The idea was to frighten the snail with a cool, fast and stylish car. So what would be better than a Fiat 126, aka Maluch? Apart from the fact that the Maluch is the only car I can distinguish without reading its label (I'm not a car specialist).
It's important to get as much reference material as possible, to avoid anatomical errors. Some anatomical "errors" are intentional for this piece however, such as the snail's mouth and teeth, but either way, it's good to know how an animal looks in real-life, even if you choose to make it cartoon-like. As well as using Google, I also like to search for references on Wikipedia, as there are often very big pictures to be found there. I always enable all available languages in order to get more results. For the snail I collected around 40 photographic references and for the car over 70 images, including blueprints. My younger brother took some photographs of Maluch details, such as the lights and side mirrors. Fig.01 and Fig.02 are just two of them.
For all objects I used box modelling, extruding and subdividing to get the right shapes. The following images (Fig03a - 03l) detail the way in which I modelled the snail's shell (well at least the general idea; there was also a lot more fine-tuning and tweaking done which isn't shown here). I started with box, and then added lines for the edge flow, and moved vertices and edges into shape. Later, I subdivided where necessary to achieve greater detail by connecting or chamfering edges (Fig.03a - i). Sometimes I pulled parts around with the Tweak/Magnet tools (which are similar to soft selection in Max) to get it into shape. The finished snail can be seen in Fig.04.
Fig.03a - i
The car model can be seen in Fig05, however car specialists have informed me that the wireframe is not clean (so don't look at it too closely) (Fig.05).
The background grass was modelled in low poly and distributed on the earth object with the hair and fur modifier (Fig.06).
I mapped all objects in Blender. Blender has a lot of advantages in contrast to Max, as far as UV mapping is concerned. I won't go much into the technical details here - just some general things...
One big problem with mapping are the seams between two separate parts of the UV map. There are ways to get rid of them totally, but I prefer to simply hide them, because it's much quicker that way. If you map an object for a still scene then make sure that you know the approximate camera angle before you start. That way you can hide the majority of the seams at the back of the model. If there are seams on the front of the model, hide them between folds or at the edges which point into the shape. Also try to avoid stretched textures. You can help yourself by putting a checker map onto your object. You will see where the checker is stretched or distorted and can fix it by adding additional seams. You'll also see how on some places the checker map is larger than on others. Generally try to avoid that; try to make the checker the same size everywhere. That way you won't have to scale your bitmaps differently for each area. Only regions with more detail should have more UV-space (that means the checker displays smaller in those regions).
You can control the size of the distribution of the map with the pin option in Blender; in Max you would do it perhaps using Soft Selection in the UV editor (which I don't recommend because it tends to distort your UV) (Fig.07).