‘We did multiple simulations of what that would actually do to someone. It was horrific.’

Inside the often gruesome visual effects of Shōgun, plus a look at crafting the storm sequence, landslide, crowds of samurai, and 1600-era Japanese environments.

When Shōgun visual effects supervisor Michael Cliett was helping to craft the world of the FX series, authenticity was always key. “We wanted all the visual effects to of course enhance the story and immerse the viewer in the story,” Cliett tells befores & afters, “but we also wanted them all to be nothing too fantastical. They needed to be grounded in reality.”

Cliett liaised with expert Japanese historians and several department heads in accurately recreating Sengoku-era Japan, and in delivering dynamic action sequences such as the ship in the storm moments and a deadly landslide.

There were also battle scenes that required that kind of authentic attention to detail, and occasionally shocking up-close moments of blood and gore where visual effects were called upon to create the most impact possible.

Here, Cliett details several key visual effects sequences he oversaw for the series.

b&a: Tell me about the storm sequence. What could you shoot for that?

Michael Cliett: For the storm sequence we had about a quarter of the ship, Toranaga’s galley, and that was placed on a gimbal. The gimbal we got from the Peter Pan & Wendy film that Disney had been shooting. To put the whole ship on a gimbal was a big undertaking, so we knew we would have to do at least three quarters of the ship in visual effects in an extension. It rocked back and forth about 34 degrees in either direction. We surrounded that with 40 foot tall inflatable bluescreens on all sides. Cameron Waldbauer, our special effects supervisor, would then dump a lot of water on that and have wind machines going as well.

By the time we got into post, I would say we were probably about a ‘level four’ just from the raw plates. Even when we got the background and the raging ocean in there, we still needed to take it up a notch. So we ended up doing a lot of CG water. We did a scan of our practical galley set, which was a very, very accurate scan that we did with LiDAR. Then we used that to matchmove our virtual set to, and use the geometry to help simulate the water.

I would say about 60-70% of the waves and the water that’s crashing into the boat and running off, that was all CG. And then we added the little droplets to the lens, or made the rain go sideways. Sometimes the flags weren’t quite moving fast enough and we really wanted them to look like they were in 100 mile an hour winds, so we had to replace those with CG.

Important Looking Pirates was the vendor on the storm. They are brilliant artists. I can’t say enough good things about ILP. Philip Engström was the visual effects supervisor and Chelsea Mirus was the project manager. ILP did a majority of our water work for the series.

b&a: I wanted to ask you about another fluid-like sequence, and that’s the landslide. How did you approach that one?

Michael Cliett: On the day of shooting, I took up a drone, and we did aerial photogrammetry of the mountain that was there right next to our practical set. For the ledge looking out at the camp right before the earthquake, that actually looked out onto a body of water. So we had to add the encampment and we swapped it out with mountains. Everything that’s surrounding the practical set on three sides was visual effects.

ILP took the photogrammetry and built a CG duplicate of the mountain that they could use to run a sim with for when the landslide came down. For the big wide shot, when you see the whole thing coming down, that was almost 99% CG. We had some practical actors on our practical set there that then we comp’ed in.

For the shot where the two Samurai get swallowed up by the ground, our stunt coordinator Lash (Lauro David Chartrand-DelValle) put two guys on these rugs that were buried under the dirt. When they were supposed to slide, they would pull the rugs and the guys would fall down and start sliding on top of the dirt, and it was a good start. We always knew we would need to add a full landslide simulation under them. In post we realized we wanted to have the two guys get swallowed up by the ground and die in that shot. So, it had to be definitive in that we couldn’t just see them slide away. They really needed to get swallowed up by the ground.

Ultimately those two practical guys were great reference for the motion of them falling down and then starting to slide. We match moved geometry to the practical samurai to interact with our full ground simulation and then had the ground swallow the poor real samurai meeting their unfortunate end.

We did find actual videos of earthquake-induced landslides that we matched to. Recently there was the earthquake in Taipei, and there’s a video of a landslide happening from that earthquake. I saw the video and I thought, ‘Well, yeah, there it is. We got it right.’

b&a: There’s a couple of battles and also crowds of soldiers in the series. What was your methodology for filming what you could but also doing crowd replication?

Michael Cliett: We never had too many extras on set. The most we had on a given shooting day was in episode four, which was for the army on the beach, where we had close to 120 extras. But of course it needed to more like 10,000. We wanted to honor the amazing costumes that our costume designer Carlos Rosario had designed and made, so we did a lot of intricate scanning. We also did photogrammetry for the extras. T. Parker, our VFX data wrangler, would take 300 photographs of an extra or a main cast member, sometimes against bluescreen. And then we could build them up that way.

In episode eight, we had a scene where they’re walking into Edo for the first time. We had 40 -ish people there. That was it. That was filmed out in a cornfield just outside of Vancouver. We had to make it look like it was rice paddies and with Mt Fuji and a city nearby, too.

ILP did the Toranaga brown army in episode five and in episode ten. Goodbye Kansas handled Yabushige’s green army and Saiki’s silver army. SSVFX was building Osaka and so they did crowds for the Osaka set extensions.

b&a: For set extensions, what were the kinds of conversations you had about what could be built for real, and how much you’d be doing in visual effects?

Michael Cliett: Well, Shōgun is historical fiction. It’s based on actual events and places and people from 1600-era Japan. So we wanted to make sure that the world we were building was also historically accurate, right down to the smallest minute detail. We spent many months researching all of this, myself and Helen Jarvis, our production designer. We really got into it as far as what the structures would look like.

Frederik Cryns, a professor of Japanese history at the University of Kyoto, he’s the world authority on this period in Japan–the Sengoku era. There’s a lot of material out there, but a lot of material isn’t accurate and isn’t true, so Frederik was able to really weed it out for us. He’d say, ‘This is what the Tenshu at Osaka Castle would’ve looked like in 1697. Here’s a blueprint. Here’s a model that they did at this museum based on this artwork, and this is as far as we know the best representation.’ It meant we had an actual blueprint before we built the main structures of the castle.

Helen Jarvis and the construction team built several city streets of Osaka along a waterfront just outside Vancouver. Those buildings were a fantastic representation of what the merchant district would’ve looked like in Osaka. Our set dec team did a phenomenal job of populating all of that. We scanned a ton of their stuff, too, some of the props and the set decorations that we could populate into the visual effects work.

b&a: There are times where we see some pretty crazy blood and gore and limbs coming off and everything. How did you manage that side of the VFX?

Michael Cliett: Most of the blood and the gore was done by Refuge VFX. They did really great work for us. Probably our most gruesome scene would be the end of episode four when the cannon chain shots are fired through Ishido’s guys. We approached that from two different directions. One is that we removed limbs and body parts from real guys by painting them out frame by frame and then animating a CG version of that arm flying off or this limb flying off. And then we also scanned all of those stunt guys, too. We scanned all of them because we knew we would have to have some of them as fully digital reproductions in order to separate their rib cage from their torso and the rest of their body.

Lash our stunt coordinator actually had his guys all set up with ratchets, which are basically hydraulic pulls that would pull them about 20, 30 feet through the air. On cue, three or four of them at a time would get pulled by a ratchet. We tied pink ribbons onto various places on their arms, on their legs, some around their midsections. Those ribbons would signify where we wanted the bodies to separate on impact.

For the guy who falls off the horse, Lash had one of his guys do the stunt. The horse reared back, his guy fell down on the ground, and that was great reference. We ended up covering him up with a digital double and matchmoving him and then painting out the practical guy frame by frame. With the CG guy, we had the chain shot go right through his middle section and his guts kind of spilled out on the ground and he rolled down.

That chain shot is supposed to be two cannonballs tethered together by a chain used for naval combat. Ships fire these at each other for maximum damage to their hulls. Those would be flying through the air at supersonic speed, and this chain would just rip through someone. We did multiple simulations in the computer of what that would actually do to someone. It was horrific. But, again, this was another good example of us trying to work out what would actually happen if a chain shot was fired through human beings.

There’s also the beheading in the first episode. Omi just out of the blue cuts off this Christian peasant’s head. It’s a shocking moment, which is exactly what we wanted the viewer to think. That went through multiple iterations of physics animations just to get the rotation of the head correct and the speed at which it drops.

My wife has seen that probably 20 times, and she still gasps every time that that happens on screen. We were down in LA for the premiere in February at the Academy Museum Theater. There’s probably a thousand people there. And when that happened, the whole theater was just like, ‘Oooh.’ There was a collective gasp, and I was just, ‘Yes, that’s great.’ The person next to me probably thought there was something wrong with me because I was just cheering the guy’s death and the beheading, but they didn’t know who I was.