Understand appearance
Intrinsic decomposition, reflectance, and material recognition grounded in real-world signals.
Workshop Focus
Where material perception meets controllable generation
Understanding and generating appearance are increasingly interdependent challenges. Shape, texture, and reflectance together define how we see the world and how models should represent it. Advances in intrinsic decomposition, material capture and recognition, and generative rendering now let us analyze and synthesize appearance with unprecedented fidelity.
Yet a key question remains: how can a deeper understanding of surface appearance support material perception and interaction, while enabling controllable, identity-preserving editing and consistent multimodal generation?
APPX brings together researchers in vision, graphics, and generative AI to explore the representations, datasets, and methods that bridge analysis and synthesis for more interpretable and controllable systems.
Model physical behavior
Inverse rendering, BRDF acquisition, and rendering-aware methods that remain faithful to physics.
Generate with control
Relighting, editing, and multimodal synthesis with stronger consistency and interpretability.
Call for Papers
Accepted CVPR 2026 papers are welcome
We invite submissions of papers accepted at CVPR 2026 that fall within the scope of appearance understanding and generation. Selected papers will be presented as posters during the workshop session, with spotlight presentations for featured works.
Invited Speakers
A lineup spanning vision, graphics, and generative modeling
Shree Nayar
Columbia University
Foundational contributions to computational photography and appearance modeling. Co-developer of the Oren-Nayar reflectance model.
Ravi Ramamoorthi
UCSD
Leading researcher in physics-based vision and rendering. Known for light transport theory, inverse rendering, and neural radiance fields.
Milos Hasan
NVIDIA Research
Focuses on realistic material appearance, inverse material estimation, and high-resolution material synthesis.
Hongzhi Wu
Zhejiang University
Known for material and reflectance acquisition, SVBRDF datasets, and learning-based high-fidelity material reconstruction.
Pieter Peers
College of William & Mary
Pioneering work on multi-view material capture, relighting, appearance reproduction, and generative appearance.
Dor Verbin
Google DeepMind
Focuses on geometry-, physics-, and graphics-informed methods for understanding and synthesizing 3D scenes.
Zhao Dong
Meta Reality Labs
Expert in 3D object and scene reconstruction, differentiable rendering, and materials and lighting modeling.
Schedule
June 4th, 1:30-5:35pm, Room 4AB
1:30 - 1:35 PM
Opening Remarks
Welcome and framing for the workshop themes.
1:35 - 2:05 PM
Invited Talk: Shree Nayar
Computational photography and appearance modeling.
2:05 - 2:35 PM
Invited Talk: Ravi Ramamoorthi
Physics-based vision, rendering, and radiance field methods.
2:35 - 3:00 PM
Invited Talk: Milos Hasan
Material synthesis and inverse material estimation.
3:00 - 3:40 PM
Coffee Break and Poster Session
Poster presentations and informal discussion with participants.
3:40 - 4:15 PM
Highlight Talks
Featured presentations from four to five selected papers.
4:10 - 4:30 PM
Invited Talk: Pieter Peers
Material capture, relighting, and appearance reproduction.
4:30 - 4:50 PM
Invited Talk: Hongzhi Wu
SVBRDF datasets and high-fidelity material reconstruction.
4:50 - 5:10 PM
Invited Talk: Zhao Dong
Reconstruction, differentiable rendering, and lighting models.
5:10 - 5:30 PM
Invited Talk: Dor Verbin
Graphics-informed methods for understanding and synthesizing 3D scenes.
5:30 - 5:35 PM
Closing Remarks
Wrap-up and directions for continued discussion.
Organizers
Elena Garces
Senior Research Scientist
Adobe Research
Valentin Deschaintre
Senior Research Scientist
Adobe Research
Iliyan Georgiev
Senior Research Scientist
Adobe Research
Giuseppe Vecchio
Research Scientist
Adobe Research
Michael Fischer
Research Scientist
Adobe Research