Pretrained Diffusion Models for Unified Human Motion Synthesis

12/06/2022
by   Jianxin Ma, et al.
0

Generative modeling of human motion has broad applications in computer animation, virtual reality, and robotics. Conventional approaches develop separate models for different motion synthesis tasks, and typically use a model of a small size to avoid overfitting the scarce data available in each setting. It remains an open question whether developing a single unified model is feasible, which may 1) benefit the acquirement of novel skills by combining skills learned from multiple tasks, and 2) help in increasing the model capacity without overfitting by combining multiple data sources. Unification is challenging because 1) it involves diverse control signals as well as targets of varying granularity, and 2) motion datasets may use different skeletons and default poses. In this paper, we present MoFusion, a framework for unified motion synthesis. MoFusion employs a Transformer backbone to ease the inclusion of diverse control signals via cross attention, and pretrains the backbone as a diffusion model to support multi-granularity synthesis ranging from motion completion of a body part to whole-body motion generation. It uses a learnable adapter to accommodate the differences between the default skeletons used by the pretraining and the fine-tuning data. Empirical results show that pretraining is vital for scaling the model size without overfitting, and demonstrate MoFusion's potential in various tasks, e.g., text-to-motion, motion completion, and zero-shot mixing of multiple control signals. Project page: <https://ofa-sys.github.io/MoFusion/>.

READ FULL TEXT

page 7

page 8

research
06/01/2023

Controllable Motion Diffusion Model

Generating realistic and controllable motions for virtual characters is ...
research
03/02/2023

Human Motion Diffusion as a Generative Prior

In recent months, we witness a leap forward as denoising diffusion model...
research
09/29/2022

Human Motion Diffusion Model

Natural and expressive human motion generation is the holy grail of comp...
research
09/06/2023

MCM: Multi-condition Motion Synthesis Framework for Multi-scenario

The objective of the multi-condition human motion synthesis task is to i...
research
04/12/2023

DreamPose: Fashion Image-to-Video Synthesis via Stable Diffusion

We present DreamPose, a diffusion-based method for generating animated f...
research
12/08/2022

MoFusion: A Framework for Denoising-Diffusion-based Motion Synthesis

Conventional methods for human motion synthesis are either deterministic...
research
06/19/2023

MotionGPT: Finetuned LLMs are General-Purpose Motion Generators

Generating realistic human motion from given action descriptions has exp...

Please sign up or login with your details

Forgot password? Click here to reset