ComfyUI Node: UniRig: Extract Skeleton

UniRig: Extract Skeleton:

The UniRigExtractSkeletonNew node is designed to extract a semantic skeleton from a 3D mesh using a machine learning-based approach. This node is part of the UniRig system, which leverages advanced algorithms to deliver high-quality skeleton extraction suitable for a variety of mesh types, including humans, animals, and objects. The node operates within an isolated environment that supports GPU dependencies, ensuring efficient processing. It requires a pre-loaded model from the UniRigLoadSkeletonModel node to function. The node is particularly beneficial for artists and developers looking to automate the rigging process, as it simplifies the extraction of skeletons from complex meshes, making it easier to animate and manipulate 3D models.

UniRig: Extract Skeleton Input Parameters:

skeleton_template

The skeleton_template parameter specifies the template to be used for skeleton extraction. It determines the naming convention and structure of the skeleton. Options include "mixamo" and "smpl", which are popular skeleton formats. If "mixamo" is selected, the node uses a VRoid-based extraction method and remaps the names to Mixamo's format. Similarly, if "smpl" is chosen, the node extracts using VRoid and converts the skeleton to the SMPL format, which includes filtering to 22 SMPL joints. This parameter is crucial for ensuring compatibility with different animation systems and workflows.

skeleton_model

The skeleton_model parameter is a required input that provides the pre-trained model necessary for skeleton extraction. This model must be loaded using the UniRigLoadSkeletonModel node and should include a valid checkpoint path. The model's configuration, including data type and attention backend, influences the extraction process. Without a valid model, the node cannot perform skeleton extraction, making this parameter essential for the node's operation.

UniRig: Extract Skeleton Output Parameters:

direct_skeleton_result

The direct_skeleton_result output contains the extracted skeleton data, including joint positions and connectivity information. This result is crucial for further processing, such as animation or rigging, as it provides the foundational structure needed to manipulate the 3D mesh. The output is generated through direct prediction from the mesh vertices and faces, ensuring that the skeleton accurately represents the underlying geometry.

norm_params

The norm_params output provides normalization parameters used during the skeleton extraction process. These parameters are important for understanding how the input mesh was processed and can be used to reverse normalization if needed. They ensure that the extracted skeleton maintains the correct proportions and alignment relative to the original mesh.

UniRig: Extract Skeleton Usage Tips:

• Ensure that the skeleton_model is correctly loaded and connected to the node to avoid runtime errors. The model should have a valid checkpoint path.
• Choose the appropriate skeleton_template based on your target animation system. Use "mixamo" for compatibility with Mixamo animations and "smpl" for SMPL-based systems.

UniRig: Extract Skeleton Common Errors and Solutions:

"skeleton_model is required for UniRig: Extract Skeleton. Please connect a UniRigLoadSkeletonModel node."

• Explanation: This error occurs when the node is executed without a connected skeleton model.
• Solution: Load a skeleton model using the UniRigLoadSkeletonModel node and ensure it is connected to the UniRigExtractSkeletonNew node.

"Skeleton checkpoint not found: <checkpoint_path>"

• Explanation: The specified checkpoint path for the skeleton model does not exist, preventing the node from accessing the necessary model data.
• Solution: Verify that the checkpoint path is correct and that the file exists. Ensure the model is properly loaded with the correct path.

"Skeleton prediction failed - no joints generated"

• Explanation: The node was unable to generate joints from the input mesh, possibly due to an issue with the mesh data or model configuration.
• Solution: Check the input mesh for errors or inconsistencies. Ensure the model configuration is correct and compatible with the mesh type.