For this project I had the idea of two cars wrestling with each other. Originally I had it the other way around, with a larger car (4x4 etc.) overpowering a smaller car. But a moment of inspiration sent me in another direction, and I decided to switch the roles of the cars, changing large verses small, to "efficient" verses "gas guzzler".
I wanted to attempt a very illustrative style for the image; quite colourful, distorted and almost cartoony, whilst keeping some realistic elements to hold it all together. I also wanted to treat the project as if it were a job, so this meant setting myself a deadline and trying to be as efficient as possible - such as using a lot of post work and models already at my disposal in my library (Fig.01).
The first thing I did for the scene was to bring in the car models, place them roughly in the right position together, and pin down an appropriate camera angle as early as possible - to make the post work easier for one thing. I had them sitting on a basic floor, which received shadows but had no diffuse as this was painted in 2D in the compositing stage.
The car models were lit by placing a few large Vray area lights - considering the nature of the scene and the use of such large area lights, I decided I didn't really need any global illumination. These lights were all set to invisible, not reflection or affecting specular, so that "lightboxes" around the scene were used for something to reflect. In the image below you can see the lightbox textures I painted in Photoshop that were used so the lightboxes were not just a pure white material (Fig.02).
In the rest of the scene, the ropes, corners and the lighting rig were made to suit the camera, so if you view them from another angle you can see that they are quite a strange shape, but this didn't matter as I only intended them to be seen through the one camera.
This is also true for a few other things: for instance, the front wheels on the car on the bottom are stretched out twice the distance they should be from underneath the body, but without that it was hard to read them clearly.
For rendering, I made several different passes, as well as bringing in other elements and painting in Photoshop itself. The base layer was all 2D from Photoshop; this included the painted floor, tyre tracks and a rough smoke type effect to break up the background (Fig.03).
The next layer was the 3D background elements; again, these were painted either to fade away, or to add highlights and depth of field (Fig.04).
For the lights on the rigging in the background, I actually took a photo of my LED lamp out of focus. I then colour corrected and warped it to match the camera's distortion. This was much more practical than faking it - when I could easily use a real one! (Fig.05)
The next layer was the 3D cars, which contained their shadow in the alpha channel of the image from Vray. Again, specific areas were colour corrected and touched up in Photoshop (Fig.06).
The final element in the composition was the volumetric lights. These were rendered separately and in two passes: one for the background lights and one for the car's headlights. These were then layered onto the other elements with 'Screen' (or you could just as well use 'Linear Dodge' which has now also been labelled correctly as 'Add' in Photoshop) (Fig.07).