High-dynamic-range (HDR) formats and displays are becoming increasingly prevalent, yet state-of-the-art image generators (e.g., Stable Diffusion and FLUX) typically remain limited to low-dynamic-range (LDR) output due to the lack of large-scale HDR training data. In this work, we show that existing pretrained diffusion models can be easily adapted to HDR generation without retraining from scratch. A key challenge is that HDR images are natively represented in linear RGB, whose intensity and color statistics differ substantially from those of sRGB-encoded LDR images. This gap, however, can be effectively bridged by converting HDR inputs into perceptually uniform encodings (e.g., using PU21 or PQ). Empirically, we find that LDR-pretrained variational autoencoders (VAEs) reconstruct PU21-encoded HDR inputs with fidelity comparable to LDR data, whereas linear RGB inputs cause severe degradations. Motivated by this finding, we describe an efficient adaptation strategy that freezes the VAE and finetunes only the denoiser via low-rank adaptation in a perceptually uniform space. This results in a unified computational method that supports both text-to-HDR synthesis and single-image RAW-to-HDR reconstruction. Experiments demonstrate that our perceptually encoded adaptation consistently improves perceptual fidelity, text-image alignment, and effective dynamic range, relative to previous techniques.
Figures & tables
Figure 1: Qualitative results demonstrating the two modes supported by the proposed X2HDR. Top: text-to-HDR generation, visualized with exposure-adjusted views at EV −4 and EV +4 , highlighting synthesized content across highlights and shadows. Bottom: single-image RAW-to-HDR reconstruction. From underexposed and overexposed RAW inputs, X2HDR reconstructs HDR structures by inpainting saturated regions and denoising in low-light areas.
Setting
JOD ↑
QPSNR⋆↑
QSSIM⋆↑
QLPIPS⋆↓
QDISTS⋆↓
LDR
9.86
38.8
0.961
0.026
0.008
HDR (Linear)
8.54
27.9
0.867
0.129
0.067
HDR (PU21)
9.44
35.6
0.965
0.020
0.010
Table 1: VAE reconstruction fidelity on 512 aligned (LDR, HDR) frame pairs. The JOD score is given by ColorVideoVDP Mantiuk et al. (2024) . Q⋆ represents a family of LDR quality measures, equipped with exposure-optimized inverse display models Cao et al. (2024) .
Figure 2: VAE reconstruction of an aligned (LDR, HDR) pair under different input encodings. Left: source LDR/HDR images fed to the VAE. Right: reconstructions (top row) for LDR, linear HDR, and PU21-encoded HDR representation, together with ColorVideoVDP perceptual error heatmaps (bottom row), where stronger colors indicate more noticeable differences. Insets report the corresponding JOD scores.
Figure 4
Figure 3: System diagram of X2HDR. For Text-to-HDR , the prompt is encoded into text tokens cp ; a noisy HDR latent z is sampled and denoised by the LoRA-adapted DiT; the final result is decoded by the frozen VAE and mapped back to linear HDR values via the inverse PU21 transform (Equation ( 2 )). For RAW-to-HDR , a RAW capture is demosaicked, PU21-encoded (Equation ( 1 )), and converted into image tokens cRAW ; conditioning on cRAW , the same denoising and decoding pipeline reconstructs the HDR output. Both branches are shown in a single diagram for simplicity; in practice, training is performed separately with a task-specific LoRA for each branch.
Method
Image Quality ↑
Text-Image Alignment ↑
Dynamic Range ↑
Time (sec) ↓
Memory (GB) ↓
LEDiff
0.488
0.623
4.3
7.4
8.6
Bracket Diffusion
0.524
0.671
11.1
557.5
38.0
Ours ( SD )
0.568
0.672
14.4
2.2
2.9
Ours ( FLUX )
0.668
0.831
14.2
11.2
33.0
Table 2: Quantitative results for text-to-HDR. We report image quality and text-image alignment scores by Q-Eval-100K Zhang et al. (2025b) (higher is better), effective dynamic range by Equation ( 6 ) in stops (higher is better), and inference cost measured by runtime and peak memory (lower is better).
Figure 4: Visual comparison on text-to-HDR generation shown at three exposure settings (EV −4 , EV +0 , and EV +4 ). X2HDR better preserves light sources at low exposure (EV −4 ) and reveals informative shadow details at high exposure (EV +4 ), demonstrating a wider effective dynamic range. In contrast, LEDiff and Bracket Diffusion often underestimate peak luminance (dimming highlights at EV −4 ) and/or lose structure in underexposed areas ( e . g ., shadow collapse in the palm tree example for Bracket Diffusion).
Figure 5: Distributions of effective dynamic range (in stops) over 100 generated HDR images.
Figure 6: Visual comparison on RAW-to-HDR reconstruction. The competing methods exhibit similar failure modes: RawHDR is prone to noise amplification and color instability at low exposure, LEDiff often trades fidelity for excessive smoothing, and Bracket Diffusion’s low conditioning resolution tends to soften edges and attenuate fine structure after upsampling. X2HDR, on the other hand, better balances denoising and detail preservation, and more consistently restores plausible content in clipped regions ( e . g ., saturated sky) while maintaining stable color appearance. For better visual comparison, we map HDR outputs to LDR using the inverse display model of Mantiuk and Heidrich Mantiuk and Heidrich (2009) , optimizing the respective exposures (parameterized by luminance percentiles) to best align perceived brightness relative to the selected GT HDR exposure.
Method
JOD ↑
QPSNR⋆↑
QSSIM⋆↑
QLPIPS⋆↓
QDISTS⋆↓
RawHDR
6.58
19.3
0.591
0.282
0.133
LEDiff
6.74
19.7
0.570
0.319
0.151
Bracket Diffusion
7.08
21.6
0.727
0.282
0.148
X2HDR (Ours)
7.57
22.5
0.741
0.174
0.089
Table 3: Quantitative results for RAW-to-HDR.
Table 4: Ablation results for text-to-HDR and RAW-to-HDR.
Figure 7: Text-to-HDR ablation on HDR representation and finetuning. Linear/gamma/log encodings severely compress bright emitters, most evident at EV −4 , leading to a reduced effective dynamic range and poorer highlight/shadow recoverability. PQ closely matches PU21, preserving plausible details, whereas removing finetuning yields a “pseudo-HDR” appearance with exaggerated saturation/contrast and visible clipping artifacts.
Figure 8: RAW-to-HDR ablation on HDR representation. Linear/gamma/log encodings introduce pronounced quantization and color distortions, whereas PQ and PU21 achieve comparable perceptual quality, with more faithful highlight recovery and reduced artifacts. In the candle scene, the background content varies between runs despite similar encoding behavior, explained by the stochasticity of the inpainting process.
Appendix figures & tables17 assets
Supplementary material from the paper’s appendix.
Appendix
Figure 9: Encoding functions (linear, gamma, log, PQ, and PU21) mapping BT.2100-range linear luminance to display-encoded values (with log-scaled x-axis).
Figure 10: RGB value distributions under different encodings.
Figure 11: Representative VAE reconstruction failures on PU21-encoded HDR images, highlighted by ColorVideoVDP error maps.
Figure 12: t-SNE visualization of VAE latent distributions under different encodings. PU21 substantially reduces the LDR-HDR mismatch, while VAE finetuning produces only marginal changes.
Method
JOD ↑
QPSNR⋆↑
QSSIM⋆↑
QLPIPS⋆↓
QDISTS⋆↓
VAE w/ finetuning
7.34
22.3
0.759
0.197
0.111
VAE w/o finetuning (default)
7.57
22.5
0.741
0.174
0.089
Appendix
Table 5: Effect of VAE finetuning on RAW-to-HDR reconstruction.
Figure 13: Sample HDR images generated by X2HDR with SD-1.5 at different exposure levels.
Figure 14: Distributions of effective dynamic range (in stops) over 100 generated HDR images. Both the SD -based and FLUX -based variants produce images with wide dynamic range.
Figure 15: RAW-to-HDR reconstruction on synthetic data with controlled logarithmic luminance gradients. Rows correspond to different chromatic gradients (yellow, green, and red), and each method is visualized at two exposure levels. X2HDR strongly suppresses sensor noise, while LEDiff and Bracket Diffusion introduce structured artifacts. All methods remain challenged in recovering perfectly smooth gradients within severely clipped regions, indicating limited generalization to these synthetic patterns.
Figure 16: Visual comparison of LDR-to-HDR reconstruction. For the overexposed input (top), X2HDR recovers highlight structure more plausibly, while LEDiff and Bracket Diffusion produce visible distortions. For the underexposed input (bottom), X2HDR reveals more faithful shadow details with fewer artifacts.
Method
JOD ↑
QPSNR⋆↑
QSSIM⋆↑
QLPIPS⋆↓
QDISTS⋆↓
LEDiff
5.74
17.7
0.569
0.252
0.130
Bracket Diffusion
5.86
18.4
0.641
0.272
0.160
X2HDR (Ours)
6.47
19.3
0.680
0.187
0.101
Appendix
Table 6: Quantitative results for LDR-to-HDR.
Figure 17: Quantitative results of the perceptual study, reported in the unit of JOD (higher is better). Bars show mean JOD across stimuli, and error bars denote 95% confidence intervals.
Figure 18: Per-image JOD scores from our perceptual study. The top and bottom panels report text-to-HDR and RAW-to-HDR results, respectively. Error bars denote 95% confidence intervals via bootstrapping. A small horizontal jitter is applied to separate methods for improved visibility.
Figure 19: Multi-resolution HDR generation results. Examples are shown at three output resolutions, covering diverse lighting conditions and luminous phenomena ( e . g ., artificial lights, fire, sunlight, and night skies). Each HDR image is visualized at EV −4 / 0 / +4 to illustrate the effective dynamic range.
Figure 20: Single-RAW multi-look HDR authoring with text guidance. Given the same RAW input (left), our RAW-to-HDR model generates multiple HDR renderings conditioned on different prompts ( e . g ., “blue sky,” “orange sky,” and “cloudy sky”), producing diverse global appearance and atmosphere while preserving scene structure.
Figure 21: Banding artifacts induced by BF16 inference in text-to-HDR generation. When rendered at EV +5 , quantization in smooth, low-luminance gradients produces visible banding (left/middle), whereas FP32 inference preserves subtle gradients and eliminates the artifacts (right).
Figure 23: Limitations. Text-to-HDR: performance may degrade for out-of-domain styles, leading to less realistic appearance. RAW-to-HDR: implausible or inconsistent hallucinations may occur for large regions that are severely underexposed or overexposed.
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.
Naomi Ken Korem, Mohamed Oumoumad, Omer Hagage +8
Lightricks · Gear Productions · Tel Aviv University
High Dynamic Range (HDR) generation remains challenging for generative models, which are largely limited to low dynamic range outputs. Recent diffusionbased approaches approximate HDR by generating multiple exposure-conditioned samples, incurring high computational cost and structural inconsistencies across exposures. We propose LatentHDR, a framework that decouples scene generation from exposure modeling in latent space. A pretrained diffusion backbone produces a single coherent scene representation, while a lightweight conditional latent to-latent head deterministically maps it to exposure-specific representations. This enables the generation of a dense, structurally consistent exposure stack in a single pass. This design eliminates multi-pass diffusion, ensures cross-exposure alignment, and enables scalable HDR synthesis. LatentHDR supports both textand image-conditioned HDR generation for perspective and panoramic scenes. Experiments on synthetic data and the SI-HDR benchmark show that LatentHDR achieves state-of-the-art dynamic range with competitive perceptual quality, while reducing computation by an order of magnitude. Our results demonstrate that high-quality HDR generation can be achieved through structured latent modeling, challenging the need for stochastic multi-exposure generation.
The rapid adoption of HDR-capable devices has created a pressing need to convert the 8-bit Standard Dynamic Range (SDR) content into perceptually and physically accurate 10-bit High Dynamic Range (HDR). Existing inverse tone-mapping (ITM) methods often rely on fixed tone-mapping operators that struggle to generalize to real-world degradations, stylistic variations, and camera pipelines, frequently producing clipped highlights, desaturated colors, or unstable tone reproduction. We introduce LumaFlux, a first physically and perceptually guided diffusion transformer (DiT) for SDR-to-HDR reconstruction by adapting a large pretrained DiT. Our LumaFlux introduces (1) a Physically-Guided Adaptation (PGA) module that injects luminance, spatial descriptors, and frequency cues into attention through low-rank residuals; (2) a Perceptual Cross-Modulation (PCM) layer that stabilizes chroma and texture via FiLM conditioning from vision encoder features; and (3) an HDR Residual Coupler that fuses physical and perceptual signals under a timestep- and layer-adaptive modulation schedule. Finally, a lightweight Rational-Quadratic Spline decoder reconstructs smooth, interpretable tone fields for highlight and exposure expansion, enhancing the output of the VAE decoder to generate HDR. To enable robust HDR learning, we curate the first large-scale SDR-HDR training corpus. For fair and reproducible comparison, we further establish a new evaluation benchmark, comprising HDR references and corresponding expert-graded SDR versions. Across benchmarks, LumaFlux outperforms state-of-the-art baselines, achieving superior luminance reconstruction and perceptual color fidelity with minimal additional parameters.