After my last post, I remembered distantly I'd written another article on this topic a while back, but time and happenstance shoved it to the bottom of the To-Post pile. One archaeological expedition later, here it is: my Totally Non-Canonical Guide To Making Good Live-Action Films From Manga And Anime.
Hollywood-only rule: Pick a property that you can "Westernize" without destroying. I've hit on this point so many times by now I run the risk of breaking my wrists. The more a particular property is tied to its cultural milieu for its net effect, the more difficult it is to relocate it without breaking everything.
Hollywood tends to think of everything as raw material to be reworked in any number of ways, and in fact the whole process of pitching ideas revolves around such plug-and-play conceptualizing: "It's South Pacific in space!" But not every such thing survives being uprooted and transplanted, so you're best off picking a property hardy enough to thrive in any soil. Blood: The Last Vampire was a good pick for a live-action production, because it had one foot in both the culture it was coming from and the culture it was being marketed to.
Keep the budget sensible. The less you spend, the less you have to make back. The global culture of filmmaking (South Africa, Eastern Europe) makes it far easier to create epic-looking movies on $30-50M budgets (District 9, Doomsday). That and having a smaller budget forces you to think that much more about your actors and your story, instead of your effects sequences. This was, ironically enough, something originally made most clear to me by the folks who graduated from the Nikkatsu exploitation-film factory.
Don't depend on star power unless your stars work cheap. This rule applies to most any movie made today, really. People are growing increasingly disenchanted with stars, and that much more hungry for stories — right at the same time moviemaking has become that much less interested in stories and that much more driven by marketing and distribution. Build the movie around characters we given a darn about — much as the best manga themselves are! — and you'll get the audience you deserve. Even Chow Yun-Fat couldn't save Dragonball: Evolution. I'm not sure he would have cared to, either.
You can't make everyone happy. You're never going to make all the fans happy. But you can keep from ticking them off more than you need to. Keep the things that make the story what it is: the attitudes, the mannerisms, the way things look and feel — and most of all, the characters themselves.
Do justice to the story that matters. If you're adapting material that runs to many volumes, don't go nuts trying to include everything. You can't make everyone happy, and you shouldn't try. Dig out the dynamics between the most crucial characters and stay true to that. If you have a self-contained arc in the first few volumes, use that; it's hard to go wrong there.
Your fans can only spread the word so far. Using an existing fanbase as your way to spread the word about a property is, I fear, a double-edged sword. Their enthusiasm is boundless, but they are not always capable of reaching the Masses of the Great Unaware through the sheer force of their enthusiasm. James Cameron was able to get most of Western civilization to see what amounted to Pocahontas in 3D with blue space elves, and did it without relying on fandom. He relied, instead, on many more time-tested Hollywood hype machines, which were correspondingly more expensive ... but also that much more far-reaching.
It doesn't even work between fans a lot of the time. As geekish as I am, my interest in Firefly/Serenity (to pick a random example) was never more than lukewarm despite many fellow fans' propagandizing it to me, and when I finally did see it for myself I was pretty unimpressed. Now is not the time to go into my whole lecture about FF/S, but I feel a lot of the ferocity behind its fandom revolves around it being an underdog that never had a chance to fully blossom, and not because the show was all that great to begin with. Send your hate mail to the usual place.
Note that I don't expect anyone to actually follow this advice, but some open discussion about how these things do and don't work should be useful in the long run. Plus, it's fun.