Exploring the creative process behind a fire ritual concept piece.
Today we are exploring the theme of Rituals as we continue our multi-part series of dramatic performance and motion capture with NIDA & Electric Lens Co. In this article we delve deeper into the process behind the creation.
The aim was to deliver an experimental, short-form video piece using motion capture which explored ideas of:
Prior to the capture session, a few creative jams were arranged and the concept was formed:
A central performer is trapped, surrounded by a mysterious group of dancers who pulsate back & forth, suggesting a ritual. The people are invisible and only their long, flowing clothing can be seen - witches flocks or pagan gowns. The surrounding dancers edge closer until they breach the protagonist’s personal space providing a sense of overwhelming claustrophobia. Build to the climax where the protagonist demands space and the others span out and disappear. The performer is left feeling alone, exposed and abandoned.
Not only was this a mesmerizing performance, it was also a great opportunity to explore a variety of technical techniques which will be discussed below:
Duplicated motion capture for cult
Cloth simulation on transparent character
Take a look at the final video and then read on to see what was involved:
Once the concept had been formed, Dreamr's creative team produced a series of moodboards to articulate the tone and style of the final piece. This was then provided to ELC's VFX team so they could start testing and experimenting with cloth and fire sims which would be used in the final production.
The moodboards and early test sims were used by the performers to understand the physical attributes of their virtual costume so they could base their movement around the flow and behaviour of the fabric.
From here Dreamr created a rough blocking to understand the positioning of the cult compared to the protagonist. We knew that we wanted to explore the technique of using one performer to control the entire ritual dancers. When a single mocap performance is applied to a line of multiple characters, the movement can come across as stale and unnatural. This is due to the most subtle of movements being in sync and losing the organicness. However, we wanted to see if this issue still occurred when the group is positioned in a circle.
After doing a quick mock-up with a mannequin in Unreal our expectations were confirmed - the cloning affects were far more subtle in a circle than in a line. This is mainly due to no too characters being aligned to the camera in the same way. It actually gave an eerie similarity without it looking too artificial and we were able to leverage this to create a kaleidoscopic feel when we cut a top-down shot of the dance.
During this time the NIDA team worked on the choreography, exploring ritualistic movements whilst taking the learnings from the cloth sims and mocap tests. The final movements were not locked down, allowing the performer to use free-form expression throughout the capture. Since there was only one performer being captured for the cult, the movements would look synchronized regardless.
It was time for the capture. This was done in a single session as two unbroken takes in NIDA's performance space. The team used an xSens motion capture system with a reference audio beat track on loop in the background and two witness cameras that could be referenced afterwards.
The capture was then processed and brought into UE5 where camera angles were decided using the simple mannequin scene.
Once this was locked down, we worked with Matt Hermans from Electric Lens Co who simulated the different effects in Houdini (cloth sim, fire sim) and produced the final visuals. This produced some amazing and hypnotic results and, if time permitted, we could have spent weeks tinkering with the inputs for this.
Once the piece was rendered out, it was packaged up and delivered back as a high res video to NIDA for circulation.
As part of a retrospective, Dreamr's director, Pete Short, shared a few observations:
It was amazing what could be achieved in such a short period of time. However, in an ideal world, pieces like this would benefit from additional rounds of real-time cloth sim with the actor's participation. This would allow the actor to truly understand the physics of their costume and refine their performance through trial and error.
Upon reflection, there was actually something really appealing about the simple mannequin test version. The blank canvas stage, abstract characters and single point light made for a clean, dramatic piece and kept the focus on the body language. It is a lesson that not every piece needs visual effects for the best results.
The circular formation for the synchronized dancers worked really well and is something we would like to explore further in another piece.
If you would like to know more about Dreamr's performance capture offering, please visit: Dreamr Performance Capture