Table of Contents
How did Sony Pictures Imageworks produce the Netflix series Love, Death + Robots’ real-time thriller? The eighth episode of volume three, titled In Vaulted Halls Entombed, is about a group of troops who have the risky task of rescuing a hostage held by terrorists. However, when they descend deep down, they encounter the ancient Lovecraftian evil.
The Sony Pictures Imageworks crew, who had previously provided VFX on everything from Jumanji to Spider-Man: Far From Home, was immediately drawn to the story.
That freedom required Chen’s team to do their most challenging experiment yet: converting a dated pipeline into one designed for real-time rendering in Unreal Engine.
The conventional pipeline was very reliable, strong, and stable. It featured a vast code base, rigs, shaders, and rendering configurations. They figuratively backed away from every last aspect they had tried to perfect as well as all of the custom software they had created.
Despite having no prior real-time production experience, a team of 45 artists produced four distinct environments, including the Afghan mountains and an evil lair, six photoreal human characters with flawlessly executed animated fights, dialogue, and mutilation FX, six distinct environments, and even a Cthulhu monster in just four months.
It was a project that, as of right now, has permanently altered how the Sony Pictures Imageworks team operates.
>>>Read more:
- Love, Death, and Robots – Adult TV Animation Series from Netflix (Part 1)
- Love, Death, and Robots – Adult TV Animation Series from Netflix (Part 2)
- Love, Death, and Robots – Adult TV Animation Series from Netflix (Part 3)
The road to Love, Death + Robots real-time thriller
Choosing the right team was the first stage in producing the short. 900 Sony Pictures Imageworks artists were surveyed to determine their interest in real-time rendering. Those that applied were thereafter interviewed and given a briefing on the project.
Once the 45-person crew was assembled, they started building what Oddy refers to as the “Nerf level”: the short’s basic world and camera setup. After completing storyboarding the 15-minute film in the engine using objects imported from ZBrush and Maya, they blocked each scene.
The three hero characters were created using scans of three actors, while the three redshirt characters were created quickly using MetaHuman Creator.
It’s interesting to note that this was the first time a MetaHuman had been included in the creation of real-time linear content; this feat, which was made possible by Sony Pictures Imageworks’ insights during the early stages of development, not only succeeded but also helped MetaHuman Creator advance.
A streamlined production
Layout and blocking were finished, and then the crew turned to animation. The initial animation pass for this was finished utilizing motion capture and Xsens suits.
There was no need to envision the final shot on set since Take Recorder allowed stage performers to see how their gestures were linked to characters in the Unreal Engine world. As a result, decisions might be made throughout production more like in a live-action movie.
Most of the time, they didn’t enter a shoot with fully developed scenes. While they were unsure of the precise method they wished to use, they were aware of the basic course of action that had to be followed.
They could then choose alternative takes quite easily using the Sequencer in Unreal Engine. That meant that instead of asking for and waiting for a re-rendered shot, they could just go through and pick another choice if the performance wasn’t the correct one.
Oddy said that the team’s flexibility to make adjustments enabled them to operate more quickly and cooperatively without being slowed down by prospective reworks.
Traditional pipelines make it more difficult to start opening up the shots without going all the way back to the pipeline’s beginning as production progresses and things become more galvanized. Real-time does not allow for going backward.
Simply put, you are operating in the present. You can always come up with fresh ideas along the way, even with a smaller crew. Even at the last minute, the movie can be recut.
Following the completion of the mocap animation pass, the results were cleaned up in Maya prior to sending the FBX rig data back to Unreal Engine.
Following the completion of the mocap animation pass, the results were cleaned up in Maya prior to sending the FBX rig data back to Unreal Engine. After that, all final-pixel rendering was completed in Unreal Engine with very little post-production.
They can see their concepts in full, rather than simply play-blasted gray images, thanks to Unreal Engine. As a result, the motion blur in their final rendered frames from Movie Render Queue matches exactly what they were seeing in the viewport. The post-production procedure was quite simple; they merely added film grain and lens flares to several pictures.
Additionally, anything that had to do with the bugs in the tunnel or the exploding monster effects was included in the render. They were not required to add anything.
Ending silos
The In Vaulted Halls Entombed team’s use of a real-time workflow encouraged artist-driven creation in addition to speeding on-set production and post. The departments at Sony Pictures Imageworks immediately broke down their silos.
Without having to wait hours for a render, artists could watch sequences play out in front of them and make changes without involving another department. As a result, people could work on multiple projects at once, improving everything from lighting to animation on the same platform.
Currently available on Netflix is an episode of Love, Death + Robots real-time thriller: Vol 3 titled In Vaulted Halls Entombed.
–
Animost – Vietnam 3D Animation Studio
hello@animost.com