NOTE: EVERY ASPECT OF THIS BLOG POST IS SIGNIFICANT TO UNDERSTANDING THE TECHNICAL AND ARTISTIC CHALLENGES I FACED.
Like with my creative intentions from day one in previous blog posts, EVERY FRAME HAD TO BE METICULOUSLY PLANNED CORRECTLY. The reason for this was that if one frame of offset on the timings, it would evidently noticeable with the style we are going for.
What you see below is a demonstration of what I mean by planning every frame. One might assume it would be more efficient to just create a crop transition in post, but in fact, from experience, the reality was that it was more practical (are artistically appropriate) to do it all with Maya.
The transition between scenes was literally ONE CAMERA moving between each individually constructed scene imported into a render set up. What I did was set up the camera in a way that was identicial to the animatic and have the camera move in a motion that would allow me to animate the visibility of various grouped assets to activate according each frame in a transition.
In fact, I came across this video essay exploring Brad Bird’s cinematic style and it actually made me think a bit more carefully about the flow of the camera before I set up the scene!
As you see below in this final render (bar rough fixes), as the camera tracks through the wall, I am gradually deactivating the hallway scene in groups, and activating the visibility on the country road, using the wall as a proximity meter to hide abrupt appearances (such as the very subtle ground that pops in for one frame in the first shot below).
But to also make the transition more smooth, there is actually an extremely subtle camera rotation I added that helps the view read the perspective without it appearing too jarring. What I noticed in the first render test of the camera, was that doing a static left to right camera track make the scene feel flat and made it actually rather weird to watch. As a result, the rotation is to just highlight the depth of the images, while I used a slight tilt to accentuate the 3D environments, thus making for a more visually striking look.
What was effectively key for this render setup to work was that the animation worked in motion and synergy with the camera. Watching the animatic and then seeing the flow of the animations, I was really looking for an offset to mirror the line art style and also to play with the energy of Zara’s character and the narration of Attracta.
As a result, as the dog moves, the camera chases, so once the dog runs out of the bedroom, I deactivate, thus triggering the country scene, then to make the final transition to the conservatory work, I actually animated the scenes movement to line up with the camera, thus making it both smooth and timed accordingly to the appropriate pace of the scene.
The several sleepless nights to get that to work is evident in both the final playblast film:
Moments such as transition to adjust Zara and then to the country road Zara and finally conservatory Zara have a very beautiful flow that I attribute to both a very orcastrated technical process and luck. The timing makes both Attracta’s narration and the animation read without complexity or confusion, thus successfully working with an audience.
With that technical and artistic obstacle proving successful in the final presentation, there was a significantly major challenge in getting the animation itself to work. Other than scaling and translation issues, loading the references caused Maya to break several of the rigs and animations in the process. I found that having multiple animations trigger at different times caused so many breaks that I thought baking the animation would be an effective solution…. It wasn’t… and I isolated the reason to the fact Tyrone had animated locators to make the dog (really smart animation tactic) and I had separate meshes for the mouth and eyes.
The more and more I tried to find a solution, I decided to go back to last semester’s notes on importing animation to Cinema 4D and I thought…. Couldn’t I just export the animations as an Alembic cache, as long as I time when the animations were due to begin?
I followed this tutorial, re-read last semester’s notes and I was successfully able to overcome the technical obstacle and get the animations to load coherently into the scene.
Overall, my technical and artistic challenges came down to this:
- get the camera to work in one go by animating the visibility of assets
- get the perspective readable by adjusting the cameras orientation
- export animations as Alembics to avoid referencing issues
- check EACH FRAME to ensure the visibilities are consistent and coherent
As a result of this careful planning and research at the beginning of the project and the extremely skillful handling of the animatic, this process was significantly more manageable because of this preparation. I was successfully able to prepare for obstacles and even overcome ones I didn’t predict, thus getting the render setup completed within 4 days, giving us one additional day before final presentation.