The Spine of Night is an independently-animated film, produced and drawn frame-by-frame by a small team of animators. Inspired by the fantasy genre of the 1970s, but not based on any specific work, the film’s characters were performed by Richard E. Grant, Lucy Lawless, Patton Oswalt, Betty Grabiel and Joe Manganiello.
Philip Gelatt and Morgan Galen King directed and edited the feature, and recently took some time to explain their filmmaking process. The following is their own words:
The place to start describing how we animated The Spine of Night is to note that it was drawn by hand, frame-by-frame, on a computer. This is important to note mostly because it is unlikely anybody will ever do it this way again, as across the industry there are technologies emerging that are attempting to approximate the look through AI and motion capture, and it’s only a matter of time before the right set of tools comes along and is virtually indistinguishable from the relatively old-fashioned approach we embarked on in 2014 — itself a holdover from the short films I’d been doing the years prior.
Fundamentally, rotoscope animation is the filming of live action footage, then using that as a direct reference for the animated elements. From the outside, this is often reduced to ‘tracing’ the footage, but that’s not really how it works — at least not in the style we were working in. While there is some element of using the hard outline of parts of the real body — the fingers of the human hand, for example — mostly what you’re doing is working from the lines of the ‘key’ animation that dictates the art style, since there are no ‘lines’ on the human body, of course, much less costumes, props, and special effects.
The reference footage is primarily useful for maintaining the sense of recognizable realism to the way the characters move through space. In cartooning, you are fundamentally working with rubbery, exaggerated ‘squash and stretch’ techniques to reproduce the feeling of movement and momentum for non-human beings, but rotoscoping allows you to treat the characters as physical beings with understandable feelings and limitations. No doubt that’s why its use over the last half-century has been tied to adult animation, what with it’s non-cartoon preoccupation with the sensual extremes of human violence and eroticism. That’s certainly the appeal for us both, at least.
From a process perspective, we started with a large warehouse space in an old papermill building in Pawtucket, Rhode Island. While a green-screen room probably would have been helpful, we filmed against the dingy white walls and pipes on a large blue rug. The rest of the set was made up with some stacked apple boxes, some not-soft-enough gym mats, and a rolling painter’s scaffold that acted as prison bars, a massive wagon, a castle wall to scale, and pretty much anything else we needed.
The actors largely dressed in costumes made from a cobbled together mix of cloth, cardboard, and bubblewrap mailers, and armed with mostly wooden dowels of different lengths, since we were primarily looking to just get the rough structures of the characters’ designs. I’d say the costuming was a mixed success — sometimes helpful, but often so floppy it obscured details you’d have liked to have seen when it moved out of position. We filmed some pickup shots, later, with shaped heavy foam, and that was much more effective.
There was extensive storyboarding to work from, pinned to the wall, and crossed off as went, but the set was loose enough that we were able to maintain a sense of improvisation, which is another area where having a team of live actors makes rotoscoping a really dynamic process. The physical and performance choices of the actors come through in ways that add diversity and eccentricity in a way that an animator alone wouldn’t think of adding.
We had originally set out to capture all the performances live — and cast the actors as such — with everybody equipped with lav mics. While this made for great scratch audio to animate from, ultimately, the audio captured that way was too rough and inconsistent for final audio mixing. I think capturing live audio would still be a great approach, but next time we’d want a more robust audio capturing setup. I’d also suggest getting the flock of starlings out of the airducts!
This would ultimately lead to us having to re-record all the dialogue from the film towards the end of the production, and only three of the original cast were part of that effort, which is why Betty Gabriel, Malcolm Mills, and Jordan Douglas Smith all resemble their animated roles.
In terms of editing the film, we created a full edit of the film as live action before beginning animation, Since each second of animation could take more a day to animate, the goal was to have the edit as tight as possible to avoid trimming any actually animated footage. This mostly held true, though there are a few partially-animated shots that were dropped eventually.
The long production time — seven years of animation — left us time to both revisit the edit, and to tighten some sequences for time — so it didn't end up being 10 years! I ultimately think the film is inarguably stronger from having so much time to revisit the edit over the years. Just living with the film and having it grow and age with you makes for a rare opportunity to revise.
From the edited feature, each shot was exported into still frames, at 12 frames per second. Using the clearest still from that shot, each of the characters visible were illustrated in the established art style of the film — something akin to the design approach of Ralph Bakshi’s rotoscoped work from the late ‘70s and early ‘80s — though I think it took on its own quality over time. That illustration was then a key frame that the animators would work from, mapping its details to the motion of the live-action actors, and embellishing that with further animated details. We had a lot of flexibility with the images because we were animating by hand — nothing was automated, so everything in every frame was as flexible as any other animation approach. Fire burns, cloaks billow in the wind, blood drips, limbs are severed, eyelines are adjusted, etc.
The linework animation team was a core group of three animators — Morgan, Brittain Scott and Ian Densford for the first couple of years, and then Morgan’s brother, Alexander King, for the remaining years. We had a few other animators come and go over the years, usually handling background characters. Finding people who could work in the style at the skill level we were demanding — to keep the film looking fairly uniform — proved to be more challenging than we were expecting, and much of the production time in the early years was spent training people how to achieve the look we wanted. They don’t really teach it in schools anymore, if they ever did.
The coloring was mostly done by two colorists, Morgan and Caleb Cliff. We colored with flat tones in regions bound by the linework, and then used an overlay of the background’s colors to add shadow and light and tone-matching to integrate those characters with the digitally-painted backgrounds — another stylistic approach rooted in Bakshi’s look.
We mostly tried to have the background paintings for each larger sequence done by the same artist, and they generally painted the elements in separate layers, so that the animated characters could move between the foreground and background elements, and so that things like clouds, smoke, and fire could be manipulated to move with the animation.
The linework animation and coloring was done in Adobe Photoshop, drawn with Wacom tablets, and the frames then assembled with the backgrounds and lighting layers in Abode Premiere. There’s a lot of other animation software used in the industry, but this pipeline seemed perfectly suited to the specific approach we were using, at least relative to where things were in 2014.
It was, perhaps, a brute-force method to accomplish our aesthetic goals — similar projects over the years have relied on a lot more automation, but it usually comes at the cost of losing the sense of the artist’s hand, and results in a smoother look that doesn’t capture the grittiness that one often associates with the classic fantasy and independent animation of the early 1980s. We suspect, though, that the tools to accomplish that look are on the horizon, if they aren’t already here, and that future efforts in this style won’t rely on an approach that requires so many man-hours of labor to accomplish.