Retargeting motion across characters with varying body shapes while preserving interaction semantics, such as self-contact and near-body proximity, remains a challenging problem. While recent geometry-aware approaches address this by maintaining spatial relationships between predefined corresponding regions, their reliance on static correspondences often struggles when the target character exhibits exaggerated body proportions. In this paper, we present a geometry-aware motion retargeting framework that preserves interaction semantics by performing proximity matching over spatially adaptive anchors. Unlike prior methods with static anchor definitions, the proposed method dynamically repositions anchors to reachable regions on the target character. This is achieved via a Transformer-based anchor refinement strategy that predicts anchor displacements and constrains the translated anchors to remain on the target character geometry through differentiable soft projection. By incorporating pose-dependent spatial structures from the source character, the adapted anchors provide structurally coherent guidance for interaction-aware retargeting. Conditioned on these anchors, a graph-based autoencoder predicts target skeletal motion that preserves the spatial configuration of the source. To encourage task-aligned optimization between anchor adaptation and motion retargeting, we adopt an alternating training scheme in which each module is optimized in turn. Through extensive evaluations, we demonstrate that our method outperforms state-of-the-art approaches in preserving interaction fidelity across diverse character geometries. Code is available at Project Page .

Text-driven motion generation has advanced significantly with the rise of denoising diffusion models. However, previous methods often oversimplify representations for the skeletal joints, temporal frames, and textual words, limiting their ability to fully capture the information within each modality and their interactions. Moreover, when using pre-trained models for downstream tasks, such as editing, they typically require additional efforts, including manual interventions, optimization, or fine-tuning. In this paper, we introduce a skeleton-aware latent diffusion (SALAD), a model that explicitly captures the intricate inter-relationships between joints, frames, and words. Furthermore, by leveraging cross-attention maps produced during the generation process, we enable attention-based zero-shot text-driven motion editing using a pre-trained SALAD model, requiring no additional user input beyond text prompts. Our approach significantly outperforms previous methods in terms of text-motion alignment without compromising generation quality, and demonstrates practical versatility by providing diverse editing capabilities beyond generation. Code is available at project page.

Despite recent advancements in learning-based motion in-betweening, a key limitation has been overlooked: the requirement for character-specific datasets. In this work, we introduce AnyMoLe, a novel method that addresses this limitation by leveraging video diffusion models to generate motion in-between frames for arbitrary characters without external data. Our approach employs a two-stage frame generation process to enhance contextual understanding. Furthermore, to bridge the domain gap between real-world and rendered character animations, we introduce ICAdapt, a fine-tuning technique for video diffusion models. Additionally, we propose a "motion-video mimicking" optimization technique, enabling seamless motion generation for characters with arbitrary joint structures using 2D and 3D-aware features. AnyMoLe significantly reduces data dependency while generating smooth and realistic transitions, making it applicable to a wide range of motion in-betweening tasks. The code and videos are available at project page.

Abstract Despite the growing accessibility of skeletal motion data, integrating it for animating character meshes remains challenging due to diverse configurations of both skeletons and meshes. Specifically, the body scale and bone lengths of the skeleton should be adjusted in accordance with the size and proportions of the mesh, ensuring that all joints are accurately positioned within the character mesh. Furthermore, defining skinning weights is complicated by variations in skeletal configurations, such as the number of joints and their hierarchy, as well as differences in mesh configurations, including their connectivity and shapes. While existing approaches have made efforts to automate this process, they hardly address the variations in both skeletal and mesh configurations. In this paper, we present a novel method for the automatic rigging and skinning of character meshes using skeletal motion data, accommodating arbitrary configurations of both meshes and skeletons. The proposed method predicts the optimal skeleton aligned with the size and proportion of the mesh as well as defines skinning weights for various meshskeleton configurations, without requiring explicit supervision tailored to each of them. By incorporating Diffusion 3D Features (Diff3F) as semantic descriptors of character meshes, our method achieves robust generalization across different configurations. To assess the performance ofour method in comparison to existing approaches, we conducted comprehensive evaluations encompassing both quantitative and qualitative analyses, specifically examining the predicted skeletons, skinning weights, and deformation quality.

Abstract Motion in‐betweening has emerged as a promising approach to enhance the efficiency of motion creation due to its flexibility and time performance. However, previous in‐betweening methods are limited to generating short transitions due to growing pose ambiguity when the number of missing frames increases. This length‐related constraint makes the optimization hard and it further causes another constraint on the target pose, limiting the degrees of freedom for artists to use. In this paper, we introduce a keyframe‐driven approach that effectively solves the pose ambiguity problem, allowing robust in‐betweening performance on various lengths of missing frames. To incorporate keyframe‐driven motion synthesis, we introduce a keyframe score that measures the likelihood of a frame being used as a keyframe as well as an adaptive keyframe selection method that maintains appropriate temporal distances between resulting keyframes. Additionally, we employ phase manifolds to further resolve the pose ambiguity and incorporate trajectory conditions to guide the approximate movement of the character. Comprehensive evaluations, encompassing both quantitative and qualitative analyses, were conducted to compare our method with state‐of‐the‐art in‐betweening approaches across various transition lengths. The code for the paper is available at https://github.com/seokhyeonhong/long-mib

Interactive motion between multiple characters is widely utilized in games and movies. However, the method for generating interactive motions considering the character's diverse mesh shape has yet to be studied. We propose a Spatio Cooperative Transformer (SCT) to retarget the interacting motions of two characters having arbitrary mesh connectivity. SCT predicts the residual of root position and joint rotations considering the shape difference between the source and target of interacting characters. In addition, we introduce an anchor loss function for SCT to maintain the geometric distance between the interacting characters when they are retargeted. We also propose a motion augmentation method with deformation-based adaptation to prepare a source-target paired dataset with an identical mesh connectivity for training. In experiments, our method achieved higher accuracy for semantic preservation and produced less artifacts of inter-penetration between the interacting characters for unseen characters and motions than the baselines. Moreover, we conducted a user evaluation using characters with various shapes, spanning low-to-high interaction levels to prove better semantic preservation of our method compared to previous studies.

Abstract Stitching different character motions is one of the most commonly used techniques as it allows the user to make new animations that fit one's purpose from pieces of motion. However, current motion stitching methods often produce unnatural motion with foot sliding artefacts, depending on the performance of the interpolation. In this paper, we propose a novel motion stitching technique based on a recurrent motion refiner (RMR) that connects discontinuous locomotions into a single natural locomotion. Our model receives different locomotions as input, in which the root of the last pose of the previous motion and that of the first pose of the next motion are aligned. During runtime, the model slides through the sequence, editing frames window by window to output a smoothly connected animation. Our model consists of a two‐layer recurrent network that comes between a simple encoder and decoder. To train this network, we created a sufficient number of paired data with a newly designed data generation. This process employs a K‐nearest neighbour search that explores a predefined motion database to create the corresponding input to the ground truth. Once trained, the suggested model can connect various lengths of locomotion sequences into a single natural locomotion.