Towards New AI-based Animation
Historically, 3D animation has been a time-consuming and costly process involving many preparatory steps. To make their characters ready to move, animators have to develop properly sculptured 3D meshes appropriate for skin weighting, rigging, and deformation. Beyond that, they have to define anatomy and movement variables set either manually (key framing) or retrieved from the motion capture using a live action footage. Thereafter, animators take a role of Frankenstein creating a frame and skeleton for their characters. This process, known as rigging, is one of the most tedious and complex parts of the animator’s work that requires precision and consideration of many details.
Many animation tools, both online and desktop, have been developed to simplify the 3D animation workflow, beginning from 3D modeling to animation itself. Recent advances in Machine Learning (ML) have made it easier to create rigging and animations in a more automated way. Animation tools pre-trained with deep learning save time on painting weights and defining motion variables.
When it comes to rigging, Mixamo, a company bought by Adobe Systems in 2015, offers a popular Auto-Rigging tool that allows the users to upload their 3D models, place some marks on them, and in one click get back a 3D mesh rig with calculated skinning weights. The service that takes just 2 minutes to complete the task, makes 90 percent of what animators typically expect from the rigging. After the work is done, animators may tweak the download adding extra bones and adjusting some weight placements.
Similarly, machine learning powers face animations in Mixamo’s Face Plus service. Released in 2013, Face Plus allows users to record facial animation data of themselves via standard webcam and, then, apply the animation to their characters inside the Unity game engine.
Contemporary 3D characters are not simple geometric figures, but complex and realistic protagonists with a wide variety of unique features, such as hair, clothing, and facial expressions. Adobe Fuse 3D character creator software allows users to create their unique 3D characters by assembling customizable body parts, clothes, and textures together. The system supports over 280 customizable attributes that account for a wide variety of game design use cases and goals. Even better, with Mixamo’s 3D characters repository you may select any character that suits you, apply animations and be on your way with the fully functional 3D personage. If you want some VFX (Visual Effects), modern animation tools like Maya and Blender support simulation of fluids, particles, cloth, clouds, smoke, and other physical processes and objects. With such tools at hand, designers and animators have unlimited possibilities in story weaving and VFX (Visual Effects).
Recent advances in deep learning and artificial generative models promise to expand existing capabilities of animation tools even further. Norah AI, a revolutionary animation tool recently released by Absentia startup, moves animation and game design industry into the AI era. It enables quick creation of all game elements, including 3D models and animations, game geometry, story weaving, and texturing, all with the minimal human input. The first version of Norah AI features Auto Rig tool, human motion simulation and blending, and Motion Editor that cover a wide variety of user cases in 3D animation and game design.
Under the hood, Norah AI uses the state-of-the-art deep learning and generative models enabling her to create complex animations, which are realistic, dynamic, and flexible. It takes time and training to become such an animation guru. For that purpose, Norah was trained on half a million of animation frames. Being an expert in all conceivable types of movements, Norah also excels in complex ones, such as dancing and combat animation. This knowledge is not just a machine memory, but machine imagination.
Since Norah learns generic patterns, animators have an option of blending and merging existing animations into the new ones using Norah’s Style Transfer tool. As a result, Norah’s technology opens up infinite opportunities for experimentation and creation of an endless repository of classy animations suitable for any purposes.
Norah’s AutoRig tool is another example of this high-end functionality. Nowadays, even the most sophisticated tools require the user’s input to produce usable rigs and skin weights. Norah’s AutoRig is different. The system creates usable rigs in the fraction of seconds without any human input. No skin weighting and placing bone markers are needed. This is an excellent illustration of the AI-first approach to animation.
Norah’s Motion Editor is equally powerful. The system allows users to make infinite number of changes, add smooth transitions along the time spectrum. This brings fine tuning control with its frame by frame editing option which links the futuristic AI driven approach to old school methods of creating animations .
