Why is 3D animation still so hard?
In the late 1970s the first 3D computer graphics software products were released, able to produce low-fidelity 3D representations of objects. In practice, it was time-consuming, expensive, and had low-quality results. One of the first popularisations of this new technology was in the original Star Wars movie:
Around the same time, the first camcorders, handheld devices that allowed for recording video to tape were being produced. The first camcorders were extremely bulky, expensive and produced low-quality video, the privilege of wealthy hobbyists and news organisations.
Fast-forward 40 or so years, practically everyone on the planet is able to record video for free (or at least, the cost of a phone, which people tend to purchase for other capabilities), on a device that fits in a pocket, at a quality matching anything produced in the previous decade. And yet,hardly anyone on the planet is able to produce media that is animated in 3D.
Why is this?
I first discovered computer graphics, like many of my peers, at school with Logo, software where you’d type in commands and get some art out (in many ways, prescient to generative AI, but I digress). It was painful to get it to do anything interesting, the results were always poor quality, but it was glorious, nonetheless. We were making art out of nothing, and having it displayed proudly on a screen!
In my early teens, I discovered CompuServe, and through that, POV-Ray. POV-Ray was free software that ran on home computers, and it was like Logo, but produced results in 3D, fully raytraced. My lifelong fascination with 3D graphics had begun.
POV-Ray worked by having you describe a scene by typing in commands to define and position geometry and their surface properties, lights, and a camera to view it all. You’d then effectively compile (“render”) the scene and get an image out. It was fairly trivial to make things like light refracting through a prism:
But it required significantly more effort to make (even slightly) more complex compositions:
I even experimented with making animated sequences, incrementally altering some of the parameters to create individual frames that could be stitched together to make a movie. At university, I documented my experience in trying to make a short film.
As much as I enjoyed producing the end result, I yearned for a way to do it that didn’t require such a massive investment of time (both for the trial-and-error required, and the many hours to render the single 320x200 pixel images before you could see what it looked like), and mental energy (it was essentially a form of programming, after all). Then in the late 90s, the first step-change in creating 3D animation happened.
In 1996, 3D Studio Max 1.0 was released. Two years after that, Alias Maya debuted*. What both of these products did was make the process of creating 3D faster and more accessible, in terms of cost, the requirements of the hardware they needed to run on, and crucially, in terms of interactivity. Users of the either products could view, create and manipulate 3D objects in a scene by dragging a mouse, without needing to wait for a lengthy render to see how the composition would look.
The next step change… never really happened. The paradigms established by the products of the late-90s are largely used today, even in newer products, and even though the functionality and quality of products has steadily improved. For example, the free and open source Blender, a favourite of independent 3D artists, is extremely capable, and utterly intimidating. And this is long before you venture down the rabbit-hole of third-party plugins and addons.
Of course, people do go on to overcome the challenge of learning how to bend these systems to their will and create some truly incredibly feats of artistry– one only needs to glance at some of the entries on pwnisher’s YouTube channel to get an idea of what individual artists are able to produce in their spare time using current tools.
Since the late 90s, the cost has gone down, even as the quality has gone up. But all the fundamentals of how 3D animation is made on a computer– the different concepts you are expected to know and understand, the challenges of translating ideas into pixels, have largely gone unchanged.
To animate a human, I first have to go through the difficult and painstaking process of sculpting the human in virtual 3D, typically starting with “primitive” objects (cubes, spheres, cones, etc), a process that can take an hour or more, even when you know what you are doing and aren’t looking to create a high fidelity result.
From there, you define and paint surfaces and textures, and crucially “rigs”, analogous to bones and muscles, which aren’t seen but will allow the human to move more realistically. All of this is incredibly hard, to the point that 3D artists will study anatomy in great detail to be able to do all of this better. Then you want to think about lighting the scene, and colour correcting the end result.
So to do 3D animation, you have to be a competent sculptor, painter, photographer, and taxidermist. Almost certainly, you should also be able to do some rudimentary programming, to be able to overcome the various challenges that the process will throw at you. Oh yeah, and you need to have all those animation skills too. See, the other big lack of innovation in the past 3 decades is computer animation is keyframes.
Keyframes were invented by Disney in the 1930s, and remain in use today (to the point they are a standard in CSS for animating web pages). The basic idea is you define the state for something at a particular point in time (or at a “key frame”), and define the state for that thing at a different point in time, with some interpolation of the intermediate points of time (frames). Originally this meant that a lead animator might draw one frame for every second of final content, with junior animators effectively drawing the intermediate frames (a process known as “in-betweening” or “tweening”). Eventually computers would take the role of tweening, and animators would just define key frames.
So everyone kind of collectively agreed that yes, this was the best approach, and even in the mid-2020s with the benefit of 20 years of innovation in UX design, coupled with massively increased computing power, that this is how a lot of animation (or at least most of the animation that has an artist’s input) is to be done; and instead the innovation would come from simulating hair and clothing, and creating new file formats.
People who have crossed the 3D animation rubicon, are of course, perfectly fine with this. They know how all the pieces fit together, and how to deploy an arsenal of different software products to achieve truly spectacular works of art.
But I believe that artistic mediums should be accessible to all, not just to the privilege of those who dedicated their entire lives and/or careers to it. What of the people who can dream up ancient civilisations but are incapable of drawing a person with correct proportions? What about the people who can reimagine an everyday scene with a novel twist, but don’t know how to write code? Did we leave them all behind? Is it ok for us to continue to do so? I believe we’ve failed to look outside of the bubble and try to welcome more people in.
Still, I guess they’ll all be fine, seeing as now anyone can create photorealistic 3D animation through simple text prompts:
*Both products exist to this day, owned by Autodesk, a company I previously worked for.