For (a random) example, what is the process behind thinking ‘I want two people sword fighting on top of a moving train in the mountains’ and seeing that formed on the screen. How are those images generated, is it purely animation, does it involve prop actors with green screens, etc.
In: Technology
Ex 3D animator here:
1. Everything gets modeled. The process is somewhat similar to making a sculpture out of toothpicks and marshmallows (connecting points and lines to form triangle and square polygons), and modern tools make this process a lot more fluid by allowing the artist to manipulate many points at once, automatically generate symmetrical models from one half of a model, and even allowing them to sculpt high-detail models like they would clay. For your example of 2 people fighting, the model (AKA a **mesh**) would be modeled to look like the characters who were fighting.
2. Texturing. **Textures** are the 2D images that are pasted onto 3D models sort of like wallpaper, giving them their color as well as things like small bumps and such that are easier to do in textures rather than in geometry. First, the model is “unwrapped,” where the 3D geometry is flattened out onto a 2D plane, and then the textures are made either in a traditional 2D painting application, or by painting them onto the model directly.
3. Rigging. “Bones,” are added to the model that allow it to be posed and animated.
4. Animation. The bones are posed, and multiple **keyframes** are made that allow for “tweening,” where the animator basically says “I want the model to be in this pose on frame #10, and this other pose on frame #27,” and the software automatically blends between them. When animating living things, **motion capture** can also be used to match the model’s animation to the captured motion of a real life person or animal. Either of these techniques could be used to animate a fight scene, it just depends on what style they’re going for.
5. Rendering. Computers don’t actually “see,” the model, all they see is a big list of points similar to [<0, 3, 2>, <3, 4, 9>, <20, 1, 68>…], and so at some point the computer has to turn those points into an image. There are 2 main ways to do it: **Raster Rendering** which is fast and used for most video games, and **Raytracing/Pathtracing Rendering** which is used for almost all movies, looks absolutely amazing, but takes a long time and involves the computer simulating millions of light rays bouncing around the scene similar to how light behaves in real life. Each frame would be rendered out individually (takes literal weeks or months to render an entire scene depending on size of the production), and then combined into a video.
6. Compositing. Think of it as sort of “video photoshop.” You’re taking the rendered out 3D elements, and combining them together with other 3D elements, 2D photo elements, as well as 2D video footage to make the final product.
Usually on a professional film, most of these steps are going to be done by separate people, and in the end it all comes together to form your finished scene. Here’s a great timelapse video showing all steps of the process where they make a dinosaur: [https://youtu.be/YsLt6lSl5ZM?si=_YAKMf0WtR8c-9ZU](https://youtu.be/YsLt6lSl5ZM?si=_YAKMf0WtR8c-9ZU)
Latest Answers