Post-apocalyptic car design
Tamás Gyermán shows you how to design and create a post-apocalyptic car using 3ds Max
In the short tutorial we can run through the process how we can create a post-apocalyptic car design in 3ds Max. You'll learn about methods and techniques, as well as design/visual development. I will show you how to prepare your scene, create the car, detail the model's topology, and understand surfaces and geometry.
I'm using basic 3ds Max for this project, with no additional plugins applied. First, open a new scene and save it in its own folder. Folder structure is essential, especially in a team project, where the folder hierarchy gives your team-mates a way to find raws and resources. Even when I'm creating personal works, using a folder hierarchy is a standard step in my creations. Names and paths must be similar and understandable. In this case I create three folders: model, texture, and references. If the project required more phases, like ZBrush or rigging, they should have their own folders. So I save my new scene into the model folder and start to work with it.
In this step we'll prepare the scene with all the raws we have. Markus Lovadina did a fantastic job with the post-apocalyptic car concept art, and now we're doing it in 3D. I push the (default) button 8 in 3ds Max, which brings up the Environment tab. Here, I load the concept art, then grab the channel and drop it into the Material Editor, using the Instance option. This is a linear workflow, so if you change something in the source, the copy is changed with it. It's important to be able to modify the images' size or placement in the background. This is very useful when we're working HDR panorama images.
The next step is to make our background element visible. Go to Views > Viewport Background and pipe in the environment background. Now we can see the image in the scene. Another way to do this is if we open the Viewport Configuration and choose Background Palette to load up an image. This is more useful for reference modeling, because we can match the viewport size with the image's size. Now let's start building.
I'm creating this great car as a high-resolution model. There are two ways to approach high-res modeling: static and subdivision. Static models have fixed topology that cannot be modified. So if you created a circle with 20 segments, it would look fine from a fair distance, but worse in a closer view. Static models work with any kind of wireframe, and are very useful for environment parts.
Subdivision topology is different, with many polygon rules, suited for more organic modeling. This car has surfaces that aren't so hard to create, so we'll build it with quad polygons. Why quads? Because the subdivision modifier creates more division inside the polygon to get fine shapes and surfaces. Five-sided polygons would also work here, because the subdivision modifier creates three good quad polygons from one five-sided polygon.
When creating subdivision models, we have to watch out for the distance between polygons. This isn't so important for flat surfaces, but much more so for bent ones. The best way is if we create polygon cuts with almost the same distance between them. If we can keep more distance between two polygons on a flat surface, this increases our polygons' efficiency: fewer polygons are easier to modify, and if we want to skin or sculpt the object, or number it with a displacement map, the topology must be in balanced distances.
Geometry is the one really important step to creating good stuff. The pipeline of subdivision modeling is very simple: create the raw shape, merge or hole out the required parts, and finally create the topology. I'm creating a small part here to show an example of my overall workflow. In the image, the red line is the shape edge, yellow is the support edge, and green is the subdivision edge. A simple box and two cylinders are what I'm creating now.
Very importantly, never forget to merge shapes into each other, as if they were real objects. If two parts are meant to connect, merge them properly, or they'll just go through each other and create holes. When you get to the subdivision stage, you'll have much better results. Of course, we should save time and polygons by interconnecting the hidden or less visible parts of the model as well.
The auto body
I create a camera object and try to capture a similar perspective as the concept art. It looks like a very tight-ranged one, with around a 30° field of view. It looks almost like an orthographic view, which is really useful because it removes perspective distortion and doesn't cheat the eyes.
I use a subdivided plane and a box object with a visible wireframe to find the perspective. Hit F4 to make the wire visible. I create a simple box and start to modify it with basic tools such as Extrude, Bevel, and Cut. After I create an object, I have to convert it into an editable poly, so the software will modify the geometry as I'd like it to. The conversion also means the software will start to mark modifications in its history, so we can go back if we do something wrong.
I use cuts, extrudes and vertex moves to find the overall shape. There are two main shapes to start with: the front and rear, drawn from the car at the bottom of the concept art. These are really useful to match the car parameters with the perspective ones. I do the same with other details, creating placeholders until I finish an overall matching shape.
Then I start to work with the level 2 car body's geometry. I use the same tools: Move, Cut, Extrude and Bevel. You can also use the Symmetry modifier to make modeling parts faster and easier. The final, full subdivision part is at the end of the modeling phase, but now I show each parts, how it start and going to be finish them inside one process image. I separate the mesh to several parts that it should be built up in real life.
Lamps and glass
As the concept art shows, there are separate plate parts for the lights. I'll handle the front parts this way too, so just separate a loop from the body by moving to Face mode, selecting a loop and detaching it. With this loop I can easily make the reflector's covering backplate, using Extrude Edges and the Move tool to create the shape. The best option for this is an 8-sided cylinder shape. 3ds Max helpfully gives us the ability to set the points into a circle, so after you have the backplate, select the required region and create an Inset. Delete the faces, then select the loop edge and around it.
In Graphite Modeling Tools, find Loops > Loop Tools and click on the Circle command. The script helps me instantly! If you're in another software and this feature isn't available, you can use Boolean cylinders to make the shape, then clear the vertices. When you work in circles, always use circles with an even number of sides: 6 is enough, 8 is perfect, 12 works fine, and 16 is also perfect. These numbers are always defined by the surrounding topology. For a realistic look, we must create a back structure to the lamp, and the lens geometry must be double-sided for Raytrace rendering (refraction).
Tubes, plates and welding
Working with tubes and plates looks very simple, but a good-looking tube connection requires a lot of time and patience. I use cylinders as a base to create my tubes for the car, and use the Quick Slice tool to create more loops where the tubes are connected. The connecting steps are the same for each of them: cut the lines, weld the points, delete the internal parts, and then remove the additional loops. For some other tubes and plates, I use additional extruded, flattened shapes to give the appearance of welded metal.
Surface elements and edges
Unfortunately this car is a post-apocalyptic one – it's not shiny and cool, but looks tough and like it's survived a lot. The body has been covered with plates, tubes, and planks. These can be done the same way as before, by cutting up primitives like boxes and cylinders and creating topology.
The important part here is to consider the thickness of materials and roughness of the edges. The plates have sharp, thin edges; wood and rubber are smoother. The geometry's edges help the brain to recognize the materials. Support edges and control edges are important at this stage. I create some wooden planks whose edges add a softer shape.
Fence and guide splines
I think the fence object is a really good part of the car, giving pretty fine shapes and details. I create six cubes and place them underneath each other, then rotate them 45? to make a diamond pattern. The touching corners should all be aligned. Select all the cubes and attach them together to make a single geometry. Now use Select Edge mode to select the edges on a zigzag path, as shown in the image. Under Edit Edges, hit Create Shape from Selection. This will create a perfect guideline for the fence wire. At each corner of the zigzag, I create one more segment in the path, and then push them to make the wire shape. In the Rendering tab, turn on the Enable in Viewport button with Radial selected to transform it into the required 3D geometry. I use low-segmented shapes to create the fence, because the poly counts could become massive after the copying and subdividing process.
I create some copies of this geometry to form the full fence, which I can easily transform using the Skin Wrap deformer, which works by creating a similar geometry that connects with the high-res fence. I create a simple plane object and connect it via Skin Wrap to control the surface of the fence.
Creating the chains
The chain is another eye-catching part for each model, and not so hard to create. I start by creating a pair of chain eyes using a Torus object. I merge them into one, and then fix the Pivot point to the rear of the mesh. In the Hierarchy tab, hit Affect Pivot Only then move it to the right place. Now the mesh can be transformed from this position and space.
The next stage is to create the full chain, using a spline and Path Deform modifier. From the Modifier list, choose Path Deform then click the Pick Path button. The chain eye geometry will shape to the path, but the scale is wrong, so scale it down to the right size. Using the Move Path button will also cause the geometry to start following the direction of the path.
Now copy the chain eye, hold Shift, and add the number of copies you want. Now the chain should work fine, but I find there's often a problem with the eyes lining up, mainly in corners. So right-click and convert it to an Editable Spline – this creates marker points so you'll be able to fix the path, or create some more random looks.
Hide the gun
The concept art shows something powerful-looking on top of the car – I imagine it's a Gatling cannon. To be loyal to the concept art, it'll be covered by some kind of fabric. A major part of this stage is cloth simulation: I create the cover's geometry with cloth simulation all around the gun. Because all of the gun's body is hidden, we're using SimGeometry, which works with bigger shapes rather than fine details.
I start making the simulation's silhouette using a few boxes and cylinders, and the cover's base mesh using a cylinder object (I delete the rear and lower parts, where the gun is connected to the car roof). This is a really simple simulation process, and the resulting geometry can be modified manually. I create the simulation geometry from the gun's group, using the Attach command. The whole mesh needs to be handled together in the Simulation properties.
I add a cloth deformer to the cloth geometry, and open the Properties tab, where I can also set the collider object. I set a few of the parameters and push the Simulate Local button. Only one thing is important here: I make two fixed parts around the cloth, which will be the parts pinned down by ropes. In Vertex mode, I select the areas that'll be tied down, and create a group from them. Then I click the SimNode button and click on the gun object – now the fixed points are functional. When you have the final cloth mesh, extrude it slightly to add thickness.
Other car details
For a realistic look, we have to handle the parts that aren't so visible too. To save polygons and time, I create placeholder silhouette geometries for the interior parts and chassis. These parts are created as subdivision models as well, but I don't add any fine details.
Tires and rims
Our car rolls on four beautiful tires, which we'll handle in this step. First I create a cylinder object – it's math time again! We need the perfect segment number to get a subdivided geometry that works well. Here, we'll go for 48 segments, because balancing the faces is always important in bent shapes. If we made an unbalanced topology, the shape wouldn't look consistently rounded.
I do a few Insets then select the parts which will become holes in the tire. I do another Inset on the selected (hole) parts. I use Swift Loop to add a subdivision control loop in a radial shape then delete the selected parts. Change to Border mode to make another selection and extrude it, moving the new edges further out to give us a thicker shape. After that, I add more support edges to the corners of the spokes to harden them up.
The rest of the tire is easy now. Use Swift Loop (under Graphite Modeling Tools) to fix the geometry subdivision. The tire has too many surface details to handle it as a connected subdivided mesh – the connected teeth and body would have about 170-200 segments, which would be terrible to handle with subdivision! So we'll just put the surface details onto the separate tire, by creating cubes and using an array object to place them. The teeth's pivot placement is very important here, because the array works in accordance with it – the correct place is in the center of the tire's body.
Rivets and screws are always cool-looking on sci-fi surfaces, but we must handle them careful, as they can dramatically increase our polygon count. I use really low-segmented base geometry to detail the surfaces, which works well enough on subdivision level 1.
If we know exactly which parts of the model will be invisible, I think we can delete these hidden parts. What do I mean by that? In real life, all things have double sides, but here we can delete the back face if we're really sure about the visibility. A render object can work well with one-sided planes, which is useful, because these parts decrease the polygon count and don't occupy UV space.
We can enhance our model's looks with some surface damage. I use freeform brushes to spin, blow, and smear a few parts on the whole mesh, making it look more real and weathered. Soft Selection is also a useful tool here, allowing you to grab and move vertices with the help of a falloff effect.
After completing the whole mesh, we can start UVW mapping it. This is a similar undertaking for each piece of geometry. We have to mark the seams and cut apart the mesh – the seams tell the unwrapping script where we want to cut out the geo. If you don't understand how it works, just think about when you're creating a box from paper: you draw six sides in a T-shape, and then stick the sides in – UVing is the same, just in reverse.
When we have all the UV islands, we should attach the parts to get an optimized, balanced full mesh with UVW map. Use the pack command to make balanced UV islands, which now become paintable.
Under Modifiers, choose the UVW Mapping deformer. Select the necessary edges, right-click, and hit break. The software creates a readable seam on the object. Now in the UV Editor, find Tools and click Relax. Here you choose the Relax By Face Angles command, and the script starts to handle your object. That is the modeling process finished. In part two we will look at adding texture to the model.