Hello, welcome to pixelation!
I think nobody has answered you yet because your pieces are already quite good with no huge apparent flaws. So congrats on that, your art is already readable, nice to look at  with good use of colors!

If I may, the small character with the long hair feels a bit confused, try maybe to simplify the hair texture to get a more "united" feel, less pure texture, more volumes.

On your last green landscape, the clouds that constitute the background lack a sense of direction. You could draw the curve which they're supposed to follow and then on another layer shape the actual clouds and give them a cleaner finish.

On your Pokémon-like creature, the shadows feel muddy. Picking a color that's a bit darker and more saturated will improve your piece overall. You could also play with hue shifting to make it more interesting (this PJ thread is very enlightening http://pixeljoint.com/forum/forum_posts.asp?TID=10695
Also a good way to improve would be to participate to some challenges; pixeljoint.com holds a different pixel art challenge each week

I gotta disagree with indigo there, I don't like the infinite scroll at all. I always feel kind of cheated on websites and apps that use this method, because I can't properly control the time I spend on them, it's designed to make you stay longer than you initially wished.

Keith's website could use some CSS, but it's pretty neat already, thanks a lot for doing this
Would it be possible to display the image when there is one? Or does it just not display on my phone's browser?

Not too bad, I recognized the style of the first Pokémon games before reading your post



Made Pikachu a bit shorter, repositioned and reshaped the ears and nose, redid part of the tail trying to fix the AA, and lastly made the overall shading a bit more dramatic (careful though, the sprite might have to be flipped horizontally so he's facing and shaded like the other gen1 sprites).
Does it help a bit ?

First of all welcome to Pixelation now sit back and relax, it's going to take a little time.
First I see you've already got some important things nailed: working with a low color count to avoid ugly blurry gradient effects, the lineart approach and the cluster approach, shading relatively to an imaginary lightsource...

which is a bit strange, because I've been drawing faces since I was a young girl. I feel very lost. I'm thinking it's probably the fact that I looked at pixel art as an entirely different art than classical art.

Let's take a look at your portrait for a second. The proportions are alright, the shading follows a up-right corner lightsource, there are few colors with good contrast, so why is it "terrible" (it's not so terrible though, it's got potential)?
Here's an edit i made (it's not perfect but I'm no pro, it'll work  ):



As you see, I didn't change an enormous amount of stuff, yet it feels "more right".
I'll try explaining each change I made and why I did so:

• Your color palette was very linear: light orange, getting progressively darker into brown. To make your colors more appealing, try using hue-shifting: say your middle-color here is brown. Try progressing towards another color value as you go darker and as you go lighter (ex.: ). Here the darker shades go toward purple, whereas the lighter ones go toward a yellowish pink (and I lightened the skin tones bc she looked a bit Donald Trump-ish with all that bright orange ). It's not simple in the beginning (took me a year to kind of understand that :') ), but experiment until you're comfortable with hue-shifting, it's well worth it. Also, keep this in mind for now: desaturate as you go lighter and saturate as you go darker.
• You're very dependent on lines: coming from traditional media, you think "Oh, I have to represent her nose/collar bone/eye/eyebrow/[thing]! Easy, I'll just draw it". While when drawing with a pen on paper it works wonderfully, in pixel art, not so much. Most of the time, in pixel art, don't represent a volume with lines, but instead by using light. A nose isn't a line, it's just more of a face; it stands out because it's in 3D, and light hits it differently from the rest of the face. so it's got a highlight, it casts a shadow. Lines don't make volume, light does. You have to learn to "think in 3D", thinking "how does my lightsource affect this element I'm trying to depict?". I don't know if it's clear enough, don't hesitate to tell me if it isn't.
• Some of your outlines are "jagged", they don't form a harmonious curve and the result is unappealing: See this video for a more in depth explanation!
• I added some AA (anti-aliasing) in various spots to soften the image. You did a good job with the AA tutorial, so you should be able to adapt the theory to a practical case pretty easily

About your last trial with the cluster approach, I'm really not an expert of the cluster thing so I'm not going to comment on that; I just want to say something about the palette you put on the left of your image. You put four linear ramps w/ 6 shades each, but only used few of those colors. My suggestion (I could be wrong tho) would be not to "build" your palette beforehand, but to build it "organically", adding a color when you need it. That way, you'll avoid unnecessary colors (by asking yourself "do I really need to add this color or could I use a very similar one that's already there?"), and you can make multiple "links" between the ramps. It's difficult to explain, but your palette shouldn't look like several disconnected independent ramps, it should look like interconnected and interdependent ramps: it will add visual coherence and appeal to your image, tying it together. For more information on this very important point (I insist, it's fundamental in understanding color, it took me way too long to get better because I was reluctant to learning this), go and read this whole PixelJoint topic, which helped me immensely. It's a lot to take in at once, but I promise it'll get easier as you progress!

That's about it, I hope I was of some help!