cascadeur-use-of-ai-tools
- Home
- cascadeur-use-of-ai-tools
This document is intended to provide insight into Cascadeur software’s use of tools (hereinafter “AI tools”) that function using machine-learning algorithms.
1. What AI and non-AI tools does Cascadeur have?
Cascadeir’s AI tools include AutoPosing, Inbetweening, AutoInterpolation, and the Video Mocap tool.
Non-AI tools include AutoPhysics, Ragdoll, Animation Unbaking, and Fulcrum Motion Cleaning. They are not considered AI tools because they don’t use machine-learning algorithms.
2. What do Cascadeur’s AI tools do? How do they work?
AutoPosing - a posing tool that uses machine-learning algorithms to predict the pose the user is trying to make.
Trained on a database of various poses, it checks which AutoPosing manipulators were activated by the user and where they are located. The activated controllers are used as “guides” to predict the pose the user is trying to build.
Inbetweening - an animation tool that uses machine-learning algorithms to produce animations between the keyframed poses.
Trained on a database of various animations, it uses several factors to predict the right animation: character poses on the keyframes, the keyframes’ timing and quantity, and the animation style (e.g., walking, running, jumping, etc.).
AutoInterpolation - an animation tool that uses machine-learning algorithms to predict and apply different interpolation and kinematics types between the keyframes, based on their timing, quantity, and stored character poses.
Video Mocap tool - an animation tool that uses machine-learning algorithms to extract poses of the performing actors from the imported video.
It uses computer vision algorithms to detect the pose the performing actor is making and then applies it to the rigged character model.
3. What data were Cascadeur’s AI tools trained on?
Cascadeur’s AutoPosing, AutoInterpolation, and Inbetweening tools were mainly trained on Nekki's proprietary animation data. The animation data consists of dozens of hours of handkey and motion capture animations produced for Nekki’s games - the Shadow Fight series, and SPINE. To supplement some of the missing pose and locomotion data, we also recorded motion capture animations specifically for Cascadeur and purchased third-party animation packs certified for AI use.
Cascadeur’s Video Mocap tool uses open-source libraries, with the added post-processing tools to improve the output quality.
4. Do Cascadeur’s AI tools gather and use my animations as training data?
No, Cascadeur doesn’t collect the animations you produce and/or import to use as a training data. Cascadeur’s AI tools are trained on Nekki’s animation data and legally purchased third-party animation data certified for AI use.
5. Do Cascadeur’s AI tools work on the local machine or the external servers?
Cascadeur’s tools work on your local machine, without needing to connect to external servers.
The Mocap tool isn’t included in the installation by default due to its large size, so it has to be downloaded from our servers. After the download, the Mocap tool uses your machine’s resources for pose calculations.
6. What goal do we keep in mind while developing our AI tools? Do we want to replace animators?
Cascadeur’s main mission is to make the 3D animation process easier for everyone, from complete strangers to the art of animation and novice animators, to experienced professionals.
This mission is carried out through several key factors:
1. Cascadeur improves the animation process through the absence of gimbal lock, the automatic IK-FK blending during interpolation, and the lack of separate IK and FK controls in the control rig.
2. Cascadeur introduces tools that streamline the different parts of the 3d animation pipeline, such as AutoPosing for pose drafting, Animation Unbaking and AutoInterpolation for splining, AutoPhysics for polishing, and more.
3. Cascadeur’s AI tools assist the 3D animation process but do not entirely replace it. The animator is still in control from start to finish and is very welcome to continue improving and polishing their animation.
Many people get discouraged when interacting with the traditional 3D animation tools, unable to produce even a barely passable animation. This stems from the technical limitations, problems and quirks of the established 3D animation tools.
Cascadeur seeks to challenge these limitations by reinventing the animation pipeline, providing the tools to make it quicker, easier and better for everyone.
1. What AI and non-AI tools does Cascadeur have?
Cascadeir’s AI tools include AutoPosing, Inbetweening, AutoInterpolation, and the Video Mocap tool.
Non-AI tools include AutoPhysics, Ragdoll, Animation Unbaking, and Fulcrum Motion Cleaning. They are not considered AI tools because they don’t use machine-learning algorithms.
2. What do Cascadeur’s AI tools do? How do they work?
AutoPosing - a posing tool that uses machine-learning algorithms to predict the pose the user is trying to make.
Trained on a database of various poses, it checks which AutoPosing manipulators were activated by the user and where they are located. The activated controllers are used as “guides” to predict the pose the user is trying to build.
Inbetweening - an animation tool that uses machine-learning algorithms to produce animations between the keyframed poses.
Trained on a database of various animations, it uses several factors to predict the right animation: character poses on the keyframes, the keyframes’ timing and quantity, and the animation style (e.g., walking, running, jumping, etc.).
AutoInterpolation - an animation tool that uses machine-learning algorithms to predict and apply different interpolation and kinematics types between the keyframes, based on their timing, quantity, and stored character poses.
Video Mocap tool - an animation tool that uses machine-learning algorithms to extract poses of the performing actors from the imported video.
It uses computer vision algorithms to detect the pose the performing actor is making and then applies it to the rigged character model.
3. What data were Cascadeur’s AI tools trained on?
Cascadeur’s AutoPosing, AutoInterpolation, and Inbetweening tools were mainly trained on Nekki's proprietary animation data. The animation data consists of dozens of hours of handkey and motion capture animations produced for Nekki’s games - the Shadow Fight series, and SPINE. To supplement some of the missing pose and locomotion data, we also recorded motion capture animations specifically for Cascadeur and purchased third-party animation packs certified for AI use.
Cascadeur’s Video Mocap tool uses open-source libraries, with the added post-processing tools to improve the output quality.
4. Do Cascadeur’s AI tools gather and use my animations as training data?
No, Cascadeur doesn’t collect the animations you produce and/or import to use as a training data. Cascadeur’s AI tools are trained on Nekki’s animation data and legally purchased third-party animation data certified for AI use.
5. Do Cascadeur’s AI tools work on the local machine or the external servers?
Cascadeur’s tools work on your local machine, without needing to connect to external servers.
The Mocap tool isn’t included in the installation by default due to its large size, so it has to be downloaded from our servers. After the download, the Mocap tool uses your machine’s resources for pose calculations.
6. What goal do we keep in mind while developing our AI tools? Do we want to replace animators?
Cascadeur’s main mission is to make the 3D animation process easier for everyone, from complete strangers to the art of animation and novice animators, to experienced professionals.
This mission is carried out through several key factors:
1. Cascadeur improves the animation process through the absence of gimbal lock, the automatic IK-FK blending during interpolation, and the lack of separate IK and FK controls in the control rig.
2. Cascadeur introduces tools that streamline the different parts of the 3d animation pipeline, such as AutoPosing for pose drafting, Animation Unbaking and AutoInterpolation for splining, AutoPhysics for polishing, and more.
3. Cascadeur’s AI tools assist the 3D animation process but do not entirely replace it. The animator is still in control from start to finish and is very welcome to continue improving and polishing their animation.
Many people get discouraged when interacting with the traditional 3D animation tools, unable to produce even a barely passable animation. This stems from the technical limitations, problems and quirks of the established 3D animation tools.
Cascadeur seeks to challenge these limitations by reinventing the animation pipeline, providing the tools to make it quicker, easier and better for everyone.