RGB$\leftrightarrow$X: Image decomposition and synthesis using material- and lighting-aware diffusion models

Abstract

The three areas of realistic forward rendering, per-pixel inverse rendering, and generative image synthesis may seem like separate and unrelated sub-fields of graphics and vision. However, recent work has demonstrated improved estimation of per-pixel intrinsic channels (albedo, roughness, metallicity) based on a diffusion architecture; we call this the RGB$\to$X problem. We further show that the reverse problem of synthesizing realistic images given intrinsic channels, X$\to$RGB, can also be addressed in a diffusion framework. Focusing on the image domain of interior scenes, we introduce an improved diffusion model for RGB$\to$X, which also estimates lighting, as well as the first diffusion X$\to$RGB model capable of synthesizing realistic images from (full or partial) intrinsic channels. Our X$\to$RGB model explores a middle ground between traditional rendering and generative models: we can specify only certain appearance properties that should be followed, and give freedom to the model to hallucinate a plausible version of the rest. This flexibility makes it possible to use a mix of heterogeneous training datasets, which differ in the available channels. We use multiple existing datasets and extend them with our own synthetic and real data, resulting in a model capable of extracting scene properties better than previous work and of generating highly realistic images of interior scenes.

References in corpus (4)

• Learning Transferable Visual Models From Natural Language Supervision
• PVT v2: Improved Baselines with Pyramid Vision Transformer
• Deep Shading: Convolutional Neural Networks for Screen-Space Shading
• Intrinsic Image Decomposition via Ordinal Shading

Cited by in corpus (7)

• IntrinsicEdit: Precise generative image manipulation in intrinsic space
• LightLab: Controlling Light Sources in Images with Diffusion Models
• Generative Detail Enhancement for Physically Based Materials
• Uncertainty for SVBRDF Acquisition using Frequency Analysis
• MaterialPicker: Multi-Modal DiT-Based Material Generation
• Aerial Path Planning for Urban Geometry and Texture Co-Capture
• Detail Enhanced Gaussian Splatting for Large-Scale Volumetric Capture