|
PoseAI is a state-of-the-art muscle-based pose authoring tool that helps in creating animation. Using learnt inverse kinematics (IK) on muscles, PoseAI estimates body positions based on constraints applied to various joints, generating natural poses and movements.
Key Features:
- Pose IK: Complete full-body poses from partial inputs with PoseAI’s advanced IK system. Experience real-time performance with inference times of just a few milliseconds, allowing body pose estimation as you edit.
- Image to Pose: Transform 2D images into rig poses PoseAI’s Image to Pose feature. Achieve accurate pose replication within approximately 1-2 seconds.
- Preview Animations: Preview animation clips directly within the editor without needing to switch tabs. Navigate through clips using the playhead to view any part of the animation. Both the timeline and ragdoll preview are conveniently displayed within the editor.
- Comprehensive Editor Script: Manage all options in a single, well-organized editor window. The editor includes a built-in control panel for body constraints and IKs.
Dependencies:
PoseAI requires the Barracuda package from the package manager.
Limitations:
- While PoseAI uses AI to estimate body poses based on constraints or images, occasional inaccuracies may occur, such as incorrect rotations of body parts. Increasing the pose data set will enhance algorithm's accurancy. We are improving the AI and in the process of making it more accurate.
- Retargeting to any rig is supported through Mecanim, but rigs must be set to humanoid to utilize the Image to Pose feature.
PoseAI 是最先进的基于肌肉的姿势创作工具，有助于创作动画。PoseAI 利用学习到的肌肉逆运动学 (IK) 技术，根据应用于不同关节的约束条件估算身体位置，从而生成自然的姿势和动作。
主要功能
- 姿势 IK：利用 PoseAI 先进的 IK 系统，从部分输入完成全身姿势。体验实时性能，推理时间仅需几毫秒，可在编辑过程中估算身体姿势。
- 图像到姿势 将 2D 图像转换为钻机姿势 PoseAI 的图像到姿势功能。在大约 1-2 秒内实现精确的姿势复制。
- 预览动画 直接在编辑器中预览动画片段，无需切换选项卡。使用播放头浏览片段，查看动画的任何部分。时间轴和布偶预览都可在编辑器中方便地显示。
- 全面的编辑器脚本 在一个组织有序的编辑器窗口中管理所有选项。编辑器包含一个内置控制面板，用于控制身体约束和 IK。
依赖性：
PoseAI 需要使用软件包管理器中的 Barracuda 软件包。
局限性：
- 虽然 PoseAI 使用人工智能来根据约束或图像估算身体姿势，但偶尔可能会出现不准确的情况，例如身体部位的旋转不正确。增加姿势数据集将提高算法的准确性。我们正在改进人工智能，使其更加精确。
- 支持重定向到任何装备
|
