I thought IvanX's second face edit was pretty spot on myself. It looks like something from Starzinger or the early 80's anime era. Long nose ridge, narrow eyes.



Rough ideas: Use panel lines to sculpt suit. Trunky 80's boots? Propped a leg up for no particular reason. Her original hair color is brunette. Green is with Varia.

I felt the gray wasn't working on that one. By enhancing saturation points, I mean... when going from dark to light, (and probably pretty desaturated), or just from one form to another, having a 50% blend is a bit boring. By adding some saturation you can say, hey, this thing has some color dimension too. I don't really mean saturated terminators (on a an overexposed red planet you'd only see black where shadowed, white where lit, and say, martian red on the terminator because it's the only place where you can fit in a normally saturated color. I'm more talking about sort of whimsical points of light, chromatic aberration perhaps. Actual color, perhaps. Subsurface scattering, perhaps. Self illumination, perhaps. It doesn't matter so much. It's just to bring in more color dimensionality.



I never uploaded this one for some reason.


Basically I'm talking about a more subtle version of painting with correct values but garbled colors.

A related discussion is what I'd call... color textures... color fields which consists of many colors upon closer inspection, i.e. dithering in pixel art, but with the intent of creating something more than a color rather than just a new value. The impressionists and pointillists did it with gusto. http://en.wikipedia.org/wiki/File:Henri_de_Toulouse-Lautrec_056.jpg It's pretty common to see artists from any time mix in blue textures into warmer colors but digital art tends to be cleaner nowadays. Paul Bonner works traditionally and likes to use a copper teal: http://www.amazon.co.uk/Out-Forests-The-Paul-Bonner/dp/1845767055 - It's sometimes a plain flat hue shift but I think the double (or triple) color texture has a different feel to it so I wouldn't call that a hue shift. It's visible on the brushstroke level.

Needless to say, this kinda stuff is difficult to do in pixel art with a small palette and so few pixels to work with. Can't quite do the dither-color thing on a tiny NES sprite (pixels are too big) and you wouldn't want to waste a color on just a saturated dot. Can play with adjacencies though, maybe colored outlines.

I'm thinking they did the daytime graphics first, then implemented nighttime later, at which point the artist had to paint in the index for window light anyways. Seems unlikely they painted the nighttime graphics first, or had the foresight to paint in a special index for daytime windows from the start, so it was still extra work, so to speak.

Edit: The best money saving approach for them would probably be to do a complete sheet with "programmer art" and then hire a guy to paint over that.

Perhaps the proper question to answer is "how to go about estimating how much time is needed"...

That Ogre Battle tileset does look like quite a bit of work. Different climate zones * night and sunset versions Sure, a lot of it is copied and color tweaked, but I don't think I could churn out a full set like this in say, 12 hours, at least not if I had to figure out the structure/system from scratch, test tiling, run into dead ends, etc.

I was talking about something abstractly like that map in terms of features (rather than quality). I'm not familiar with the game, but I'm guessing it has several maps, each with different features. There might even be some climate zones.

If you have 100 tiles and spend 5-10 minutes on each, working nonstop, that's 12.5 hours, which you multiply by pi because all time estimates should be multiplied by pi.

Basically, there will be a lot of extra time on top because the something was done wrong (miscommunication) and needs to be redone, and maybe the mountains looked peculiar so you spend an hour or two fiddling with those, and some beach seams needs fixing, then there's the little towns and buildings which can be unexpected extra assets. And woops, client wanted combinatorial roads too. And then the tiles needed to be fiddled together on a sprite sheet a certain way. And then there's the first hour of two before starting, just figuring out how to do stuff efficiently, and understanding the client's needs and... tolerances. Maybe he wants a few initial style tests?

But like I said, if the artist knew exactly what to do and got free reins (client will be left with whatever comes out), maybe a full days work (probably split into 3 because communication lag). It's a lot different to just draw something and to draw something for someone else. I'm also assuming the artist isn't sloppy and puts some love into cliff edges and stuff.

If we're talking Ludumdare speed & quality, maybe 4hrs. Perhaps the same for just replicating the particular map that I linked in about the same quality. But that's probably not what the client asks for unless it's some kind of fan game.

Edit: I see now that OP mentioned SNES "quality" so it seems we're talking lower end here. A smaller studio could keep the costs down by figuring out what is needed, limiting the amount of blind alleys run into... but since the post was so vague...