---
OA_place: publisher
OA_type: gold
_id: '21474'
abstract:
- lang: eng
  text: Rendering novel, relit views of a human head, given a monocular portrait image
    as input, is an inherently underconstrained problem. The traditional graphics
    solution is to explicitly decompose the input image into geometry, material and
    lighting via differentiable rendering; but this is constrained by the multiple
    assumptions and approximations of the underlying models and parameterizations
    of these scene components. We propose 3DPR, an image-based relighting model that
    leverages generative priors learnt from multi-view One-Light-at-A-Time (OLAT)
    images captured in a light stage. We introduce a new diverse and large-scale multi-view
    4K OLAT dataset of 139 subjects to learn a high-quality prior over the distribution
    of high-frequency face reflectance. We leverage the latent space of a pre-trained
    generative head model that provides a rich prior over face geometry learnt from
    in-the-wild image datasets. The input portrait is first embedded in the latent
    manifold of such a model through an encoder-based inversion process. Then a novel
    triplane-based reflectance network trained on our lightstage data is used to synthesize
    high-fidelity OLAT images to enable image-based relighting. Our reflectance network
    operates in the latent space of the generative head model, crucially enabling
    a relatively small number of lightstage images to train the reflectance model.
    Combining the generated OLATs according to a given HDRI environment maps yields
    physically accurate environmental relighting results. Through quantitative and
    qualitative evaluations, we demonstrate that 3DPR outperforms previous methods,
    particularly in preserving identity and in capturing lighting effects such as
    specularities, self-shadows, and subsurface scattering.
acknowledgement: This work was supported by the ERC Consolidator Grant 4DReply (770784)
  and Saarbrücken Research Center for Visual Comput- ing, Interaction, and AI. We
  thank Oleksandr Sotnychenko for helping us with setting up data capture. Finally,
  we thank Shrisha Bharadwaj for discussions, proofreading and innumerable support.
article_number: '108'
article_processing_charge: No
arxiv: 1
author:
- first_name: Pramod
  full_name: Rao, Pramod
  last_name: Rao
- first_name: Abhimitra
  full_name: Meka, Abhimitra
  last_name: Meka
- first_name: Xilong
  full_name: Zhou, Xilong
  last_name: Zhou
- first_name: Gereon
  full_name: Fox, Gereon
  last_name: Fox
- first_name: B. R.
  full_name: Mallikarjun, B. R.
  last_name: Mallikarjun
- first_name: Fangneng
  full_name: Zhan, Fangneng
  last_name: Zhan
- first_name: Tim
  full_name: Weyrich, Tim
  last_name: Weyrich
- first_name: Bernd
  full_name: Bickel, Bernd
  id: 49876194-F248-11E8-B48F-1D18A9856A87
  last_name: Bickel
  orcid: 0000-0001-6511-9385
- first_name: Hanspeter
  full_name: Pfister, Hanspeter
  last_name: Pfister
- first_name: Wojciech
  full_name: Matusik, Wojciech
  last_name: Matusik
- first_name: Thabo
  full_name: Beeler, Thabo
  last_name: Beeler
- first_name: Mohamed
  full_name: Elgharib, Mohamed
  last_name: Elgharib
- first_name: Marc
  full_name: Habermann, Marc
  last_name: Habermann
- first_name: Christian
  full_name: Theobalt, Christian
  last_name: Theobalt
citation:
  ama: 'Rao P, Meka A, Zhou X, et al. 3DPR: Single image 3D portrait relighting with
    generative priors. In: <i>Proceedings SIGGRAPH Asia 2025 Conference Papers 2025</i>.
    Association for Computing Machinery; 2025. doi:<a href="https://doi.org/10.1145/3757377.3763962">10.1145/3757377.3763962</a>'
  apa: 'Rao, P., Meka, A., Zhou, X., Fox, G., Mallikarjun, B. R., Zhan, F., … Theobalt,
    C. (2025). 3DPR: Single image 3D portrait relighting with generative priors. In
    <i>Proceedings SIGGRAPH Asia 2025 Conference Papers 2025</i>. Hong Kong, Hong
    Kong: Association for Computing Machinery. <a href="https://doi.org/10.1145/3757377.3763962">https://doi.org/10.1145/3757377.3763962</a>'
  chicago: 'Rao, Pramod, Abhimitra Meka, Xilong Zhou, Gereon Fox, B. R. Mallikarjun,
    Fangneng Zhan, Tim Weyrich, et al. “3DPR: Single Image 3D Portrait Relighting
    with Generative Priors.” In <i>Proceedings SIGGRAPH Asia 2025 Conference Papers
    2025</i>. Association for Computing Machinery, 2025. <a href="https://doi.org/10.1145/3757377.3763962">https://doi.org/10.1145/3757377.3763962</a>.'
  ieee: 'P. Rao <i>et al.</i>, “3DPR: Single image 3D portrait relighting with generative
    priors,” in <i>Proceedings SIGGRAPH Asia 2025 Conference Papers 2025</i>, Hong
    Kong, Hong Kong, 2025.'
  ista: 'Rao P, Meka A, Zhou X, Fox G, Mallikarjun BR, Zhan F, Weyrich T, Bickel B,
    Pfister H, Matusik W, Beeler T, Elgharib M, Habermann M, Theobalt C. 2025. 3DPR:
    Single image 3D portrait relighting with generative priors. Proceedings SIGGRAPH
    Asia 2025 Conference Papers 2025. SA: SIGGRAPH Asia, 108.'
  mla: 'Rao, Pramod, et al. “3DPR: Single Image 3D Portrait Relighting with Generative
    Priors.” <i>Proceedings SIGGRAPH Asia 2025 Conference Papers 2025</i>, 108, Association
    for Computing Machinery, 2025, doi:<a href="https://doi.org/10.1145/3757377.3763962">10.1145/3757377.3763962</a>.'
  short: P. Rao, A. Meka, X. Zhou, G. Fox, B.R. Mallikarjun, F. Zhan, T. Weyrich,
    B. Bickel, H. Pfister, W. Matusik, T. Beeler, M. Elgharib, M. Habermann, C. Theobalt,
    in:, Proceedings SIGGRAPH Asia 2025 Conference Papers 2025, Association for Computing
    Machinery, 2025.
conference:
  end_date: 2025-12-18
  location: Hong Kong, Hong Kong
  name: 'SA: SIGGRAPH Asia'
  start_date: 2025-12-15
date_created: 2026-03-22T23:04:35Z
date_published: 2025-12-14T00:00:00Z
date_updated: 2026-03-23T14:45:58Z
day: '14'
ddc:
- '000'
department:
- _id: BeBi
doi: 10.1145/3757377.3763962
external_id:
  arxiv:
  - '2510.15846'
file:
- access_level: open_access
  checksum: a3dc426cdf7bbd84a192e5140bb3bb49
  content_type: application/pdf
  creator: dernst
  date_created: 2026-03-23T14:41:07Z
  date_updated: 2026-03-23T14:41:07Z
  file_id: '21479'
  file_name: 2025_SiggraphAsia_Rao.pdf
  file_size: 57903731
  relation: main_file
  success: 1
file_date_updated: 2026-03-23T14:41:07Z
has_accepted_license: '1'
language:
- iso: eng
month: '12'
oa: 1
oa_version: Published Version
publication: Proceedings SIGGRAPH Asia 2025 Conference Papers 2025
publication_identifier:
  isbn:
  - '9798400721373'
publication_status: published
publisher: Association for Computing Machinery
quality_controlled: '1'
scopus_import: '1'
status: public
title: '3DPR: Single image 3D portrait relighting with generative priors'
tmp:
  image: /images/cc_by_nc.png
  legal_code_url: https://creativecommons.org/licenses/by-nc/4.0/legalcode
  name: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)
  short: CC BY-NC (4.0)
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
year: '2025'
...