In my opinion with art tests it's not really about the quality - it's about the impression. As long as you make people go "wow" and don't allow any major incompetence to show through - the job is yours.
The audience never knows what they want to see. They usually expect something similar to what they've already seen, but that impedes your ability to impress or even bring something new to the table. Which is exactly what we need from people in this industry and what the audience really expects.
So my advice would be to go through all the versions of the test you can find on-line, make a bullet-point list of qualities that you think made those tests impressive and another one for all the mistakes you must not repeat.
Then go through all the games the studio has made, the whitepapers they've published, their history on wikipedia or elsewhere, the art dumps they post online and the artists they employ and the test specs they've given you. Try to figure out what makes them tick, what makes them them.
Whatever info you find, try to use it in some capacity to show the studio that you're more intimate with their processes then they expect, that you're trying to conform to their ways and would be a great fit for them right off the bat.
After a thorough analysis add a few original points to the first list. Something of your own that you think will take this test to the next level. And then make sure that all your actions reinforce the vision you set for yourself and avoid the mistakes you've outlined.
It's important to remember that since it's all about impression you can go out of the boundaries(not specs) of what the test initially requires. In fact, I always would. Make an impressive piece of artwork and not an impressive art-test. Allow yourself no excuses for whatever restrictions you have. If the test doesn't look good to a person absolutely ignorant to your constraints, then chances are you'll have a hard time impressing the people you need to impress.
I'm not saying that people don't get jobs by just following the guidelines, but when you are going to take no small amount of time to fight for a place at a particular studio you must really want that job.
So go all out or go home.
If you're going to half-ass it you might not even start at all, 'cause there's always going to be someone more motivated who will crush you with sheer enthusiasm. Show the company that you are that guy.
One last thing that is crucial for you to know, is that your employment doesn't really depend on your test. There is a million factors at play here. From other applicants to internal studio stuff to local labor laws. Unfortunately more then a few people with good tests, resume and attitude end up turned down in the end of the day. Keep that in mind and don't demotivate. Even if it doesn't fully depend on you whether you'll get the job in each specific case, in the long run it is all about you. So above everything else - persist. And you will do great.
- Research the team
- Research the test
- Outline the traits worth repeating/improving, mistakes worth avoiding
- Figure out your unique input/twist on the test
- Impress. Overdeliver if you have to
- Go do another one