@Rosse:
While induction works well within a game, or something to that effect, it would not work at all with a single sprite or tile; when I looked at that ampersand for the first time(it being bold, it caught my eyes before I read the sentence) I saw "
and," not a big, scary demon; but If I was, say, playing Nethack, then yes, it could pass as a demon, only then though. This is actually the opposite of what I was trying to create; I wanted something instantly recognizable to avoid confusion of what, exactly, this supposed to be, so that someone who knew what a Starcraft sprite was, he could say, "yeah, that looks right," or someone who didn't know what a Starcraft Marine was, could google it, and say, "oh yes, that does look like what he made."
I think I found a bridge over the gap in our views; we're both striving for information, Helm defines this very well.
detail is pixel placements. If you can use less colours and still maintain the pixel placements that you've done, then this is good. This is relatively easy. If you take a 256 colour pic and you turn it into 128 shades of gray, it is very possible even automatically, a machine will induce no loss of detail. Information isn't pixel placement, though, it is WHAT the pixels are. What colour, what they signify as texture and symbol.
What I was trying to say in my post was that, in order to have this essential information, he needed to place the pixels with more of an idea of what he was trying to create. In other words, he had a bunch of volumetric balls and other shapes, with a small bit of detail, but not enough to inform the viewer of what, exactly, he was attempting to create. I'm not saying that you should bog the sprite down with tons of details like I implied with my first post to this topic, and I apologize for that; I really do value simplicity and lighting, but if you focus too much on these it can, and will, make your sprite look just as bad as if you had over-detailed it. This also reminds of a post I made a while back:
But you must be careful when applying shades. Just remember that the fewer the better, as far as colors go; the reason for this is that when you have, say, fifteen colors in a little 32x32 square, it ends up being so smooth and flowing that none of the few details (that you can actually fit into that amount of space) really show up.
On the other hand, if you try to cram too much detail into your 'lil sprite, it takes on a noisy quality; which is, most definitely, not what you would want. Because most, if not all, of the volume would pack its bags and go off in search of something that could make good use of its illusion.
What I was saying here was that you need to find a good balance between volume and detail to really have an informative sprite, at any size.
@Xestro:
Your sprite is looking much, much better than what you had when you started out. Just a few crits to push it even farther:
First, it'd be a good idea to get rid of the highlight his (not our) left shoulder plate.
Second, I don't think that the dither is necessary for a sprite at this size, and dithering never really works with animations anyway.
Third, he's a little off balance, just move his torso and arms a pixel to the right and it should look better; a good way to check balance is to mirror the sprite, then work on it for a bit from that perspective.
Fourth, contrast, you need more of this; the colors are just a bit too close to each other.
Fifth(this is more of a nitpick), the visor looks as though it is bending towards the sprite's head a little bit, you could fix this by not using any buffer shades from the highlight to the shadow to create a thinner looking top-piece.
Other than those few things, though, it's looking really solid.