mlsanityData to 14 Oct 2026

Generative Relightable Avatars

arXiv 2606.22718 ↗First posted 21 Jun 2026cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

We present Generative Relightable Avatars (GRA), a person-specific method for photorealistic free-view rendering and environment-map relighting of full-body humans. We postulate that modeling fine-grained appearance details is inherently a one-to-many problem that can benefit from a generative formulation. In contrast to fully regressive relightable avatar methods, GRA follows a hybrid approach that combines controllable, physics-grounded relighting with probabilistic refinement. Starting from a tracked animated mesh, we optimize material parameters in UV-space and render a coarse relit appearance under a target HDR environment map. Next, we refine the textures with a feed-forward model to capture pose-dependent texture dynamics and illumination effects beyond simplified reflectance assumptions. Finally, a fine-tuned video-to-video diffusion model transforms the physically grounded renderings into temporally coherent, high-detail videos while preserving 3D control, with an error-recycling strategy for generating long videos. Experimental evaluations demonstrate our method's improved perceptual quality over prior relightable avatar baselines. Project Page: https://vcai.mpi-inf.mpg.de/projects/GRA/

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 40 of its 49 references that are in the map. First posted on arXiv 21 Jun 2026.

Builds on

40 of 49 references.

References · 40

  1. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  2. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
  3. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  4. 2022Flow Matching for Generative Modelingcited by 5,702
  5. 2024Structured 3D Latents for Scalable and Versatile 3D Generationcited by 770
  6. 2025VACE: All-in-One Video Creation and Editingcited by 479
  7. 2025Hunyuan3D 2.0: Scaling Diffusion Models for High Resolution Textured 3D Assets Generationcited by 356
  8. 2022NeuS2: Fast Learning of Neural Implicit Surfaces for Multi-view Reconstructioncited by 314
  9. 2019Learning to Regress 3D Face Shape and Expression from an Image without 3D Supervisioncited by 176
  10. 2025Native and Compact Structured Latents for 3D Generationcited by 175
  11. 2017Learning to Predict Indoor Illumination from a Single Imagecited by 161
  12. 2024RGBleftrightarrowX: Image decomposition and synthesis using material- and lighting-aware diffusion modelscited by 161
28 more
  1. 2021Real-time Deep Dynamic Characterscited by 133
  2. 2024Neural Gaffer: Relighting Any Object via Diffusioncited by 91
  3. 2025Stable Video Infinity: Infinite-Length Video Generation with Error Recyclingcited by 73
  4. 2024SwitchLight: Co-design of Physics-driven Architecture and Pre-training Framework for Human Portrait Relightingcited by 67
  5. 2024Fr'echet Video Motion Distance: A Metric for Evaluating Motion Consistency in Videoscited by 58
  6. 2023Relightful Harmonization: Lighting-aware Portrait Background Replacementcited by 57
  7. 2025DiffusionRenderer: Neural Inverse and Forward Rendering with Video Diffusion Modelscited by 57
  8. 2022Relighting4D: Neural Relightable Human from Videoscited by 41
  9. 2024DifFRelight: Diffusion-Based Facial Performance Relightingcited by 35
  10. 2024MeshAvatar: Learning High-quality Triangular Human Avatars from Multi-view Videoscited by 34
  11. 2024Holo-Relighting: Controllable Volumetric Portrait Relighting from a Single Imagecited by 32
  12. 2023Relightable and Animatable Neural Avatar from Sparse-View Videocited by 28
  13. 2025Lux Post Facto: Learning Portrait Performance Relighting with Conditional Video Diffusion and a Hybrid Datasetcited by 28
  14. 2023Blind Video Deflickering by Neural Filtering with a Flawed Atlascited by 27
  15. 2025SynthLight: Portrait Relighting with Diffusion Model by Learning to Re-render Synthetic Facescited by 24
  16. 2025Relightable Full-Body Gaussian Codec Avatarscited by 22
  17. 2023IntrinsicAvatar: Physically Based Inverse Rendering of Dynamic Humans from Monocular Videos via Explicit Ray Tracingcited by 20
  18. 2022RANA: Relightable Articulated Neural Avatarscited by 12
  19. 2024Surfel-based Gaussian Inverse Rendering for Fast and Relightable Dynamic Human Reconstruction from Monocular Videocited by 12
  20. 2025Comprehensive Relighting: Generalizable and Consistent Monocular Human Relighting and Harmonizationcited by 12
  21. 2023Relightable Neural Actor with Intrinsic Decomposition and Pose Controlcited by 11
  22. 2024Generative Portrait Shadow Removalcited by 10
  23. 2025PractiLight: Practical Light Control Using Foundational Diffusion Modelscited by 7
  24. 2025MatAnyone 2: Scaling Video Matting via a Learned Quality Evaluatorcited by 7
  25. 2024NECA: Neural Customizable Human Avatarcited by 3
  26. 2025DNF-Avatar: Distilling Neural Fields for Real-time Animatable Avatar Relightingcited by 2
  27. 2025Relightable Holoported Characters: Capturing and Relighting Dynamic Human Performance from Sparse Viewscited by 2
  28. 2025ReLumix: Extending Image Relighting to Video via Video Diffusion Modelscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026D-Rex : Diffusion Rendering for Relightable Expressive Avatars ↗
  2. 2025Hyper Diffusion Avatars: Dynamic Human Avatar Generation using Network Weight Space Diffusion
  3. 2025GAS: Generative Avatar Synthesis from a Single Image
  4. 2026Monocular Avatar Reconstruction via Cascaded Diffusion Priors and UV-Space Differentiable Shading
  5. 2026AvatarDynamizer: From Static to Dynamic Human Avatars via Generative Dynamic Textures
  6. 2025ViSA: 3D-Aware Video Shading for Real-Time Upper-Body Avatar Creation
  7. 2026Pixel Cube: Diffusion-based Portrait Video Relighting Through Realistic Lighting Reproduction
  8. 2024GenCA: A Text-conditioned Generative Model for Realistic and Drivable Codec Avatars