Template-free and Rig-free Animation Transfer using Kinetic Codes

Sanjeev Muralikrishnan, Niladri Shekhar Dutt, Niloy J. Mitra · 2025

Read Template-free and Rig-free Animation Transfer using Kinetic Codes by Sanjeev Muralikrishnan, Niladri Shekhar Dutt, Niloy J. Mitra (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-04831.txt.

Description

Sanjeev Muralikrishnan, Niladri Shekhar Dutt, and Niloy Mitra present SMF, in a 2025 preprint, a way to transfer motion onto new characters without templates or rigs. Animation retargeting turns sparse keypoint motion into a full moving mesh. Older methods need a template shape or an artist-built rig. They also jitter or fail on unseen bodies and moves. SMF learns from sparse motion alone, with no dataset-specific labels. Its core is Kinetic Codes, an autoencoder that packs motion into a rich latent code. Separate predictors handle spatial and temporal gradients of the deformation.

Tests span several public motion sets plus raw video from one camera. Characters varied widely in shape and mesh structure. On a leading benchmark, SMF set a new best for handling motion it never trained on. The latent code is credited for the calm, coherent output. Code and weights are public on the project page.

Catalog record

CREATORSanjeev Muralikrishnan, Niladri Shekhar Dutt, Niloy J. Mitra
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-04831

Browse the full library · About the Tower of Babel