HDR Video Generation via Latent Alignment with Logarithmic Encoding

2026-04-13

AI Summary

This work addresses the challenge of high dynamic range (HDR) video generation, which is hindered by distribution mismatches arising from training existing generative models on bounded, perceptually compressed data. The authors propose a lightweight fine-tuning approach that leverages log encoding to naturally align HDR video content with the latent space of pretrained video generation models. Coupled with a training strategy grounded in camera degradation modeling, this method obviates the need for retraining the encoder or designing specialized architectures. By effectively harnessing the model’s preexisting visual priors, the approach generates high-quality HDR videos across diverse scenes and under extreme lighting conditions, demonstrating that HDR content can be efficiently handled without reconstructing the generative model itself.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

High dynamic range (HDR) imagery offers a rich and faithful representation of scene radiance, but remains challenging for generative models due to its mismatch with the bounded, perceptually compressed data on which these models are trained. A natural solution is to learn new representations for HDR, which introduces additional complexity and data requirements. In this work, we show that HDR generation can be achieved in a much simpler way by leveraging the strong visual priors already captured by pretrained generative models. We observe that a logarithmic encoding widely used in cinematic pipelines maps HDR imagery into a distribution that is naturally aligned with the latent space of these models, enabling direct adaptation via lightweight fine-tuning without retraining an encoder. To recover details that are not directly observable in the input, we further introduce a training strategy based on camera-mimicking degradations that encourages the model to infer missing high dynamic range content from its learned priors. Combining these insights, we demonstrate high-quality HDR video generation using a pretrained video model with minimal adaptation, achieving strong results across diverse scenes and challenging lighting conditions. Our results indicate that HDR, despite representing a fundamentally different image formation regime, can be handled effectively without redesigning generative models, provided that the representation is chosen to align with their learned priors.

Problem

• HDR video generation
• generative models
• dynamic range
• image representation
• latent space

Innovation

• HDR video generation
• latent alignment
• logarithmic encoding
• pretrained generative models
• camera-mimicking degradations