It was great working on shots where a character’s face transforms into a ZQN. In the moment where the transformation occurs, blood vessels start to rise to the surface and break out all over the victim. We used CG to create this moment but to get it to look right an extremely difficult type of tracking was required. We took one frame, cut a mask out of it and then used it to spread the effect. But as the effect spread to shadowy areas in the original footage, it was insanely difficult to make the effect match the same shadowy feel of the original footage. We placed markers where necessary on the actor but using these markers to figure actual measurements on across the skin surface became arduous. The skin often expanded and contracted so even with the marker points, slippage was unavoidable. We used Nuke’s Mesh Up and Gridwarp points to constrain the points and as we controlled their relationship to each other via “expressions” we constructed the transformation such that it actually looked like it was flush to the skin’s surface.
It probably would have worked better if we attached more markers to the surface of the skin for this effect. But during shots, we often saw sweat cover markers in the middle of a take so if it’s difficult to judge whether or not increasing the amount of markers would have helped. As expected, editing these blood vessels parts took the most work and we eventually would have to delete the markers from the shot. Sure putting more markers in could have helped us, but we’d also have to spend more time deleting each one too. Luckily we didn’t have too many to erase (laugh). All we could do was feel things out.

Within the A-Rank ZQNs there is a character of the ultimate rank called “High Jump”. Unlike other ZQNs, his character required various types prosthetics and VFX depending on the scene. To have a zombie that transformed from scene to scene probably almost never occurs in other zombie films.
When we worked on the destroyed model, a Japanese prosthetics team made us a detailed model which we then 3D scanned. It was useful for us to receive a model from the prosthetics team for two reasons. One, we were able to work on a CG ZQN model that was the exact same design as the non-CG model. Two, by scanning a near complete model and being able to look at the actual model while working, we were able to do accomplish two things: shorten CG production time, and spend more time refining details in the model. Including High Jump’s final scene, we were able to destroy ZQNs using pieces created by the character team and created the explosion of small detailed pieces not through boolean operation but part by part by hand. So we were able to scatter even small pieces like the eye and tongue. These models were then used as a base by the effects team to create secondary destruction. A broken piece hit by a bat would then once more break. From here, blood would then ooze out.

Regarding the amount of blood seen, we thought through trial and error whether to go for a realistic amount vs a stylistic showy look. There were shots where we definitely pushed what was a realistic amount of blood.

We’d try to render it once, then realize there wasn’t enough blood and then we’d add more. Sometimes adding blood was just for aesthetic purposes too.

The ZQN eye is completely CG. At the beginning we considered using contacts on the eyes but applying them to a ton of actors for every scene proved to be expensive so we decided to try out doing the eyes using VFX. Regarding contacts of course there was some concern about safety but the actors themselves also just seemed to hate the idea. In traditional zombie films, eyes shine, get bloodshot and then turn white during zombie transformation scenes. In this film after talking with the Director we decided to create the “white eye” look by congesting the eyeball with blood, then adding a sense of cloudiness to the iris. When the iris is still black the victim is still human but the moment that he turns fully into a ZQN is when the pupil itself gets cloudy. Here we congested the pupil via CG. We created the eye by downloading 3D eye model into Nuke and then adding a texture to it.

This time there was a considerable amount of work where we had to adjust the color of the on-set footage to blend with the HDR footage. For all of these, we took color charts and performed color matching on rendered physics-based CG footage after it had been composited. This was a necessary part of the process for this project. In the past, there was a few times where we matched color purely with our eye. With the work we did for the eye and the process of using Nuke to download the model data, we learned a lot on this project.