Skip to main navigation Skip to search Skip to main content

IDOL: Unified Dual-Modal Latent Diffusion for Human-Centric Joint Video-Depth Generation

  • Yuanhao Zhai
  • , Kevin Lin
  • , Linjie Li
  • , Chung Ching Lin
  • , Jianfeng Wang
  • , Zhengyuan Yang
  • , David Doermann
  • , Junsong Yuan
  • , Zicheng Liu
  • , Lijuan Wang
  • SUNY Buffalo
  • Microsoft USA
  • Advanced Micro Devices

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Scopus citations

Abstract

Significant advances have been made in human-centric video generation, yet the joint video-depth generation problem remains underexplored. Most existing monocular depth estimation methods may not generalize well to synthesized images or videos, and multi-view-based methods have difficulty controlling the human appearance and motion. In this work, we present IDOL (unIfied Dual-mOdal Latent diffusion) for high-quality human-centric joint video-depth generation. Our IDOL consists of two novel designs. First, to enable dual-modal generation and maximize the information exchange between video and depth generation, we propose a unified dual-modal U-Net, a parameter-sharing framework for joint video and depth denoising, wherein a modality label guides the denoising target, and cross-modal attention enables the mutual information flow. Second, to ensure a precise video-depth spatial alignment, we propose a motion consistency loss that enforces consistency between the video and depth feature motion fields, leading to harmonized outputs. Additionally, a cross-attention map consistency loss is applied to align the cross-attention map of the video denoising with that of the depth denoising, further facilitating spatial alignment. Extensive experiments on the TikTok and NTU120 datasets show our superior performance, significantly surpassing existing methods in terms of video FVD and depth accuracy.

Original languageEnglish
Title of host publicationComputer Vision – ECCV 2024 - 18th European Conference, Proceedings
EditorsAleš Leonardis, Elisa Ricci, Stefan Roth, Olga Russakovsky, Torsten Sattler, Gül Varol
PublisherSpringer Science and Business Media Deutschland GmbH
Pages134-152
Number of pages19
ISBN (Print)9783031726323
DOIs
StatePublished - 2025
Event18th European Conference on Computer Vision, ECCV 2024 - Milan, Italy
Duration: Sep 29 2024Oct 4 2024

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume15073 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference18th European Conference on Computer Vision, ECCV 2024
Country/TerritoryItaly
CityMilan
Period09/29/2410/4/24

Fingerprint

Dive into the research topics of 'IDOL: Unified Dual-Modal Latent Diffusion for Human-Centric Joint Video-Depth Generation'. Together they form a unique fingerprint.

Cite this