Your postdoc gives you an opportunity to demonstrate two key traits of successful researchers: independence and flexibility. Both of these traits are essential for succeeding in any tenure-track faculty position. Consequently, building a reputation for these traits is essential for being hired into a tenure-track faculty position. Faculty hiring committees look for people who will succeed on their own, and who will remain successful over several decades as the field evolves around them.
Independence
Right now your research reputation is deeply entangled with your advisor's. The research community knows your advisor better than they know you. If you've never published without your advisor, the community has no direct evidence of your research abilities. Even if you did most of the work, and even if your advisor tells everyone that you did most of the work, the community will give your advisor more credit than you. After all, your advisor's opinion isn't exactly unbiased; they have a vested interest in making you look good.
The only way to build an independent research reputation is to publish independently. At a minimum, this means publishing strong papers without your advisor. Even better is to write strong papers with new ideas/techniques that can't be traced back to your advisor. Solo followup papers on joint work with your advisor are good, but some people will still think the key ideas are your advisor's, and you're just cleaning up the low-hanging fruit. Better to strike out completely on your own.
Flexibility
Most PhD theses focus on a narrow collection of closely related problems, which are solved using similar ideas/tools/techniques. No matter how strong your thesis results are, if you've only ever done one "thing", then people will wonder if that's all you can do. So if all your research has been on reinforcement learning of human motion for animation synthesis from Kinect data, the research community will think of you as someone who works on reinforcement learning of human motion for animation synthesis from Kinect data. But what you want is for people to think of you as an expert in reinforcement learning, or an expert in animation synthesis -- or even better, an expert in machine learning more generally.
The only way to develop a reputation for broader expertise is to demonstrate broader expertise. That doesn't mean you have to suddenly switch to a completely different field — you do want to exploit the expertise you already have. But something significant needs to change—for example, statistics instead of reinforcement, or animals instead of humans, or robot motion planning instead of animation synthesis, or YouTube videos instead of Kinect data, or all of the above—that requires you to develop new ideas and new techniques.