通过合成曝光包围实现线性图像生成
Linear Image Generation by Synthesizing Exposure Brackets
把高动态范围线性图像拆成曝光包围序列、再用 DiT 流匹配逐段生成，为需要保留极端高光与阴影的生成任务（HDR 合成、RAW 域生成、专业后期编辑）提供了一个可复用的表示与建模方案。
原文依据 we represent a linear image as a sequence of exposure brackets, each capturing a specific portion of the dynamic range
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论一张照片的生命始于光子撞击传感器，其信号经过复杂的图像信号处理（ISP）流水线，最终产生显示参考图像。然而，这类图像已不再忠实于入射光，其动态范围被压缩，并被主观偏好风格化。相比之下，RAW 图像记录的是非线性色调映射之前的直接传感器信号。经过相机响应曲线校正与去马赛克后，它们可被转换为线性图像，这是一种场景参考表示，直接反映真实辐照度，且对传感器相关因素具有不变性。由于图像传感器具有更好的动态范围与位深，线性图像比显示参考图像包含更丰富的信息，为用户在后期处理中留出更大的编辑空间。尽管有这一优势，当前的生成模型主要合成显示参考图像，这从根本上限制了下游编辑。本文研究文本到线性图像生成任务：以文本提示为条件，合成一幅高质量、场景参考、保留完整动态范围的线性图像，用于专业后期处理。生成线性图像具有挑战性，因为潜变量扩散模型中预训练的 VAE 由于更高的动态范围与位深，难以同时保留极端高光与阴影。为此，我们将线性图像表示为一系列曝光包围，每个曝光包围捕获动态范围的一个特定部分，并提出一种基于 DiT 的流匹配架构，用于文本条件曝光包围生成。我们进一步展示了下游应用，包括文本引导的线性图像编辑以及通过 ControlNet 实现的结构条件生成。
英文原文摘要 1767 字符
The life of a photo begins with photons striking the sensor, whose signals are passed through a sophisticated image signal processing (ISP) pipeline to produce a display-referred image. However, such images are no longer faithful to the incident light, being compressed in dynamic range and stylized by subjective preferences. In contrast, RAW images record direct sensor signals before non-linear tone mapping. After camera response curve correction and demosaicing, they can be converted into linear images, which are scene-referred representations that directly reflect true irradiance and are invariant to sensor-specific factors. Since image sensors have better dynamic range and bit depth, linear images contain richer information than display-referred ones, leaving users more room for editing during post-processing. Despite this advantage, current generative models mainly synthesize display-referred images, which inherently limits downstream editing. In this paper, we address the task of text-to-linear-image generation: synthesizing a high-quality, scene-referred linear image that preserves full dynamic range, conditioned on a text prompt, for professional post-processing. Generating linear images is challenging, as pre-trained VAEs in latent diffusion models struggle to simultaneously preserve extreme highlights and shadows due to the higher dynamic range and bit depth. To this end, we represent a linear image as a sequence of exposure brackets, each capturing a specific portion of the dynamic range, and propose a DiT-based flow-matching architecture for text-conditioned exposure bracket generation. We further demonstrate downstream applications including text-guided linear image editing and structure-conditioned generation via ControlNet.