Question to ask yourself: Where do you see yourself in your perfect world in five years? What company are you with? It's morning and you're emailed a prompt with 5 things to do. They're all awesome, and they're so damn good you don't know which to start on first. What are they? That's the good stuff you want in your portfolio. Cheesy and vague? Potentially, but it works.
Not sure if you see it, but you have a HUGE discrepancy between your video game art and work that seems to be genuinely yours. Bioshock, Gears, Star Wars, all awesome minus some creatively concealed background environment work (except for the wood floor, me gusta). But then you have your own work, and suddenly it takes a dive, and while I'm certain you were referencing images of Marcus Fenix and the Sith, I'm also certain that images of "moving fire" and "reptile anatomy" and "grassy valleys" were not referenced as much. If they were, my total apologies, but from what I see, I don't see a lot of solid referencing and studying in your personal images as was done in the Bioshock and Star Wars and Gears images. All about the ref's
In terms of setting up yourself on the internet with a portfolio, check out carbonmade.com. Came across it, and I currently use it (davidszilagyi.carbonmade.com). If you want, you can also start a blog on blogspot and throw several images onto a page like Feng Zhu does (www.artbyfeng.com).
Anatomy's looking solid, as it's only going to get better in the next months. And for more info, take the initiative to your heroes. Go check out the credits to your favorite games, find the concept artists, and hunt down their emails (as they're workin people and probably are always looking for the newest and best work out there). Some might not get back to you, but then you get gems like getting a call back from your hero Ben Wanat from EA and getting his email
It's the advice that even I hate to give myself. If you suck at something, attack it, but attack it in a way where you can master a very small managable piece at a time. Case and point, I suck at painting eyes. I went on google's art project site and found a master painting of some US president. Zoomed in and screen capped that pic, threw it into photoshop, made a layer on top of it and painted out the top lid, and painted it back in using only logic and guess and check from surrounding colors. Painted out the actual eye, did the same thing, guess and check until it worked. Lower lid. Then eyelashes. Then touch ups.
In the words of Bill Evans, if you try to approximate the whole thing, you build on vagueness. Attack small pieces until you can demolish them no problem. And if that means honestly going ALL the way back and taking care of perspective issues, then all the better. Do that for an hour straight, perspective after perspective. If the next step is "well, what the hell actually goes INTO an environment?" Try this: Take any one of your images, imagine what's really going on in the environment. Write it down, make a list of words or hollywood moments that'll happen. Write down as many TANGIBLE things in the environment (Ask "In this environment, what could drop on my foot/pin me down/ break my fall/kill me?"). Then google those images.
If you're still having a rough time, take a general area and type into Youtube "video tour [general area]." Odds are through searching a little, you'll find videos of those little human elements that just make sense in an environment (toll booths, caution wet floor signs, tool racks, O2 tanks, warning signs, coat hangers, lockers, trash cans). The more "lived in" environments, the more people are going to trust you with the huge set pieces. That being said, don't go crazy, there's still an action/character to take the main stage
A HUGE shortcut I use. Google Sketchup, build some shapes, play with the angles and FOV (google it, Uncharted 2 concept art was built on a lot of sketchup). Export a jpeg into photoshop of your shot, throw some cgtexture.com goodness onto your buildings and ground. Play with your lighting and shadows (I use overlay brushes and layers heavily here). Add in happy details, play with it until your image works. Bam, hollywood level concept art
A good trick to ask yourself (I do this all the time) is look at some pro portfolios out there, and ask yourself "If they had ONLY an hour to do this from start to finish, what would they do to make this happen? What actually matters?"