Pith. sign in

REVIEW 3 major objections 6 minor 68 references

Deep Tone Mapping Operator for High Dynamic Range Images

T0 review · 3 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

Pith's one-line read DeepTMO claims a single multi-scale conditional GAN can tone-map varied HDR scenes into high-resolution LDR images in about 0.02 seconds, without per-scene tuning, matching or exceeding classical operators.

desk verdict Useful architecture study and a fast, parameter-free TMO, but the headline perceptual claim rests on a miscalculated binomial test and a partly circular TMQI evaluation. read the letter →

arxiv 1908.04197 v1 pith:5F5TUA3F submitted 2019-08-12 eess.IV cs.CVcs.GR

classification eess.IVcs.CVcs.GR
keywords highdynamicrangeimagingtonemappinggenerativeadversarialnetworksconditionalGANmulti-scalearchitectureTMQIimagequalityassessmentdeeplearning
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that one trained network, DeepTMO, can replace per-scene parameter tuning for tone mapping high dynamic range images. DeepTMO is a multi-scale conditional GAN that takes 32-bit linear HDR luminance as input and outputs a 1024x2048 tone-mapped LDR image in about 0.02 seconds. Its training targets are generated automatically by ranking 13 classical tone mapping operators with the TMQI quality metric and taking the best-scoring output for each scene. On 105 unseen test images, the multi-scale model reaches a mean TMQI of 0.88, the highest among all compared operators, and a pairwise subjective study favors DeepTMO for most tested scenes. If the result holds, it would remove the slow, tedious parameter tuning step from high-quality HDR visualization.

What carries the argument

The machinery is a multi-scale conditional GAN together with TMQI-based target selection. The generator has two branches: a coarse branch processing a 2x down-sampled HDR input and a fine branch processing the full-resolution input, so the output combines global structure with local detail. The discriminator is similarly two-scale, applying a PatchGAN classifier to the original and down-sampled image pairs; feature-matching loss from discriminator layers and a perceptual loss from a pre-trained deep network stabilize the adversarial training and suppress blurring, tiling and saturation artifacts. TMQI, an objective index combining structural fidelity and naturalness, is the mechanism that converts unlabeled HDR data into training pairs by picking the best-ranked output among thirteen classical TMOs for each scene.

What would settle it

Recruit expert observers to tune each of the 13 classical TMOs per scene on the 105 test images, then run a pairwise preference test between DeepTMO and the best tuned classical output for each scene. If the per-scene tuned outputs are preferred significantly more often, the claim that DeepTMO matches the best subjective quality without any parameter tuning fails.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that a two-scale generator and a two-scale PatchGAN discriminator, jointly trained with adversarial, feature-matching and perceptual losses, produce artifact-free tone-mapped outputs from linear HDR input. The generator compresses dynamic range directly in the luminance channel, with color restored by the ratio formula $C_{\mathrm{out}} = \frac{C_{\mathrm{in}}}{L_{\mathrm{in}}} L_{\mathrm{out}}$. Targets are chosen per scene as the highest-TMQI output among thirteen classical TMOs under default parameters, which lets the network learn a manifold over those operators instead of requiring hand-designed losses. In the quantitative comparison, DeepTMO's multi-scale model attains mean TMQI 0.88 +/- 0.06 over 105 test images, the best score in the table, and averages 0.0209 seconds per 1024x2048 image. The paper reports that in a forced-choice subjective test, observers significantly preferred DeepTMO outputs over the best TMQI-rated classical outputs for most scenes spanning humans, indoor, dark/noisy, structures and landscapes.

Load-bearing premise

The model is only as good as its training targets, which are chosen by ranking 13 tone mapping operators with the TMQI metric at default settings; if TMQI does not match what humans consider the best tone-mapped image, the network inherits that blind spot and the reported quality scores share the same bias.

Editorial extensions

If this is right

  • One trained model can tone-map a wide variety of real-world HDR content, including indoor, outdoor, human, dark/noisy, and landscape scenes, with no per-scene parameter tuning.
  • Near real-time tone mapping at 1024x2048 in about 0.02 seconds makes the approach practical for high-resolution displays and interactive applications.
  • The multi-scale generator-discriminator combination removes the blurring, tiling and saturation artifacts observed with single-scale cGAN designs.
  • Automatically generated TMQI-based targets allow training on unlabeled HDR data at scale, avoiding expensive subjective annotation.
  • Pairwise subjective testing suggests human observers prefer DeepTMO outputs to the best TMQI-ranked classical TMO outputs for most tested scenes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the targets are selected by TMQI, which the paper notes is color-blind, DeepTMO's saturated colors and halo rings around bright sources may be symptoms of the target selector; a different objective metric would likely change these failure modes.
  • The same pipeline could accept any target-selection rule, so upgrading the quality metric or expanding the pool of TMOs (or optimizing their parameters) should raise the ceiling of what the network can imitate.
  • The color reproduction step is a fixed ratio formula; learning color jointly with luminance in one network could address the over-saturation limitation without post-processing.
  • The model is only demonstrated on still images; applying DeepTMO to HDR video frames would test whether the scene adaptability transfers to temporal content.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper proposes DeepTMO, a cGAN-based tone mapping operator that maps 32-bit linear HDR luminance directly to a tone-mapped LDR image at 1024×2048 resolution without per-scene parameter tuning. Training targets are generated by ranking 13 classical TMOs with the TMQI metric and selecting the best-scoring output for each HDR image. The paper compares four generator/discriminator scale combinations, adopts a two-scale architecture with feature-matching and VGG perceptual losses, and reports a mean TMQI of 0.88 on 105 test images, outperforming the 13 baselines. Subjective support is claimed from a 20-observer pairwise study with 15 scenes. The authors also discuss limitations including TMQI's imperfection, halo artifacts around bright regions, and occasional over-saturation.

Significance. The core idea is timely and practically relevant: a fast, automatic, scene-adaptive TMO that avoids per-scene parameter search would be valuable for HDR display and processing pipelines. The paper's strengths are its systematic ablation of single- versus multi-scale generators and discriminators, the clear motivation for each loss term, the use of a reasonably large training set, and a candid limitations section that identifies exactly where the method fails (TMQI target selection, halos, saturation). However, the two pillars of the headline claim ('highest TMQI score' and 'significantly preferred subjective quality') are not currently established: the quantitative evaluation uses the same metric and baseline pool that generated the training labels, and the subjective significance threshold is miscalculated. If the authors can supply an independent evaluation and a correctly analyzed subjective study, the paper would be a useful contribution; with the present evidence, the strong perceptual claims must be substantially qualified.

major comments (3)
  1. [Section IV-A and Table II] The training-target oracle and the headline quantitative evaluation use the same objective function (TMQI) over the same 13 default-parameter TMOs. Because every training label is the TMQI-best output of that pool, the mean TMQI of 0.88 in Table II and the statement that DeepTMO 'achieves highest score' largely measure how faithfully the network reproduces the TMQI-optimal member of the pool; they do not independently establish perceptual quality. The paper itself concedes in Section VII-A that TMQI 'is not as perfect as the human visual system,' and Fig. 14 shows TMQI-best outputs that are not visually preferable. I recommend adding an evaluation with an independent objective quality metric (or with TMO baselines whose parameters are optimized), and separating the claim 'best under TMQI' from 'highest subjective quality' in the abstract and conclusions.
  2. [Section VI-B.3, Eq. (5), Fig. 12] The binomial significance threshold in the subjective study is miscalculated. The paper states that 13 votes out of 20 reach the 95% significance level because B(13,20,0.5)=0.9423, but 0.9423 is the cumulative probability P(X≤13); the one-sided p-value for observing 13 or more votes is 1−P(X≤12)≈0.132. The correct one-sided 95% critical value is 15 of 20 (P(X≥15)≈0.021), and the 5% low-side threshold is 5, not 6. Therefore the conclusion that DeepTMO is 'significantly preferred' for most scenes is not supported by the reported data. Please recompute the test (making clear whether N is participants or judgments, since each pair was shown twice), report p-values or confidence intervals, and handle multiple comparisons if per-scene significance is claimed.
  3. [Section VII-A, Figs. 15-16] The limitations section explicitly documents halo artifacts around high-illumination regions (Fig. 15) and occasional over-saturation requiring a separate color-correction step (Fig. 16), attributing the halos to insufficient training data. These are not purely future-work caveats; they directly affect the abstract's unqualified 'high-subjective quality' and 'wide spectrum of real-world scenes' claims. The paper should either quantify the frequency and severity of such artifacts on the 105-image test set, provide artifact-aware subjective evidence, or qualify the claims to content without extreme illumination sources.
minor comments (6)
  1. [Table II] The row labeled 'DeepTMO (Single G - Single G)' appears to be a typo; it should presumably read 'Single G - Single D' to match the four architecture combinations described in the text.
  2. [Fig. 12] The x-axis labels are scene indices only, with no legend or category markers, making it impossible to verify the per-category claims (Humans, Dark/Noisy, Indoor, Structures, Landscapes) from the figure; please add category annotations or a grouped bar chart.
  3. [Abstract and Section III] The term 'parameter-free' is overstated: inference requires no per-scene tuning, but the learned weights, the loss weights β=γ=10, the number of scales, and the normalization choices are manually selected hyper-parameters. 'Tuning-free at inference time' would be a more accurate description.
  4. [Section VI-A] The computation-time claim would be more informative if the timings of the 13 baseline TMOs on the same hardware and resolution were reported; currently only DeepTMO's inference time is given, so the 'fast' claim lacks a quantitative baseline comparison.
  5. [Sections II-C and VI] Since the paper positions itself against learning-based HDR/LDR methods, adding a quantitative comparison with the learning-based baselines mentioned in Section II-C (e.g., [41] or [52]) on the same test set would substantially strengthen the evaluation; the current Table II compares only with classical hand-crafted TMOs.
  6. [Minor typographical issues] There are several small typos: 'Dyanmic' in the Index Terms, 'arent manipulated much' in Section V-2, and 'fairing the best' in Section VI-A.

Circularity Check

1 steps flagged · score 4.0 of 10

TMQI-based target selection and evaluation share the same metric and TMO pool, so the quantitative 'highest score' is partly inherited from the label oracle; the subjective test's binomial threshold is also miscalculated.

  1. fitted input called prediction [Section IV-A (target selection), Section VI-A (quantitative comparison), Table II]
    "To find the target tone mapped image for each training HDR scene, we selected 13 classical TMOs ... For each HDR scene, we initially rank the obtained tone-mapped outputs from all the 13 TMOs using the TMQI metric. Then, the best scoring tone mapped output is selected as the 'target' ... To obtain the target test image, we follow a similar paradigm as provided in Section IV-A. ... The final results show that our proposed tone mapping model adapts for the variety of scenes and hence, achieves highest score."

    The reference 'best quality tone-mapped images' are defined as the TMQI argmax over the same 13 default-parameter TMOs, and DeepTMO is trained to regress to those exact targets. In the limit of perfect training, TMQI(DeepTMO(H)) equals TMQI(target(H)) = max_j TMQI(TMO_j(H)), so a top Table II score would be inherited from the label-generation rule rather than being an independent perceptual validation. Because the same metric and the same TMO pool serve both as the label generator and as the headline evaluation instrument, the quantitative comparison cannot independently confirm the subjective-quality claim. The paper's own Section VII-A caveat that TMQI 'is not as perfect as the human visual system' further undercuts treating the inherited high score as perceptual evidence.

full rationale

One circular step is present, but it is confined to the quantitative comparison. In Section IV-A the training targets are constructed as the TMQI-best output among 13 default-parameter TMOs; Section VI-A builds the test targets the same way, and Table II then uses TMQI on the same 13 TMOs to report that DeepTMO 'achieves highest score.' Since the network is trained to reproduce those TMQI-argmax targets, a near-perfect fit would by construction match the best pool member's TMQI, so the headline numerical superiority is at least partly inherited from the label-generation oracle rather than being independent evidence of perceptual quality. The paper itself concedes in Section VII-A that TMQI 'is not as perfect as the human visual system' and shows TMQI-best outputs that are not visually desirable, which further weakens the quantitative claim as support for 'high-subjective quality.' Separately, the pairwise subjective study in Section VI-B is intended to break this loop, but its reported 13/20 (and 6/20) significance thresholds are not correct one-sided binomial critical values at the 95%/5% levels (the former requires at least 15/20), so the perceptual claim is not independently established. These are correctness and evidence-weight concerns rather than additional circularity. The core engineering content—the cGAN architecture, multi-scale design, loss-function analysis, and runtime measurements—does not depend on the circular loop, so the paper is only partially circular.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the TMQI-selected training targets, the representativeness of the 13-TMO pool with default parameters, and the generalization of a cGAN trained on 698 images. The network weights and several hyperparameters are fitted or hand-chosen; no new physical entities are introduced.

free parameters (3)
  • beta (feature-matching loss weight) = 10
    Set by hand in Section III-D; controls the contribution of the FM loss to the total objective.
  • gamma (perceptual loss weight) = 10
    Set by hand in Section III-D; controls the contribution of the VGG perceptual loss.
  • number of scales in multi-scale generator/discriminator = 2
    Chosen by the authors in Section III-C based on ablation comparisons; not derived from theory.
assumptions (4)
  • domain assumption TMQI is a valid proxy for subjective quality of tone-mapped images.
    Used in Section IV-A to rank 13 TMOs and select training targets; the paper admits in Section VII-A that TMQI is not as perfect as the human visual system.
  • domain assumption The 13 selected classical TMOs with default parameters provide a representative space of high-quality tone-mapped images.
    The training targets are chosen as the best of these 13 outputs; the paper notes in Section IV-A that parameter tuning could improve each TMO, so the pool may be suboptimal.
  • domain assumption A cGAN trained on 698 images with augmentation generalizes to unseen HDR scenes without per-scene tuning.
    This is the core learning assumption of the paper, implicit in the problem formulation in Section I and the training setup in Section V.
  • domain assumption The multi-scale generator and discriminator architecture from [51] transfers to tone mapping at 1024x2048 resolution.
    The paper adapts the Pix2PixHD design in Section III-C; the effectiveness of this transfer is an empirical assumption tested only on the paper's own dataset.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Tone Mapping Operator for High Dynamic Range Images." pith.science (2026). https://pith.science/paper/5F5TUA3F

@misc{pith2026190804197,
  author       = {Pith},
  title        = {Pith review of: Deep Tone Mapping Operator for High Dynamic Range Images},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5F5TUA3F}},
  note         = {Machine review of arXiv:1908.04197}
}
read the original abstract

A computationally fast tone mapping operator (TMO) that can quickly adapt to a wide spectrum of high dynamic range (HDR) content is quintessential for visualization on varied low dynamic range (LDR) output devices such as movie screens or standard displays. Existing TMOs can successfully tone-map only a limited number of HDR content and require an extensive parameter tuning to yield the best subjective-quality tone-mapped output. In this paper, we address this problem by proposing a fast, parameter-free and scene-adaptable deep tone mapping operator (DeepTMO) that yields a high-resolution and high-subjective quality tone mapped output. Based on conditional generative adversarial network (cGAN), DeepTMO not only learns to adapt to vast scenic-content (e.g., outdoor, indoor, human, structures, etc.) but also tackles the HDR related scene-specific challenges such as contrast and brightness, while preserving the fine-grained details. We explore 4 possible combinations of Generator-Discriminator architectural designs to specifically address some prominent issues in HDR related deep-learning frameworks like blurring, tiling patterns and saturation artifacts. By exploring different influences of scales, loss-functions and normalization layers under a cGAN setting, we conclude with adopting a multi-scale model for our task. To further leverage on the large-scale availability of unlabeled HDR data, we train our network by generating targets using an objective HDR quality metric, namely Tone Mapping Image Quality Index (TMQI). We demonstrate results both quantitatively and qualitatively, and showcase that our DeepTMO generates high-resolution, high-quality output images over a large spectrum of real-world scenes. Finally, we evaluate the perceived quality of our results by conducting a pair-wise subjective study which confirms the versatility of our method.

Figures

Figures reproduced from arXiv: 1908.04197 by the authors.

Figure 2
Figure 2. Comparison between CNN (encoder-decoder) with Lprp-loss (perceptual) and DeepTMO (single-scale). variability of training-set scenes to a range [0,255]. Thus, the trained model effectively predicts a fairly mean luminance value for most of the pixels in output images to minimize the overall loss function. Another simple idea could be to use TMQI directly as loss function. However, due to the mathematical design of TM… view at source ↗
Figure 1
Figure 1. Comparison between CNN (encoder-decoder) with L1-loss and DeepTMO (single-scale). Inlets in row 2 show that DeepTMO yields sharp and high resolution output, whereas the CNN results in blurred outputs. c) Motivation for GANs: To achieve the desired TMO, one solution is to use a simple L1 or perceptual (Lprp) loss function [43] with an encoder-decoder architecture as utilized in the past by various inverse-TMOs for ge… view at source ↗
Figure 3
Figure 3. DeepTMO Training Pipeline. B. DeepTMO (Single-Scale) [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: DeepTMO (single-scale) generator and discriminator architecture. The generator in (a) is an encoder-decoder architecture. Residual blocks in (c) consist of two sequential convolution layers applied to the input, producing a residual correction. Discriminator in (b) con…
Figure 6
Figure 6. Figure 6: DeepTMO multi-scale generator architecture. While the finer generator Go has the original image as its input, the input to Gd is a 2× down-sampled version. composed of three components given by: G (F ront) o , G (Res) o and G (Back) o . As illustrated in [PITH_FULL_IM…
Figure 5
Figure 5. Figure 5: Impact of Multi-scale Discriminator and Generator. introducing a Multi-D, the noisy patterns observed in Single￾D, are suppressed to a large extent (for e.g., as seen in Fig. 5a and Fig. 5b). However, we still witness minor traces of these artifacts due to Single-G’s v…
Figure 7
Figure 7. Figure 7: DeepTMO (single-scale) with/without FM and VGG Loss. a) Impact of Feature Matching and Perceptual Loss term: Both LFM and LLprp loss terms act as guidance to the adversarial loss function preserving overall natural image statistics and training without both these terms…
Figure 8
Figure 8. Figure 8: Batch Normalization vs. Instance Normalization. 1) Instance Vs. Batch Norm: We use instance normaliza￾tion [61], which is equivalent to applying batch normaliza￾tion [62] using a batch size equal to 1. The efficacy of the instance-norm is showcased in [PITH_FULL_IMAGE…
Figure 9
Figure 9. Figure 9: Comparison between our DeepTMO outputs and outputs from top-2 ranked tone-mapped scenes on TMQI metrics for a variety of real-world scenes including indoor, scenes with structures, landscape, dark/noisy scenes. In brackets we show corresponding TMQI scores [PITH_FULL_…
Figure 10
Figure 10. Figure 10: Comparison between DeepTMO and targets, highlighting the zoom-ins with the corresponding HDR-linear input. completeness, we also provide scores achieved by target tone￾mapped outputs. The curves clearly show that the generated tone mapped images for DeepTMO compete cl…
Figure 12
Figure 12. Figure 12: Subjective Test Results. Preference probability of our DeepTMO over best performing target TMOs for 15 scenes repre￾senting 5 different scene categories. as shown in [PITH_FULL_IMAGE:figures/full_fig_p011_12.png]
Figure 11
Figure 11. Figure 11: Quantitative performance comparison of best performing DeepTMO with the target TMOs. b) Computation Time: Inference is performed on test￾images of size 1024 × 2048 and takes on an average 0.0187 sec. for single-scale and 0.0209 sec. for multi-scale designs, TABLE II: …
Figure 13
Figure 13. Figure 13: Computation time in seconds. realistic and appealing to him/her. Participants were provided with unlimited time to make their decision and record their choice. The experiment was divided into a training and test session, where training involved each participant being …
Figure 15
Figure 15. Figure 15: Halo effect. (a) DeepTMO output, (b) DeepTMO trained with log-scaled values, (c) and (d) 2 top ranking TMO outputs. VII. CONCLUSION, LIMITATIONS AND FUTURE WORK Designing a fast, automated tone-mapping operator that can reproduce best subjective quality outputs from a…
Figure 16
Figure 16. Figure 16: Color Correction. (a) DeepTMO, (b) and (c) are the color corrected DeepTMO controlled by parameter s from [67] [PITH_FULL_IMAGE:figures/full_fig_p012_16.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

68 extracted references · 50 canonical work pages

  1. [52]

    A generative adversarial network for tone mapping hdr images,

    V . A. Patel, P. Shah, and S. Raman, “A generative adversarial network for tone mapping hdr images,” in Computer Vision, Pattern Recognition, Image Processing, and Graphics , Singapore, 2018, pp. 220–231

  2. [1]

    Visualization of high dynamic range images,

    A. Pardo and G. Sapiro, “Visualization of high dynamic range images,” IEEE Transactions on Image Processing , vol. 12, no. 6, pp. 639–647, June 2003

  3. [2]

    High dynamic range image compression by optimizing tone mapped image quality index,

    K. Ma, H. Yeganeh, K. Zeng, and Z. Wang, “High dynamic range image compression by optimizing tone mapped image quality index,” IEEE Transactions on Image Processing , vol. 24, Oct 2015

  4. [3]

    Gradient-based tone mapping for rate-distortion optimized backward- compatible high dynamic range compression,

    D. Gommelet, A. Roumy, C. Guillemot, M. Ropert, and J. L. Tanou, “Gradient-based tone mapping for rate-distortion optimized backward- compatible high dynamic range compression,” IEEE Transactions on Image Processing, pp. 5936–5949, Dec 2017

  5. [4]

    Learning-based Adaptive Tone Mapping for Keypoint Detection,

    A. Rana, G. Valenzise, and F. Dufaux, “Learning-based Adaptive Tone Mapping for Keypoint Detection,” in IEEE International Conference on Multimedia & Expo (ICME’2017) , Hong Kong, China, Jul. 2017

  6. [5]

    Dufaux, P

    F. Dufaux, P. Le Callet, R. Mantiuk, and M. Mrak, High Dynamic Range Video: From Acquisition, to Display and Applications . Academic Press, 2016

  7. [6]

    McCann and A

    J. McCann and A. Rizzi, The art and science of HDR imaging , 01 2012

  8. [7]

    Learning-Based Tone Mapping Operator for Image Matching,

    A. Rana, G. Valenzise, and F. Dufaux, “Learning-Based Tone Mapping Operator for Image Matching,” in IEEE International Conference on Image Processing (ICIP’2017). Beijing, China: IEEE, 2017

Show all 68 references
  1. [8]

    Evaluation of tone mapping operators using a high dynamic range display,

    P. Ledda, A. Chalmers, T. Troscianko, and H. Seetzen, “Evaluation of tone mapping operators using a high dynamic range display,” ACM Transactions on Graphics (TOG) , pp. 640–648, 2005

  2. [9]

    Evaluation of Tone Mapping Operators for HDR-Video,

    G. Eilertsen, R. Wanat, R. K. Mantiuk, and J. Unger, “Evaluation of Tone Mapping Operators for HDR-Video,” Computer Graphics F orum, 2013

  3. [10]

    Imagenet classification with deep convolutional neural networks,

    A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet classification with deep convolutional neural networks,” in Advances in neural infor- mation processing systems , 2012, pp. 1097–1105

  4. [11]

    Generative adversarial nets,

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y . Bengio, “Generative adversarial nets,” in Advances in neural information processing systems , 2014

  5. [12]

    Conditional generative adversarial nets,

    M. Mirza and S. Osindero, “Conditional generative adversarial nets,” arXiv preprint arXiv:1411.1784 , 2014

  6. [13]

    Image-to-image translation with conditional adversarial networks,

    P. Isola, J. Zhu, T. Zhou, and A. A. Efros, “Image-to-image translation with conditional adversarial networks,” CVPR, 2017

  7. [14]

    Optimizing tone mapping operators for keypoint detection under illumination changes,

    A. Rana, G. Valenzise, and F. Dufaux, “Optimizing tone mapping operators for keypoint detection under illumination changes,” in 2016 IEEE 18th International Workshop on Multimedia Signal Processing (MMSP), Sep. 2016, pp. 1–6

  8. [15]

    Applicationspecific tone mapping via genetic program- ming,

    K. Debattista, “Applicationspecific tone mapping via genetic program- ming,” Computer Graphics F orum, vol. 37, no. 1, pp. 439–450, 2017

  9. [16]

    Feature learning for the image retrieval task,

    A. Rana, J. Zepeda, and P. Perez, “Feature learning for the image retrieval task,” in Computer Vision - ACCV 2014 Workshops . Springer International Publishing, 2015, pp. 152–165

  10. [17]

    Precomputed real-time texture synthesis with markovian generative adversarial networks,

    C. Li and M. Wand, “Precomputed real-time texture synthesis with markovian generative adversarial networks,” in European Conference on Computer Vision . Springer, 2016, pp. 702–716

  11. [18]

    Photo-realistic single image super-resolution using a generative adversarial network,

    C. Ledig, L. Theis, F. Husz ´ar, J. Caballero, A. Cunningham, A. Acosta, A. Aitken, A. Tejani, J. Totz, Z. Wang et al. , “Photo-realistic single image super-resolution using a generative adversarial network,” arXiv preprint arXiv:1609.04802, 2016

  12. [19]

    Banterle, A

    F. Banterle, A. Artusi, K. Debattista, and A. Chalmers, Advanced High Dynamic Range Imaging: Theory and Practice , February 2011

  13. [20]

    A visibility matching tone reproduction operator for high dynamic range scenes,

    G. W. Larson, H. Rushmeier, and C. Piatko, “A visibility matching tone reproduction operator for high dynamic range scenes,” IEEE Transactions on Visualization and Computer Graphics , pp. 291–306, Oct. 1997

  14. [21]

    Adaptive logarith- mic mapping for displaying high contrast scenes,

    F. Drago, K. Myszkowski, T. Annen, and N. Chiba, “Adaptive logarith- mic mapping for displaying high contrast scenes,” Computer Graphics F orum, pp. 419–426, 2003

  15. [22]

    An adaptive sampling technique for multidimensional integration by ray-tracing,

    C. Schlick, “An adaptive sampling technique for multidimensional integration by ray-tracing,” in Photorealistic Rendering in Computer Graphics. Springer, 1994

  16. [23]

    Two methods for display of high contrast images,

    J. Tumblin, J. K. Hodgins, and B. K. Guenter, “Two methods for display of high contrast images,” ACM Trans. Graph. , pp. 56–94, Jan. 1999

  17. [24]

    Adaptive gain control for high dynamic range image display,

    S. Pattanaik and H. Yee, “Adaptive gain control for high dynamic range image display,” in Proceedings of the 18th Spring Conference on Computer Graphics , ser. SCCG ’02. ACM, 2002, pp. 83–87

  18. [25]

    Spatially nonuniform scaling functions for high contrast images,

    K. Chiu, M. Herf, P. Shirley, S. Swamy, C. Wang, and K. Zimmerman, “Spatially nonuniform scaling functions for high contrast images,” in Proceedings of Graphics Interface ’93 , ser. GI ’93, Toronto, Ontario, Canada, 1993, pp. 245–253

  19. [26]

    Fast bilateral filtering for the display of high-dynamic-range images,

    F. Durand and J. Dorsey, “Fast bilateral filtering for the display of high-dynamic-range images,” in 29th Annual Conference on Computer Graphics and Interactive Techniques , ser. SIGGRAPH ’02, 2002

  20. [27]

    A perceptual framework for contrast processing of high dynamic range images,

    R. Mantiuk, K. Myszkowski, and H. P. Seidel, “A perceptual framework for contrast processing of high dynamic range images,” ACM Trans. Appl. Percept., vol. 3, no. 3, Jul. 2006. 14

  21. [28]

    Gradient domain high dynamic range compression,

    R. Fattal, D. Lischinski, and M. Werman, “Gradient domain high dynamic range compression,” ACM Trans. Graph. , vol. 21, no. 3, pp. 249–256, Jul. 2002

  22. [29]

    Learning-based tone mapping operator for efficient image matching,

    A. Rana, G. Valenzise, and F. Dufaux, “Learning-based tone mapping operator for efficient image matching,” IEEE Transactions on Multime- dia, pp. 1–1, 2018

  23. [30]

    Evaluation of HDR tone mapping methods using essential perceptual attributes,

    M. Cadik, M. Wimmer, L. Neumann, and A. Artusi, “Evaluation of HDR tone mapping methods using essential perceptual attributes,” Computers and Graphics , pp. 330 – 349, 2008

  24. [31]

    Objective quality assessment of tone-mapped images,

    H. Yeganeh and Z. Wang, “Objective quality assessment of tone-mapped images,” IEEE Transactions on Image Processing , vol. 22, no. 2, pp. 657–667, Feb 2013

  25. [32]

    Fsitm: A feature similarity index for tone-mapped images,

    H. Z. Nafchi, A. Shahkolaei, R. F. Moghaddam, and M. Cheriet, “Fsitm: A feature similarity index for tone-mapped images,”IEEE Signal Processing Letters, vol. 22, no. 8, pp. 1026–1029, Aug 2015

  26. [33]

    Hdr image reconstruction from a single exposure using deep cnns,

    G. Eilertsen, J. Kronander, G. Denes, R. K. Mantiuk, and J. Unger, “Hdr image reconstruction from a single exposure using deep cnns,” ACM Transactions on Graphics (TOG) , vol. 36, no. 6, p. 178, 2017

  27. [34]

    Expandnet: A deep convolutional neural network for high dynamic range expansion from low dynamic range content

    D. Marnerides, T. Bashford-Rogers, J. Hatchett, and K. Debattista, “Expandnet: A deep convolutional neural network for high dynamic range expansion from low dynamic range content.” 2018

  28. [35]

    Deep inverse tone mapping using LDR based learning for estimating HDR images with absolute luminance,

    Y . Kinoshita and H. Kiya, “Deep inverse tone mapping using LDR based learning for estimating HDR images with absolute luminance,” CoRR, vol. abs/1903.01277, 2019

  29. [36]

    Hybrid loss for learning single-image-based HDR reconstruction,

    K. Moriwaki, R. Yoshihashi, R. Kawakami, S. You, and T. Naemura, “Hybrid loss for learning single-image-based HDR reconstruction,” CoRR, vol. abs/1812.07134, 2018

  30. [37]

    Deep reverse tone mapping,

    Y . Endo, Y . Kanamori, and J. Mitani, “Deep reverse tone mapping,”ACM Transactions on Graphics (Proc. of SIGGRAPH ASIA 2017) , vol. 36, no. 6, nov 2017

  31. [38]

    Deep high dynamic range imag- ing of dynamic scenes,

    N. K. Kalantari and R. Ramamoorthi, “Deep high dynamic range imag- ing of dynamic scenes,” ACM Transactions on Graphics (Proceedings of SIGGRAPH 2017) , vol. 36, no. 4, 2017

  32. [39]

    Deep joint demo- saicking and denoising,

    M. Gharbi, G. Chaurasia, S. Paris, and F. Durand, “Deep joint demo- saicking and denoising,” ACM Transactions on Graphics (TOG), vol. 35, no. 6, p. 191, 2016

  33. [40]

    Bilateral guided upsampling,

    J. Chen, A. Adams, N. Wadhwa, and S. W. Hasinoff, “Bilateral guided upsampling,” ACM Transactions on Graphics (TOG) , vol. 35, no. 6, p. 203, 2016

  34. [41]

    Deep bilateral learning for real-time image enhancement,

    M. Gharbi, J. Chen, J. T. Barron, S. W. Hasinoff, and F. Durand, “Deep bilateral learning for real-time image enhancement,”ACM Trans. Graph., Jul. 2017

  35. [42]

    Deep feature consistent deep image transformations: Downscaling, decolorization and HDR tone mapping,

    X. Hou, J. Duan, and G. Qiu, “Deep feature consistent deep image transformations: Downscaling, decolorization and HDR tone mapping,” CoRR, 2017

  36. [43]

    Perceptual losses for real- time style transfer and super-resolution,

    J. Johnson, A. Alahi, and L. Fei-Fei, “Perceptual losses for real- time style transfer and super-resolution,” in European Conference on Computer Vision. Springer, 2016, pp. 694–711

  37. [44]

    Semantic image inpainting with perceptual and contextual losses,

    R. Yeh, C. Chen, T. Y . Lim, M. Hasegawa-Johnson, and M. N. Do, “Semantic image inpainting with perceptual and contextual losses,” arXiv preprint arXiv:1607.07539 , 2016

  38. [45]

    Deep recursive hdri: Inverse tone mapping using generative adversarial networks,

    S. Lee, G. Hwan An, and S.-J. Kang, “Deep recursive hdri: Inverse tone mapping using generative adversarial networks,” in The European Conference on Computer Vision (ECCV) , September 2018

  39. [46]

    Deep chain hdri: Reconstructing a high dynamic range image from a single low dynamic range image,

    S. Lee, G. H. An, and S. Kang, “Deep chain hdri: Reconstructing a high dynamic range image from a single low dynamic range image,” IEEE Access, pp. 49 913–49 924, 2018

  40. [47]

    Unpaired image-to-image translation using cycle-consistent adversarial networks,

    J. Zhu, T. Park, P. Isola, and A. A. Efros, “Unpaired image-to-image translation using cycle-consistent adversarial networks,” arXiv preprint arXiv:1703.10593, 2017

  41. [48]

    Unsupervised image-to-image trans- lation networks,

    M. Liu, T. Breuel, and J. Kautz, “Unsupervised image-to-image trans- lation networks,” arXiv preprint arXiv:1703.00848 , 2017

  42. [49]

    Adversarial inversion: Inverse graphics with adversarial priors,

    H. Tung, A. Harley, W. Seto, and K. Fragkiadaki, “Adversarial inversion: Inverse graphics with adversarial priors,” arXiv preprint arXiv:1705.11166, 2017

  43. [50]

    Photographic image synthesis with cascaded refinement networks,

    Q. Chen and V . Koltun, “Photographic image synthesis with cascaded refinement networks,” in IEEE International Conference on Computer Vision, ICCV 2017, V enice, Italy, October 22-29, 2017 , 2017

  44. [51]

    High- resolution image synthesis and semantic manipulation with conditional gans,

    T. Wang, M. Liu, J. Zhu, A. Tao, J. Kautz, and B. Catanzaro, “High- resolution image synthesis and semantic manipulation with conditional gans,” arXiv preprint arXiv:1711.11585 , 2017

  45. [53]

    Generating images with perceptual similar- ity metrics based on deep networks,

    A. Dosovitskiy and T. Brox, “Generating images with perceptual similar- ity metrics based on deep networks,” in Advances in Neural Information Processing Systems, 2016, pp. 658–666

  46. [54]

    Image style transfer using convolutional neural networks,

    L. A. Gatys, A. S. Ecker, and M. Bethge, “Image style transfer using convolutional neural networks,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2016, pp. 2414–2423

  47. [55]

    Very deep convolutional networks for large-scale image recognition,

    K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” arXiv preprint arXiv:1409.1556 , 2014

  48. [56]

    Imagenet large scale visual recognition challenge,

    O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein et al., “Imagenet large scale visual recognition challenge,” International Journal of Computer Vision, vol. 115, no. 3, pp. 211–252, 2015

  49. [57]

    Fairchild

    M. Fairchild. (2007) The hdr photographic survey. [Online]. Available: http://www.rit-mcsl.org/fairchild/HDR.html

  50. [58]

    A tone mapping algorithm for high contrast images,

    M. Ashikhmin, “A tone mapping algorithm for high contrast images,” pp. 145–156, 2002

  51. [59]

    Photographic tone reproduction for digital images,

    E. Reinhard, M. Stark, P. Shirley, and J. Ferwerda, “Photographic tone reproduction for digital images,” ACM Trans. Graph., pp. 267–276, Jul. 2002

  52. [60]

    Least squares generative adversarial networks,

    X. Mao, Q. Li, H. Xie, R. Y . K. Lau, Z. Wang, and S. P. Smolley, “Least squares generative adversarial networks,” in IEEE International Conference on Computer Vision, ICCV 2017, V enice, Italy, October 22- 29, 2017 , 2017, pp. 2813–2821

  53. [61]

    Instance normalization: The missing ingredient for fast stylization,

    D. Ulyanov, A. Vedaldi, and V . Lempitsky, “Instance normalization: The missing ingredient for fast stylization,” preprint arXiv:1607.08022, 2016

  54. [62]

    Batch normalization: Accelerating deep network training by reducing internal covariate shift,

    S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate shift,” CoRR, vol. abs/1502.03167, 2015

  55. [63]

    Automatic differentiation in pytorch,

    A. Paszke, S. Gross, S. Chintala, G. Chanan, E. Yang, Z. DeVito, Z. Lin, A. Desmaison, L. Antiga, and A. Lerer, “Automatic differentiation in pytorch,” 2017

  56. [64]

    Adam: A method for stochastic optimization,

    D. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980 , 2014

  57. [65]

    The rank analysis of incomplete block designs — I. The method of paired comparisons,

    R. A. Bradley and M. E. Terry, “The rank analysis of incomplete block designs — I. The method of paired comparisons,” Biometrika, vol. 39, pp. 324–345, 1952

  58. [66]

    Towards high dynamic range extensions of hevc: subjective evaluation of potential coding technologies,

    P. Hanhart, M. Rerabek, and T. Ebrahimi, “Towards high dynamic range extensions of hevc: subjective evaluation of potential coding technologies,” Applications of Digital Image Processing XXXVIII , p. 95990G, 2015

  59. [67]

    Color correction for tone mapping,

    R. Mantiuk, R. Mantiuk, A. Tomaszewska, and W. Heidrich, “Color correction for tone mapping,” Computer Graphics F orum, 2009

  60. [68]

    Learning from weak and noisy labels for semantic segmentation,

    Z. Lu, Z. Fu, T. Xiang, P. Han, L. Wang, and X. Gao, “Learning from weak and noisy labels for semantic segmentation,”IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 39, no. 3, pp. 486–500, March 2017

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.