It sounds like you are mostly putting the cart before the horse.
You don't need to make a big production of starting to work on game projects you plan to sell. You don't even need to form a legal company, although if you're going to involve other people in the production process at all, you probably should, and there are potentially some tax benefits for doing so (discuss this with a certified accountant or tax planner).
In the US a good option for this is to form an LLC. This involves only a few forms and a small fee (consult the website or office of your local government for details), and it lets you have a legal entity shielding you from total liability and provides a useful entity to assign intellectual property to in the event you bring a partner/collaborator/employee on.
As for employees: this is where you start to get ahead of yourself. Stay far, far away from hiring people as you can. You can bring on collaborators or partners (be sure to consult a lawyer and draw up contracts stipulating the ownership of their own, terms of their work, compensation, et cetera), or you can purchase work from third-party artists/developers/whatever. But as soon as you start talking about "hiring" people, people's expectations of that arrangement change, to the tune of expecting a salary and benefits. Salary alone can run you anywhere from 60k to 90k per year, approximately, depending on who you are hiring for what and where they are physically and in their career. Benefits are another several thousands of dollars on top of that. At this point you've gone from maybe a few hundred dollars in LLC filing fees to needing have a cash flow that will support nearly 100k a year. That's a huge jump, and you don't even have a product yet.
And the project? "A zombie-survival FPS" is totally doable, but "astonishing graphics" tend to cost money, both in terms of an artist who can produce them and a programmer who can build or leverage the technology to render them well in the game. That's starting to feel like where you'll need to scope the project back because you may not have the capital; if you need both a programmer and an artist because you can't do one or both yourself you're now talking about two employees, so maybe 200k a year. Conservatively.
To more directly answer your questions:
How much does it cost, first to found, then to run a game studio (yearly)?
If you can't at least come up with ballpark numbers for this yourself, you really have no business running a studio. Every company is going to operate differently, with different costs and overhead, and managing and planning for that is going to be a critical part of your role as the founder of the studio. Unless you can afford to simply hire somebody to manage it. Sit down and think about what you (think you) want, specifically, and then go out and start doing research on those specific things. How much do programmers cost on average in your area? Are you going to want an office? What's office space rent like in that area? Electrical and internet bills? Et cetera.
What are the roles in a game studio and how much everyone should be paid?
You don't hire the positions "a studio should have." You hire the positions you need. Ideally, in the beginning, that's as few as possible (zero). But if you have a critical gap in your skills, you're going to need to fill that. Hiring somebody to do it is one way, but establishing contract work arrangements is probably better. Frankly I'd worry less about who to hire and for what positions and more about finding a good accountant and a good lawyer and how to pay them. Rather than try to staff up to make your dream project, find a project you can make yourself without hiring anybody, maybe buying a few assets or whatever.
Do I need to have already founded the Studio to start a Kickstarter campaign?
Individuals can run KickStarter campaigns, as far as I know (consult the Terms of Service for details). So no, you don't need to have formed the company, although it's probably a good idea for the liability shield. What's more important is having an actual product and a plan that will convince people to hand over money and be something you can actually achieve. I would try to build as much of your project as you can without even considering Kickstarter. Not only will you then have something that more closely resembles the final product (which is more interesting to donators), but you also reduce the risk the project will fail (since it's further along) and the scope of the things you're asking for Kickstarter money for are likely to be smaller. For example, maybe you can build your game with cheap/free assets off Turbosquid and the like to get the gameplay, mechanics and technical details down. Then you can find a good artist, work out a contract, and engage a Kickstarter campaign to cover the actual cost of that contract to get your final art in the game.
I really think at this point the only things you should consider doing are:
- form an LLC or whatever the equivalent is in your area if you're not in the US
- work on your project yourself, without involving the hiring of employees or the pomp and flash of a "studio" until you have something resembling a fun, salable product
- then go from there