RelightFormer: Feed-forward Generative Transformer for Multiview Object Relighting

• 类型：arxiv
• 标识：2609.07414
• 链接：https://arxiv.org/abs/2609.07414
• 主分类：multimodal
• 形态：method
• TLDR：Image relighting is traditionally tackled via complex inverse rendering pipelines, which suffer from ill-posed optimization, or single-image generative models that ignore crucial multi-view cues necessary for understanding 3D geometry and material interactions. To address these limitations, we introduce a feed-forward generative Transformer for direct single- and multi-view image relighting that entirely bypasses explicit intrinsic property estimation. Adapted from a video foundation model, our architecture features a latent illumination module that dynamically injects target environment maps
• 待LLM分类：否
• 标题中文：RelightFormer：用于多视角物体重光照的前馈生成式 Transformer
• TLDR中文：图像重光照传统上依赖复杂的逆向渲染流程，不仅面临不适定优化问题，也难以像单图像生成模型那样利用对理解三维几何与材质交互至关重要的多视图线索。为此，我们提出一种前馈式生成 Transformer，可直接对单视图与多视图进行重光照，完全绕过显式本征属性估计。该架构由视频基础模型适配而来，核心是一个潜在光照模块，可动态注入目标环境贴图。
• 来源文件：
• /inbox/tom/_candidates/2026-09-09-agent-rag-longcontext-candidates.json