so sorry if i did something gastly xD
I'm sorta stuck as to how to proceed with this. I want to expand it, adding more science objects (wires, machines etc.) But i am unsure if it would get muddied or if i should introduce new colors ^^

[astraldata]

Hi there again! Long time no see! Good to see you got back to pixel art!

I'm going to be a little harsh here, so bear with me -- I just feel honest critique goes a long way when trying to improve on something.

So, first thing, is that you've done pretty well with 12 colors to get across what you were trying to convey. Unfortunately, I feel like your palette could have been used a little more effectively in the sense that you seemed to put most of it toward shadow gradients, etc. These entries actually caused you to have a 13th color somehow in the darkest shadows on the floor. I suggest you get a program like Graphics Gale if you haven't already (it has a feature to count colors in the image). I actually removed a color in my edit that could have been rendered with an existing color (i.e. the wires' gray color), so my color count is now 11 after removing that extra color in the shadows.

You went pretty heavy on the dithering and it mucked up the clarity of your image, so I tried to show you how it could look better with less dithering and more solid colors. There was some dithering I didn't even see zoomed-in, which means it really was unnecessary. I do appreciate the subtle color transitions you were aiming for with the dithering, but with color-restricted pixel art, this is not usually something you want to devote specific colors to, but instead you want to make sure all the colors you have evolve naturally into one another, creating a dynamic, but organic, palette -- and you've accomplished this pretty well, so congrats.

Not sure if you were aiming for a great composition for this, or were just practicing forms, but wires were definitely useful in that. I got the feeling of a Resident-Evil sort of house-laboratory feel from the colors and dithering, so I edited accordingly:



The biggest trouble I had with your image was the subtle transitions that were very close to the same color except normally in hue. Many of the colors in the dino's housing were replaced easily by other colors in your palette, which could have freed you up to use them for other stuff. I had a whole extra color leftover out of your 12 color restriction by getting rid of the gray on the wires, as mentioned before, which allowed me to use the background gray as a middle shade for the wires, since the intention was to blend them in fairly well. I stole a little focus from the main subject matter using the lighter wires, but I added more interest to the composition as a whole in doing so. It also gave me an excuse to use them to frame the empty space in the rest of the composition.

Not sure why there's a shadow being cast into the darkest color in the image, but I removed it to show a little more of the 'dark' mood you appeared to be going for with your color scheme. I left some dithering in there on the oxygen tank in case you could come up with a more purposeful use for it, since it kept the framing looking nice, but it really should be modified a bit or removed.

As you can see though, there was no need to add in more colors to use the existing elements in the scene to make it work compositionally. You can do a lot with the rendering to solve issues such as 'needing' a mid-tone on the wires. The end result was only 11 colors (I started with 15 before I removed the white, then ended up with 13, cleared a grey color and found that hidden dark grey dither in the shadows, leaving us with only 11.).

Though, as a side note, most color-restricted palettes will generally restrict to about 15-16 colors (SNES style) -- if you're going for any sort of variety in painterly detail, as you appeared to be going for, so take it as a point of pride you got this composition looking fairly workable in the 12-13 color range you did. It's possible to restrict more, and still get a pretty workable result, as NES games have shown time and time again, but for a nice image to look at, this wasn't far off -- I just gave you a few techniques you might consider reinterpreting on your own next time.