Occasionally you get a hands off client that lets you do your thing. Unfortunately, a client who trusts you to make judgments and decisions without too much oversight is a double edged sword. You run the risk of making a bad decision and allowing it to become written in stone before the client sees it (who then freaks out). At the same time, being enabled to make decisions allows you to move towards completion on an accelerated schedule, which is great if it's a fixed price engagement.
More often you find clients that micromanage the project. They will make requests like 'bold that' and 'make this green' and 'put that at the top' when doing those things don't really make any sense. This sort of feedback can be great to have if it's on target, but is a time consuming (and expensive) process.
Mitigating the risk of responsibility against compromised visual design (and usability, functionality, et al) is a huge challenge. The balancing act is dependant on the people you work with and how you start the project. Be up front about the process of turning ideas into working solutions, and always include the client in the decision process along the way. Short but flexible processes for making collaborative decisions (like giving two, and only two, comps of a design) will keep things moving and let everyone feel like they are in control, which creates the 'buy in' you need to get a project completed (and paid for).