Groundbreaker

Groundbreaker was a graduation project as a student of Multimedia. This time, I wanted to create something representing the industry standard of 3D animation. My love for animals was another reason for me to pursue this theme. The narrative follows the personification of pride and envy as a character. The visual inspiration and character development came from Disney's Zootopia. From writing the script to rendering and editing the final output, it was a studio-like experience to create a short film with modern techniques. The set was modelled and optimized for smooth animation and rendering. The pre and post-production followed a production pipeline used by studios. Since this was a 3D film it had many other layers of production like modeling, rigging, texturing, shading, lighting, animation, and rendering. All these processes were hardware and software-intensive but ultimately, the final product was satisfying.

Process and Technique

Pre-production and modelling

The initial requirement was a script. I wanted to depict the age-old concepts of pride and envy in a modernized and friendly way. After screenwriting this story, I developed character outfits and expressions through sketching and 2d vector software. It was important to create an expression sheet as most of the story revolves around the opposite personalities of the characters and therefore, their expressions. After the finalization of the characters and their appearance, the 3D production process was started. I used Autodesk Maya to model and customize the two characters and their props (shovels, sunglasses, etc.) I had to keep the models within polygon limits as this wasn't a visualization project and would be used in full-fledged animation scenes. I rigged the character with armatures with switchable inverse kinematics and forward kinematics.

The Rig

Even though every project involves technical solutions to some degree, one such occasion occurred during this project when I had to tackle a gimbal lock situation during the rigging stage.

The rigging phase is tricky as it requires an intricate understanding of the three-dimensional skeletal structuring of an armature that makes it animatable. With a manual rig, each bone and joint is carefully placed as per the character’s physiological feature — in this case, it was a biped hybrid rabbit. The rig (skeleton) interacts with the mesh of the model and allows us to animate the mesh in desired ways. Each joint has a weightage assigned to it; it determines the degree of influence a joint has on the mesh in the targeted area. During the weight painting, all joints must have the right amount of influence for the animation to be smooth. The trickiest part, however, was the switch between forward and inverse kinematics; it allowed the character to perform complex tasks. The problem first occurred after the rig was complete, I switched the character from forward to inverse kinematics resulting in a torn mesh, and the clavicle joint went into a gimbal lock; rendering the character unusable. This had occurred due to a dynamic morphing issue; the design had modular/interchangeable parts that had caused a misalignment between the IK setup and the mesh. It also occurred because the IK system relying on Euler angles caused an overlap of rotational axes. The solution had to be a step-by-step fix of the rig:

Adjusting weightage: The first step involved in solving this was removing any and every unintended influence from the mesh. I implemented a smart paintbrush tool that identifies overlapping vertices between parts and distributes weights evenly. A weight-based, adaptive, node blending system was also put into action.

Implementing rotational control system: The clavicle is locked unnaturally when reaching extreme positions, limiting the rotational freedom. With a rotational control system, a gimbal lock could be avoided. However, it wasn’t as reliable in terms of precise linear motions and rotations were disruptive. Using quaternion rotations would be the last resort as it would make the rig more complex and lead to rendering issues. The rotational control system did not work as expected and I had to switch to quaternion rotations.

Debugging the IK chain: This turned out to be the most important step. By replacing the Euler angle-based IK solver with a quaternion one, the gimbal lock was eliminated. With simplified interpolations between rotated coordinates and the representation of rotations as 4D vectors instead of 3D angles the rotations were smooth and uninterrupted.

Pole vector adjustments: The pole targets were corrected and re-aligned to avoid any overlaps. The IK constraint settings were then adjusted to reorient the pole angle parameter in order to avoid any twisting issues that might trigger a gimbal lock on the joint.

Rig optimization: The dual hybrid IK/FK system allowed seamless switching, but also made the rig more complex. For this complex anatomy, I used an auto-rig plugin to analyze and optimize the rig with auto-detection; it ensured optimal bone placement in real time. The rig automatically adapted to different forms and movements without manual intervention. I also decided to add procedural controllers to enable dynamic adjustment of pole target placements.

Testing and refinement: This stage involved rigorous testing of the rig in extreme positions to verify and identify any rigging errors. To further refine the rig, more weight painting was adjusted along with specific pole vector joint adjustments that required more attention e.g. shoulder, elbows, etc.

Resolution :

1. Fluid and natural animations without any jitters or shakes became possible. The rotational limitations had smoother animations due to the equipping of the rig with an anticipatory procedural pole vector system that analyzed and optimized the rig in real-time.

2. The Gimbal lock was eliminated with the help of quaternion rotations. The joint issue was resolved, allowing free movements in any direction.

3. The hybrid solution involving mathematics, machine learning, and a procedural system was used to solve an IK issue in an innovative and personalized way.

4. The rig was now very well adapted to more complex settings and motions. It had now acquired scalability and flexibility for hybrid-complex creatures where Euler-based systems would fail.

The experience taught me how to deal with technical difficulties within a time constraint. It not only solved the gimbal lock issue at hand but also set me on a path of research and discovery to come up with an innovative solution to the problem. It allowed me to explore a 3D rigging solution with a modern outlook in 3D animation workflow.