We used an abandoned mall in Korea as the setting of the Outlet Mall. As expected, wall posters and traffic signs etc. were Korean so we had to alter them using CG. At the top of the mall’s entrance, we created a CG gate and sign which we composited into shots. As for the overturned cars in the front of the mall, we brought in actual ones from Japan.

We probably should have placed a large green screen behind the mall before we shot. But such a screen would have changed the production costs substantially. We also could have used a small green screen that we moved for each shot but we weren’t using a motion control camera so background matching would have further added to the costs and the time, so we shot this scene completely without a green screen.

Not only for these shots, but throughout the film there were a lot of shots where ZQNs without arms, legs, and or heads came in contact with the background so it became necessary to mask by hand each time.

We also shot the highway scene in Korea as well. The background looked like Korea so through CG we added Japanese traffic signs. Because of budget restrictions we could only replace guard rails and center lines within a 500m area. So areas where we could not afford to physically replace guard rails and white lines were changed in CG.

On this project, there was a lot of partial CG work and on top of that tracking on models were expected to be of the highest precision. We weren’t employing a new technique but we were more mindful than we traditionally are. We often would discover tracking slips between frames when we played shots back on repeat. The precision of our tracking technology was one source of the slips but the errors were also probably due to the noise caused by the relatively shallow focus used in the live action shoot. Even when we would search frame by frame through a bad clip, we would never find the point where the slippage started. We had no simple solution and so all we could do was continually replay clips over and over while fine tuning and adjusting the tracking. For instance, in the scene where numbers of planes pass over a pedestrian bridge, there were no references in the sky that we could use so matching the shakiness of the camera was pretty difficult. If you look at the buildings in the background, it seems to match but the buildings are probably much farther.

Slippage adjustment lasted until the final stage of production. The film was shot handheld and without a camera rig so subtle blurring was unavoidable. This was one of the reasons it became difficult to determine how much the foreground should match the background.

Although we mainly created blood we also put a lot of energy into the fire and smoke effects. In regards to the moment in the first half of the movie where a storm of smoke rushes out of a city apartment, we used After Effects to composite the effects with the original footage. We made 10 different smoke patterns to allow us to best blend the composite. I personally think that our best work was the scene where, from the highway, you can see the smoke billowing from the distant apartment buildings.

The effects team workflow consisted of us first prepping the effect data and then just inserting them into a render-ready scenes created by the character team. Thanks to this workflow it made it easier to comp scenes later. The footage exported by the effects team and the footage exported by the character team were rendered in exactly the same environment so it made work smoother and we were able to learn, through trial and error, how to best show the blood and it’s movement.

In regards to quality, the first footage was extraordinarily good so it made the quality level of the work done on it look even higher. Doi-san, while on set, was able to make sure we had great footage to work from, which I think proved to be very crucial.