Cyclops shading

[shark jacobs]

I'm pretty new to pixel art, this is my first attempt at creating something with shading, i'm not super happy with how it turned out but am not sure what to do aside from fiddling pixel by pixel and zooming back out to see if it looks better



here's the preliminary work i did for it
http://imgur.com/1zvT3aR

[astraldata]

It's not horrible. The construction (silohuette, porportions, etc.) could be improved greatly, but since you asked for shading advice, the one thing I would suggest is to keep the huge jump in contrast down if you're going for a more soft/cartoony feel. By this, I mean to add in a buffer color to the dark shadows to prevent them from being so sharp in so many places. It's a huge mistake to go with so few colors unless you're going for an NES look purposefully. Otherwise, 3 colors is simply not enough, even for a monochromatic image of such a fairly low resolution like this. You can think of this "buffer" color to smooth the edges of where the highlight and the shadow blend in critical locations on your image, so that you can suggest roundness in things like the belly or crease in the thigh against the belly a bit more smoothly and make it appear more rotund.

The one other critique I want to give is to only add a highlight colors on places of 'shadow' where you want to emphasize something's presence in the overall piece. In the case of the (our) right arm, the highlight should be more prominent across the forearm (and the other arm's should not express its brightest highlight) to indicate that it is closer to/further from the viewer's presence if this is a character that will be animated. You can add more blending shades (buffer colors) if you need to represent more facets of the form more realistically however.

As a side note -- if this were a scene-specific sort of object or a painting with a very definite light source from the left (as your current image suggests), what you have could somewhat suffice, but the depth problem still remains with that deepest arm being given too liberal of a highlight (or any highlight at all really) when compared to the arm which is supposed to be closest to us (judging by the perspective you seem to have). This all can still be fixed in your image even with 4 color values on an object with a prominent light-source by following my previous advice.

[shark jacobs]

I tried to follow some of your advice
before

after


you were right, softening the shading helped a lot!
I added highlights to the close arm and reduced them on the far one. i didn't remove them altogether because i'm wary of adding many more colours, (part of the reason i'm getting into pixel art is because i like the idea of super minimal palettes) but maybe i'll see if i can do it with my existing colours

[astraldata]

Definitely looks a ton better!

One caveat I want to mention about getting into pixel art solely for the appeal of minimal palettes is to not realize what is truly "minimal" in regards to the style of art you're going for.

To put it more simply, a "colorful" NES sprite tends to consist of around 4 indexed colors (one of them is transparency -- which your original image adheres to somewhat well, barring technical NES tile-chunk voodoo) while an SNES sprite tends to support a lot more colors (usually up to 15+transparency for even the most detailed sprites, though this number depends on more technical voodoo I wont go into).

With that being said, I usually aim for sprites with palettes no greater than ~15 color entries (which sounds like a lot, but when you start seriously trying to design a character the size of your cyclops (or larger) around those limitations, you tend to have to make compromises that will sometimes actually improve the design, and your artwork simultaneously, by forcing you reuse colors in a more creative way, thus forcing you to develop a more varied palette and improve your sense of color selection on top of that). I personally would use no less than 6-8 colors for a 2 'color' sprite, or 10-12 for a 3 'color' sprite, which still gives me a few more palette entries for things like eyes and details like necklaces, trim/lace, or whatever. You'll quickly see that these 15 entries disappear rather fast. ;P

Although this is almost beyond the scope of the topic, since we're discussing minimal/restricted palettes, a game level's palettes should stick to the idea of using multiples of 8 or 16 (i.e. 16x1=16 colors, 16x2=32 colors, 16x3=48 colors, etc.) and perhaps even use a universal palette for all levels. A minimal palette for a single tileset layer would be no less than 8 for NES quality (since multiple bg 'layers' didn't really exist on the hardware as far as I know), and 32 for SNES quality since you would likely want to make a level with at least 3-4 differently colored objects and walkable surfaces, though separate BG layer palettes can usually vary in number of entries from those of the tileset layer palette depending on the amount of detail you're going for.

Hopefully this helps you or anyone else wondering about what is "minimal" for minimal palette selection for game graphics.

[Manupix]

Pretty good for a first try! There's a very convincing sense of light in this already.

not sure what to do aside from fiddling pixel by pixel and zooming back out to see if it looks better

That's how it's done =)
If your software allows it, opening two windows of the same document helps, one at 2x or 1x to check and one large to work.

As far as shading goes, I mostly prefer your first version. For one thing the contrast is more lively, and the transitions are consistent with the small pixel size. The new version looks flatter.

Astraldata's advice about using light to emphasize nearness is very interesting (I've never thought of that before), but in this case it conflicts with lighting realism, and I'm partial to that.
Brightness of a body part depends on the angle the light hits it, as well as distance from the light source (inverse square law). We don't know the source distance (anything from close to infinity - sun), it might give better presence to assume it's reasonably close.
The near arm is angled away from the light, more than the far one, and it could also be farther, so it should be darker. Specifically, the upper arm is more angled away, and was rightly darker than the forearm in the 1st version, even despite the greater distance. This made the arm volume easier to read.
On the other hand, the inside of the far arm and leg receives reflected light from the torso and other leg, so the lower contrast / brighter shadow there makes sense.
A brighter highlight tone might be useful for those parts facing the light and closer to it, such as the top of the head, his right shoulder, maybe upper torso.
A few pixels of shadow to encase the eye would greatly help, too.

[Fizzick]

Definitely looks a ton better!

One caveat I want to mention about getting into pixel art solely for the appeal of minimal palettes is to not realize what is truly "minimal" in regards to the style of art you're going for.

-everything else-

I agree with you in part, but truly it varies greatly from work to work. It is not always necessary to use more than three shades/hues for each color in a sprite, though it may be called for. If the cyclops truly is only orange, four shades should do fine.

In addition, limited palettes are not solely for the purpose of mimicking retro game platforms. It can also be used as an exercise in minute color control, and overall piece unification through 'webbed' palettes. In this case the emulation of technical limitations serves as an arbitrary number rather than an artistic facet of the work.

But I don't claim to be an expert on the subject, and everyone limits for different purposes.

[astraldata]

I tried to show the critique I was referring to in an attempt to explain it better. That fourth color I called a "buffer" color because it was meant to 'soften' rather than add another facet to the form as you can see around the chest, arm, and stomach areas. I kept the dark color sharp as Manupix had suggested because I agree that the dramatic feel it created via contrast was pretty useful. The only other thing I did with the shading itself was make the light source more clear (in front, above, and center to the character's front) and showed what I meant about removing the 'highlight' color on the distant arm by making the highlight the next color down in the ramp, using value to indicate distance/depth in this case.

Shading aside, I edited your character's outline so that it might work better on backgrounds of even the exact colors of your sprite, though I didn't go into that too much on your sprite (I only hinted at it in some places) so that you might have a chance to finish it up yourself. I didn't add any additonal colors, even though it may look as if I did. I shaded this as if it were going to be a game sprite so the conventions I used were the same as in my first post so that you might be able to see the difference between your updated version and mine critique-wise.

@Fizzick
I agree with your points, and I mentioned some of them indirectly in my original post about it if I'm not mistaken, but I was trying to be as general as possible about how "minimal" is not always < 4 color values unless you're a) going for a specific style, or b) only need 3 shades to represent something (such as on really small sprites), of which I assumed the latter was understood. But either way, thanks for reiterating that for anyone who didn't catch it elsewhere. I see a lot of people going *too* minimal when it's clear they want a different result than what they're getting. Thus the reason I mentioned that fourth shade.

@Manupix
I wanted to mention that I agree with your assessment about how my suggestion to use value to indicate nearness does tend to take away from realism (for example, in nature, in the daytime especially, things are very high-contrast), but for the sake of clarity and to prevent the 'flatness' you mentioned with such a low number of colors on such a large-ish sprite, exaggeration of depth is definitely a good thing in most cases with characters of this size since it adds its own sense of dynamism to the image. That may just be a matter of style preference though. I just know that, as far as in my edit, and most game sprites (especially old-school SNES) not attempting to be photorealistic (as I assumed was the case in this sprite), this is the way it's usually done. As I also suggested in a prior post, for tilesets and backgrounds, I'd do it a bit differently (probably more the way you're suggesting most likely -- more realistic lighting, less attention to how the sprite looks mirrored or on various flat-colored backgrounds of a similar color to the sprite itself, etc.)

Granted, there are games that get away without doing that level of shading on their sprites, but usually there is a very different background value scale compared to the higher value intensity of the sprites (i.e. the BGs are usually very dark or not as colorful/vibrant as the game sprites themselves, etc.) The method I suggest is just most useful to sprites existing on bright/colorful backgrounds. The other types of games are usually deemed more 'realistic' looking though. Just food for thought.