As for visual debugging, these tools are really important because they a) convey information in a much denser form than you'd get from a tool and b) convey information which you can digest without already knowing the answer.
For example, unless you'd seen the oversampling in the picture, you probably wouldn't think to ask "Hey, do we have a lot of overlapping coverage in our walk here?". Similarly, in the case of the Nebraska problem, statistical methods would probably give you the a-okay until you actually saw the results in action, at which point you'd know something was off.
In games and graphics it's pretty much essential that you have some set of visualization debug setup to easily see what's going on--in very complex systems like physics engines it can be almost impossible to tell that something is off without having debug rendering enabled.
You have no idea ( http://trenchescomic.com/tales/post/9810 ).
As for visual debugging, these tools are really important because they a) convey information in a much denser form than you'd get from a tool and b) convey information which you can digest without already knowing the answer.
For example, unless you'd seen the oversampling in the picture, you probably wouldn't think to ask "Hey, do we have a lot of overlapping coverage in our walk here?". Similarly, in the case of the Nebraska problem, statistical methods would probably give you the a-okay until you actually saw the results in action, at which point you'd know something was off.
In games and graphics it's pretty much essential that you have some set of visualization debug setup to easily see what's going on--in very complex systems like physics engines it can be almost impossible to tell that something is off without having debug rendering enabled.