Relighting is a computer graphics and generative AI technique that computationally alters the illumination of an existing image or video, repositioning or replacing light sources to produce a different lighting environment without re-capturing the scene. Modern approaches use neural rendering models trained on diverse lighting conditions, enabling portrait relighting, object relighting, and scene relighting from a single input image. Open-source workflows in tools such as ComfyUI implement IC-Light and similar diffusion-based pipelines that rival or exceed proprietary solutions.
Semantic Classification
Content
- Image Based Lighting Node-Based Diffusion Pipeline Interface lllyasviel/IC-Light: More relighting! (github.com)
- Relighting Relight Better than Magnific - Relighting v5 (better UX) | ComfyUI Workflow (openart.ai)
- Magnific AI Relight is Worse than Open Source - YouTube Computational Image Relighting Technique
- https://zju3dv.github.io/IntrinsicAnything/ relighting model
- Apple Inc Technology Corporation facial relighting
- Apple facial relighting
- Relighting AI Video [2402.18848] SwitchLight: Co-design of Physics-driven Architecture and Pre-training Framework for Human Portrait Relighting (arxiv.org)
- Beeble AI Lighting for Filmmakers
- Relighting Relight Better than Magnific - Relighting v5 (better UX) | ComfyUI Workflow (openart.ai)
- Magnific AI Relight is Worse than Open Source - YouTube Computational Image Relighting Technique