The average gaming computer spends a lot of time putting together models. It does this using a collection of triangles, essentially ‘glued’ together to create a three-dimensional shape. These structures tend to be built by specialists, usually on high-end workstation machines.
If you’re interested in games, then you might wonder if it’s worth dipping your toes in the world of 3D modelling. But, can you make a start with the gaming machine you’ve already got?
You might expect that the computer used to put those models together would share a lot with the one that’s designed to just render and display them in a game environment. But, as we’ve discovered in recent entries here, things are a little bit more complicated than that. Whether you’ve already got a gaming rig and you’re wondering whether you can safely use it to create 3D models of your own, or you’re in the market for a new machine, it’s worth knowing what components will make the difference.
Let’s delve into exactly what separates a high-end 3D-modelling rig from the equivalent gaming rig.
Lots of vertices
Today, 3D models of complex shapes (like faces and people) tend to go through a multi-stage process.
To begin with, a high-res mesh is sculpted from a kind of virtual clay, using a program like Blender or Zbrush. These sculptures offer exceptionally fine detail, with every wrinkle and blemish modelled. Once you get really good at sculpting, you might find that you need millions and millions of vertices to get your vision across – and when you’re not good at it, you might find that you end up using more than you need, anyway. All of those polygons need to be stored in RAM – which means that those working with high-poly models might prefer 16GB to 32GB.
Of course, when you’re designing a game world, then you have to make sure that the models you’re using are as lightweight as possible. Every extra polygon, after all, is going to burden the machine on which the game is ultimately played. What’s more, we need to worry about things like topology, which will determine how your model deforms (if it’s being animated).
For these reasons and more, 3D artists mostly prefer to work backwards from a super-high-detail model and trace a low-poly, deformable version over the top. This is called retopology. It’s not the only way of modelling – the introduction of ultra-fast asset streaming on the new consoles (and on high-end PCs whose graphics cards can pull data directly from the storage drive) is likely to change things.
All of this is a very longwinded way of saying: you probably need a little more horsepower than you’d be using in a game, especially if you’re sharing assets between multiple users. If a 3D artist collaborator sends you a high-poly model to open, then you’ll want to be reasonably sure that you can at least open it without your machine dying a horrible death.
When you’re learning the ropes, of course, you might not bother diving into sculpting at all, and just try and make low-poly models, to begin with. In many cases, this will encourage you to get a good grasp of the fundamentals before you start piling on the detail: you can get started with simple 3D models even on a relatively puny machine.
In a game world, all the frames are rendered just as they’re being displayed. The longer it takes for a given frame to render, the slower the framerate will be.
But what if you’re creating models and animations that won’t be rendered in real-time? You might be putting together hyper-realistic models for architectural visualisation, or you might have aspirations to one day work for Pixar, creating expressive models and scenes for them to inhabit.
In such cases, you’ll need to make a switch to a non-real-time render engine, like Cycles, which is one of several options in Blender. If you’ve ever heard stories about how long it took to render a given impressive scene in a big-budget CGI movie, then this is what they’re talking about. If you’re piling on effects and using huge numbers of samples to generate the image, then the render time is going to quickly skyrocket.
Faster render times will help you to get results and progress your skillset, as well as making easy adjustments. If you’re rendering a fancy scene in your favourite modelling program, then there will always be a sense in getting that render back as quickly as possible. Moreover, you’ll be able to experiment more without spending as much time waiting around for the results.
Stacking modifiers & multiple threads
3D modelling directly strains the CPU. This goes especially when you begin to stack modifiers on your model. Modifiers are a form of post-processing that selectively tweaks your model based on pre-defined criteria. For example, a surface subdivision modifier might double the amount of geometry in a model, creating curvy bits and softening edges, without making the model itself more complicated to work with.
Modifiers stack on top of one another. For example, you might be building a car, which you intend to be symmetrical. You’d therefore build an enclosure for one headlight and create a mirror so that the same thing is duplicated on the other side of the car. Then you might apply a bevel modifier to sharpen certain edges, and a subdivision modifier to smooth others.
While all 3D modelling software approaches this sort of thing differently, they all share a common drawback: these processes must be performed sequentially, on a single core. This is because they’re necessarily being done one after the other. As such, you might get better performance on a CPU with fewer cores and fewer threads, but a higher clock speed.
The exception here, of course, occurs when you’ve got several different models in the same scene, as is often the case with complex interior scenes. Here, many cores might be advantageous.
Putting together a flexible rig
If you’re just starting on your journey in the world of 3D, then you might not have determined where you’d like to specialise. In this instance, it’s better to have a more generalised machine that can do plenty of things well. This might include fiddling with shaders, physics, and rigging and animation.
Fortunately, this flexibility is exactly what’s required of a decent games machine. Different games tax your machine in different ways, and so a custom games machine from Cyberpower will tend to offer a balance of strengths.
Once you’ve got a strong idea of what you need and why you might tweak your build to suit your purposes. For example, if you’re concentrating entirely on 3D modelling, then you might put more emphasis on fast single-threaded CPU performance; if you’re going to be using light-bounces and hair simulation, you might go for something more generalised.