Research map / Video / Video generation
Length-Aware Motion Synthesis via Latent Diffusion
Alessio Sampieri, Alessio Palma, Indro Spinelli, Fabio Galasso
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use length-aware variable-size latent instead of fixed-dimension latent (ablation) | LADiff · HumanML3D | R-Precision Top-3: 0.776 -> 0.792 | Improved |
| use length-aware latent vs fixed latent on MM-Dist (ablation) | LADiff · HumanML3D | MM-Dist: 3.137 -> 3.054 | Improved |
| use length-aware latent vs fixed latent on FID (ablation) | LADiff · HumanML3D | FID: 0.179 -> 0.182 | No clear effect |
| choose r=48 frames per latent subspace over r=16 or 64 (ablation) | LADiff · HumanML3D | R-Precision Top-3: 0.778 (r=16) / 0.780 (r=64) -> 0.792 | Improved |
| activate all latent subspaces for every sequence (r=all) (ablation) | LADiff · HumanML3D | FID: 0.182 -> 0.445 | Hurt |
| add denoising VAE input noise of 33% (ablation) | LADiff · HumanML3D | FID: 0.222 -> 0.182 | Improved |
| apply DVAE noise on latent instead of input (ablation) | LADiff · HumanML3D | R-Precision Top-3: 0.792 -> 0.735 | Hurt |
| increase DVAE input noise from 33% to 50% (ablation) | LADiff · HumanML3D | R-Precision Top-3: 0.792 -> 0.787 | Hurt |
| add retrieval augmentation to LADiff | LADiff · HumanML3D | FID: 0.182 -> 0.054 | Improved |
| LADiff vs ReMoDiffuse retrieval-augmented on KIT-ML | LADiff · KIT-ML | FID: 0.155 -> 0.386 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- AttT2M — AttT2M: Text-Driven Human Motion Generation with Multi-Perspective Attention Mechanism (ICCV 2023) · better on 4 of 5 shared results
- ReMoDiffuse — ReMoDiffuse: Retrieval-Augmented Motion Diffusion Model (ICCV 2023) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Seamless Human Motion Composition with Blended Positional Encodings (CVPR 2024 · similarity 0.87)
- Rethinking Diffusion for Text-Driven Human Motion Generation: Redundant Representations, Evaluation, and Masked Autoregression (CVPR 2025 · similarity 0.87)
- AMD: Autoregressive Motion Diffusion (AAAI 2024 · similarity 0.87)
- Human Motion Diffusion as a Generative Prior (ICLR 2024 · similarity 0.87)
- Optimizing Diffusion Noise Can Serve As Universal Motion Priors (CVPR 2024 · similarity 0.87)
- Towards Detailed Text-to-Motion Synthesis via Basic-to-Advanced Hierarchical Diffusion Model (AAAI 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
