There’s something truly astounding about deconstructing an object: the everyday becomes much less so when you break it down into its constituent parts. Take a look inside any day-to-day device and you’ll find intricate mechanisms that work together in the aid of one, singular goal. Within the pins, cogs and wheels there are flashes of human ingenuity; the engineering relaying a story of how the object came to be.
It’s that sense of wonder that underpins Discovery Science’s enlightening Machines: How They Work. In the show, meticulous CG is used to break apart commonplace items like guitars, escalators and sprinklers. The camera swoops through and around the hundreds – if not thousands – of parts that turn these inanimate objects into utilitarian aids or instruments of art.
Montréal’s Digital Dimension was behind each of these gloriously detailed vignettes, working to pull the objects apart in 3D exploded diagrams before gradually pulling them back together.
And it was no easy task. Machines: How They Work required 160 shots at 45 seconds each, revealing a dizzying array of varied contraptions over two-hours of full-CG runtime. Given the series’ strict television turnarounds, Digital Dimension needed to deliver 16 shots every two weeks to stay on deadline. The pressure on rendering was staggering – Digital Dimension needed to render 36,000 frames per week.
But, despite the demands on quality and quantity alike, Digital Dimension knew it could turn to Redshift to get the job done.
We can rebuild you…
It bears repeating: 36,000 frames rendered once a week. That’s impressive enough of a feat on its own; more so when you consider the results. Breathtaking in their polish and detail, each of the 160 shots reveals a gloriously detailed breakdown of the object in question, perhaps showcasing the individual pins and bolts that enable a paintball gun to fire at such high velocity, or the internal mechanisms that allow an ATM to safely distribute cash.
“The Machines: How They Work project was something of a sequel to a project completed the year prior called Rise of the Machines,” explains Francois Beaudry, CG supervisor at Digital Dimension. “Rise of the Machines was completed using an older rendering solution, however, and that meant we needed to simplify as much of the 3D as we could.
“For the project to be possible within the timeframe, it was necessary to remove most shadows and reflections, or simulate reflections through the use of environment maps,” he continues. “Only the first and last 40 frames of each shot were rendered with full quality to match the plate, which we transitioned from once the object was exploded.”
For Machines: How They Work, things were different.
“This time, we had a bit of a surprise for the client…” says Beaudry. “Redshift allowed us to maintain a consistent level of quality throughout all shots, while still achieving much faster render times than the year prior. This really simplified our workflow, as we didn’t need two sets of shaders or render settings per shot. That made the whole process much more manageable than the year prior – while also looking much better!”
Quality and quantity – you can have both
Beaudry’s experience with Redshift stretches long before his time with Digital Dimension. During a former professional role, Beaudry was in fact a Redshift beta tester, battle testing the product on the front lines before it was released to the public.
“Two things have always attracted me to Redshift,” reveals Beaudry. “First was obviously its incredible speed – and it really is incredible. But there are other benefits too that shouldn’t be underestimated. Redshift’s GPU-based nature makes for a very low cost for deployment – simply adding a graphics card to get up and running is a very favorable option for any studio environment. It allows us to use any workstation, old or new, while gaining a huge boost in production.
“In the case of moving on from our previous rendering solution, that boost was immense,” he continues. “We were able to cut render times per machine to five per cent of what they were prior.”
A 95 per cent drop in per machine rendering time is very much something to shout about, given how much free time it gave the Digital Dimension team to iterate. Faster renders meant more time to look at each rendered element of the show’s 3D break-apart diagrams, ensuring each individual element stood up to the quality of the last.
“Even when testing Redshift, it was obvious that it was going to completely change how we worked,” says Beaudry. “Rapid iteration is the backbone of an efficient creative process, and with Redshift our artists could work much faster, which in turns mean they could push whatever they were working on much further in the same amount of time.
“Instead of settling on version two or three, the artists could push their renders to version 10 or more!” he continues. “As you can imagine, this greatly increases the quality that an artist can achieve. And not just that, but hey can actually experiment with different ideas, instead of simply going with the safe solution. If you want to stand out in this field, that’s how you do it.”
Biased towards biased
It’s not just Redshift’s raw speed that has impress Beaudry and his colleagues, however. They have also found reason to praise the impact this speed has on other aspects of the software’s make-up.
“Thanks to its speed, Redshift doesn’t suffer from some of the problems that often arise with other biased renderers,” explains Beaudry. “For instance, when using other biased renderers on the market you often have to spend a lot of time optimizing your renders. I’d often get flicker in my global illumination, meaning I’d have to bake it to prevent that from happening. That takes time, and time is a very precious resource when you’re working on a project like Machines.
“With Redshift, however, the line between having to bake your GI or simply increasing your settings so nothing flickers gets completely blurred – the time difference is truly miniscule,” he beams. “We also recently worked on the operators cinematics for the recently released Rainbow Six, and only two shots had baked GI. You just don’t get that kind of quality with other biased tools.”
The human touch
Technological prowess isn’t the only thing Redshift offers Digital Dimension. There’s the human element to consider, too – and to Beaudry at least, that’s just as important as many of the technical specs.
“The support offered by the Redshift team is nothing short of amazing,” he says. “I’m consistently blown away by how fast they get fixes ready whenever issues arise. It feels like they’re a part of my team – I know that if an issue comes up, Redshift will be there to help.”
That’s important on any project, but on a project like Machines: How They Work, with deadlines and quality expectations ever-present, it’s vital.
And thankfully, the results have more than impressive. Indeed, when the Digital Dimension team sent its first WIP of an exploded object, showcasing the new and consistent level of quality, the client was “quite ecstatic”, to say the least.
“Redshift truly helped keep this project on track, allowing for easier renders, faster, and with a higher overall quality,” concludes Beaudry. “If you’re achieving two of those things, you’re onto a winner. Hitting all three out of the park is something else entirely. There’s nothing everyday about that.”