I like to tackle my models head on.  For me its easier and faster to generate a concept by diving in, rather than stopping to think about it first.  I tend to problem solve and not over think my models when I do this.  Because of my extensive training in physical sciences, I tend to make my models to 1 : 1 scale.  This way if I find a physical reference object I can measure it in the real world and translate those dimensions much smoother into Maya.  I always try and balance my models based on where they are in the shot, allowing me to simultaneously practice low poly and high poly modeling in the same project.  I excel best at the creation of hard surface object models, environments, and digital sets.

When it comes to dynamics, however, I take a more methodical approach.  Its where the scientist in me kicks in.  I step back and determine what I am trying to accomplish with the effect and the work through a number of solutions in Maya to develop the base portion of that effect, by utilizing my knowledge of real world physics and dynamics.  Then I move the rendered out effect asset into After Effects, and let the artist in me take over.  In After Effects I begin embelishing the effects in an effort to not only make it more dramatic/cinematic, but to also try and achieve the final look I set out to create.

When it comes time to put it all together with live video, I like to try and establish a number of looks and test them out to see which one works best.  I start by going through each shot and tweaking different aspects so that my CG elements match my Live elements.  Once I have all my shots locked down I like to go back over them all as an edited whole and begin adjusting the overall values of color, light, noise/grain, etc. to create the desired mood and feel of the over all piece.