Recent advances in text-to-image (T2I) diffusion models have enabled remarkable control over various attributes, yet precise color specification remains a fundamental challenge. Existing approaches, such as ColorPeel, rely on model personalization, requiring additional optimization and limiting flexibility in specifying arbitrary colors. In this work, we introduce ColorWave, a novel training-free approach that achieves exact RGB-level color control in diffusion models without fine-tuning. By systematically analyzing the cross-attention mechanisms within IP-Adapter, we uncover an implicit binding between textual color descriptors and reference image features. Leveraging this insight, our method rewires these bindings to enforce precise color attribution while preserving the generative capabilities of pretrained models. Our approach maintains generation quality and diversity, outperforming prior methods in accuracy and applicability across diverse object categories. Through extensive evaluations, we demonstrate that ColorWave establishes a new paradigm for structured, color-consistent diffusion-based image synthesis.

Blind all-in-one image restoration models aim to recover a high-quality image from an input degraded with unknown distortions. However, these models require all the possible degradation types to be defined during the training stage while showing limited generalization to unseen degradations, which limits their practical application in complex cases. In this paper, we introduce ABAIR, a simple yet effective adaptive blind all-in-one restoration model that not only handles multiple degradations and generalizes well to unseen distortions but also efficiently integrates new degradations by training only a small subset of parameters. We first train our baseline model on a large dataset of natural images with multiple synthetic degradations. To enhance its ability to recognize distortions, we incorporate a segmentation head that estimates per-pixel degradation types. Second, we adapt our initial model to varying image restoration tasks using independent low-rank adapters. Third, we learn to adaptively combine adapters to versatile images via a flexible and lightweight degradation estimator. This specialize-then-merge approach is both powerful in addressing specific distortions and flexible in adapting to complex tasks. Moreover, our model not only surpasses state-of-the-art performance on five- and three-task IR setups but also demonstrates superior generalization to unseen degradations and composite distortions.

Visual illusions in humans arise when interpreting out- of-distribution stimuli: if the observer is adapted to certain statistics, perception of outliers deviates from reality. Recent studies have shown that artificial neural networks (ANNs) can also be deceived by visual illusions. This revelation raises profound questions about the nature of visual information. Why are two independent systems, both human brains and ANNs, susceptible to the same illusions? Should any ANN be capable of perceiving visual illusions? Are these perceptions a feature or a flaw? In this work, we study how visual illusions are encoded in diffusion models. Remarkably, we show that they present human-like brightness/color shifts in their latent space. We use this fact to demonstrate that diffusion models can predict visual illusions. Furthermore, we also show how to generate new unseen visual illusions in realistic images using text-to-image diffusion models. We validate this ability through psychophysical experiments that show how our model-generated illusions also fool humans.

Decomposing a video into a layer-based representation is crucial for easy video editing for the creative industries, as it enables independent editing of specific layers. Existing video-layer decomposition models rely on implicit neural representations (INRs) trained independently for each video, making the process time-consuming when applied to new videos. Noticing this limitation, we propose a meta-learning strategy to learn a generic video decomposition model to speed up the training on new videos. Our model is based on a hypernetwork architecture which, given a video-encoder embedding, generates the parameters for a compact INR-based neural video decomposition model. Our strategy mitigates the problem of single-video overfitting and, importantly, shortens the convergence of video decomposition on new, unseen videos. Our code is available at: https://hypernvd.github.io/.

White balance (WB) correction in scenes with multiple illuminants remains a persistent challenge in computer vision. Recent methods explored fusion-based approaches, where a neural network linearly blends multiple sRGB versions of an input image, each processed with predefined WB presets. However, we demonstrate that these methods are suboptimal for common multi-illuminant scenarios. Additionally, existing fusion-based methods rely on sRGB WB datasets lacking dedicated multi-illuminant images, limiting both training and evaluation. To address these challenges, we introduce two key contributions. First, we propose an efficient transformer-based model that effectively captures spatial dependencies across sRGB WB presets, substantially improving upon linear fusion techniques. Second, we introduce a large-scale multi-illuminant dataset comprising over 16,000 sRGB images rendered with five different WB settings, along with WB-corrected images. Our method achieves up to 100\% improvement over existing techniques on our new multi-illuminant image fusion dataset.

Advancements in imaging technology have enabled hardware to support 10 to 16 bits per channel, facilitating precise manipulation in applications like image editing and video processing. While deep neural networks promise to recover high bit-depth representations, they still have issues such as banding artifacts, loss of fine texture details, and inadequate preservation of subtle color gradients, due to their reliance on scale-invariant image information, limiting performance in certain scenarios. In this paper, we introduce a novel approach that integrates a super-resolution architecture to extract detailed a priori information from images. By leveraging interpolated data generated during the super-resolution process, our method achieves pixel-level recovery of fine-grained color details. Additionally, we demonstrate that spatial features learned through the super-resolution process significantly contribute to the recovery of detailed color depth information. Experiments on benchmark datasets demonstrate that our approach outperforms state-of-the-art methods, highlighting the potential of super-resolution for high-fidelity color restoration.

Color imaging algorithms - such as color correction, spectral estimation and color constancy - are developed and validated with spectral reflectance data. However, the choice of the reflectance data set - used in development and tuning - not only affects the results of these algorithms but it also changes the ranking of the different approaches. We propose that this fragility is because it is difficult to measure/sample enough data to statistically represent the large number of degrees of freedom apparent in spectral reflectances. In this paper, we propose that the space of reflectance data should not be sampled but, rather, integrated. Specifically, we advocate that the convex closure of a reflectance data set - all convex combinations of all spectra - should be used instead of discrete reflectance samples. To make the integration computation tractable, we approximate these convex closures by their enclosing hyper-cube in a privileged coordinate system. We use color correction as an exemplar color imaging problem to demonstrate the utility of our approach.

3D lookup tables (3D LUTs) are a key component for image enhancement. Modern image signal processors (ISPs) have dedicated support for these as part of the camera rendering pipeline. Cameras typically provide multiple options for picture styles, where each style is usually obtained by applying a unique handcrafted 3D LUT. Current approaches for learning and applying 3D LUTs are notably fast, yet not so memory-efficient, as storing multiple 3D LUTs is required. For this reason and other implementation limitations, their use on mobile devices is less popular. In this work, we propose a Neural Implicit LUT (NILUT), an implicitly defined continuous 3D color transformation parameterized by a neural network. We show that NILUTs are capable of accurately emulating real 3D LUTs. Moreover, a NILUT can be extended to incorporate multiple styles into a single network with the ability to blend styles implicitly. Our novel approach is memory-efficient, controllable and can complement previous methods, including learned ISPs. Code at https://github.com/mv-lab/nilut

Progress in lighting estimation is tracked by computing existing image quality assessment (IQA) metrics on images from standard datasets. While this may appear to be a reasonable approach, we demonstrate that doing so does not correlate to human preference when the estimated lighting is used to relight a virtual scene into a real photograph. To study this, we design a controlled psychophysical experiment where human observers must choose their preference amongst rendered scenes lit using a set of lighting estimation algorithms selected from the recent literature, and use it to analyse how these algorithms perform according to human perception. Then, we demonstrate that none of the most popular IQA metrics from the literature, taken individually, correctly represent human perception. Finally, we show that by learning a combination of existing IQA metrics, we can more accurately represent human preference. This provides a new perceptual framework to help evaluate future lighting estimation algorithms. To encourage future research, all (anonymised) perceptual data and code are available at https://lvsn.github.io/PerceptionMetric/.

We present a method that, given two different views of the same scene taken by two cameras with unknown settings and internal parameters, corrects the colors of one of the images making it look as if it was captured under the other camera settings. Our method is able to deal with any standard non-linear encoded images (gamma-corrected, logarithmic-encoded, or any other) without requiring any previous knowledge of the encoding. To this end, our method makes use of two important observations. First, the camera imaging pipeline from RAW to sRGB can be well approximated by considering just a per-pixel shading and a color transformation matrix, and second, for correcting the images we only need to estimate a single matrix -that will contain information from both of the original images- and an approximation of the shading term (that emulates the non-linearity). Our proposed method is fast and the results have no spurious artifacts. The method outperforms the state-of-the-art when compared with other methods that do not require knowledge of the encoding used. It is also able to compete with -and even surpass in some cases- methods that consider information about image encoding.

Color harmony refers to combinations of colors that look pleasing together. We present a novel strategy to harmonize an image's colors using color-palette manipulation and color naming. Palette-based color manipulation is a method that extracts a few colors to represent the image. Modifying the palette colors modifies the color appearance of the image. A color-naming model is a mechanism to categorize colors into a fixed number of basic color terms. Working from a color-naming model, we derive a set ofprototype colorsand demonstrate that mapping an image's extracted color palette to the nearest prototype colors effectively harmonizes the image's colors. This straightforward approach yields visually compelling, outperforming more complex color harmony methods.

Burst image restoration attempts to effectively utilize the complementary cues appearing in sequential images to produce a high-quality image. Most current methods use all the available images to obtain the reconstructed image. However, using more images for burst restoration is not always the best option regarding reconstruction quality and efficiency, as the images acquired by handheld imaging devices suffer from degradation and misalignment caused by the camera noise and shake. In this paper, we extend the perception-distortion tradeoff theory by introducing multiple-frame information. We propose the area of the unattainable region as a new metric for perception-distortion tradeoff evaluation and comparison. Based on this metric, we analyse the performance of burst restoration from the perspective of the perception-distortion tradeoff under both aligned bursts and misaligned bursts situations. Our analysis reveals the importance of inter-frame alignment for burst restoration and shows that the optimal burst length for the restoration model depends both on the degree of degradation and misalignment.

Recent advances in camera designs and imaging pipelines allow us to capture high-quality images using smartphones. However, due to the small size and lens limitations of the smartphone cameras, we commonly find artifacts or degradation in the processed images. The most common unpleasant effects are noise artifacts, diffraction artifacts, blur, and HDR overexposure. Deep learning methods for image restoration can successfully remove these artifacts. However, most approaches are not suitable for real-time applications on mobile devices due to their heavy computation and memory requirements.In this paper, we propose LPIENet, a lightweight network for perceptual image enhancement, with the focus on deploying it on smartphones. Our experiments show that, with much fewer parameters and operations, our model can deal with the mentioned artifacts and achieve competitive performance compared with state-of-the-art methods on standard benchmarks. Moreover, to prove the efficiency and reliability of our approach, we deployed the model directly on commercial smartphones and evaluated its performance. Our model can process 2K resolution images under 1 second in mid-level commercial smartphones.

Abstract Achieving visually consistent colors across multiple images is important when images are used in photo albums, websites, and brochures. Unfortunately, only a handful of methods address multi‐image color consistency compared to one‐to‐one color transfer techniques. Furthermore, existing methods do not incorporate high‐level features that can assist graphic designers in their work. To address these limitations, we introduce a framework that builds upon a previous palette‐based color consistency method and incorporates three high‐level features: white balance, saliency, and color naming. We show how these features overcome the limitations of the prior multi‐consistency workflow and showcase the user‐friendly nature of our framework.

In the quality evaluation of high dynamic range and wide color gamut (HDR/WCG) images, a number of works have concluded that native HDR metrics, such as HDR visual difference predictor (HDR-VDP), HDR video quality metric (HDR-VQM), or convolutional neural network (CNN)-based visibility metrics for HDR content, provide the best results. These metrics consider only the luminance component, but several color difference metrics have been specifically developed for, and validated with, HDR/WCG images. In this paper, we perform subjective evaluation experiments in a professional HDR/WCG production setting, under a real use case scenario. The results are quite relevant in that they show, firstly, that the performance of HDR metrics is worse than that of a classic, simple standard dynamic range (SDR) metric applied directly to the HDR content; and secondly, that the chrominance metrics specifically developed for HDR/WCG imaging have poor correlation with observer scores and are also outperformed by an SDR metric. Based on these findings, we show how a very simple framework for creating color HDR metrics, that uses only luminance SDR metrics, transfer functions, and classic color spaces, is able to consistently outperform, by a considerable margin, state-of-the-art HDR metrics on a varied set of HDR content, for both perceptual quantization (PQ) and Hybrid Log-Gamma (HLG) encoding, luminance and chroma distortions, and on different color spaces of common use.

Gamut mapping is the problem of transforming the colors of image or video content so as to fully exploit the color palette of the display device where the content will be shown, while preserving the artistic intent of the original content's creator. In particular, in the cinema industry, the rapid advancement in display technologies has created a pressing need to develop automatic and fast gamut mapping algorithms. In this article, we propose a novel framework that is based on vision science models, performs both gamut reduction and gamut extension, is of low computational complexity, produces results that are free from artifacts and outperforms state-of-the-art methods according to psychophysical tests. Our experiments also highlight the limitations of existing objective metrics for the gamut mapping problem.

By capturing a more complete rendition of scene light than standard 2D cameras, light-field technology represents an important step towards closing the gap between live action cinematography and computer graphics. Light-field cameras accomplish this by simultaneously capturing the same scene under different angular configurations, providing directional information that allows for a multitude of post-production effects. Among the practical challenges related to capturing multiple images simultaneously, a very important problem is the fact that the different images do not perfectly match in terms of color, which severely complicates all further processing. In this work we adapt and extend to the light-field scenario a color stabilization method previously proposed for standard multi-camera shoots, and demonstrate experimentally that it provides an improvement over the state-of-the-art techniques for light-field imaging.

We present a color matching method that deals with different non-linear encodings. In particular, given two different views of the same scene taken by two cameras with unknown settings and internal parameters, and encoded with unknown non-linear curves, our method is able to correct the colors of one of the images making it look as if it was captured under the other camera's settings. Our method is based on treating the in-camera color processing pipeline as a concatenation of a matrix multiplication on the linear image followed by a non-linearity. This allows us to model a color stabilization transformation among the two shots by estimating a single matrix -that will contain information from both of the original images- and an extra parameter that complies with the non-linearity. The method is fast and the results have no spurious colors. It outperforms the state-of-the-art both visually and according to several metrics, and can handle HDR encodings and very challenging real-life examples.

Visual illusions teach us that what we see is not always what is represented in the physical world. Their special nature make them a fascinating tool to test and validate any new vision model proposed. In general, current vision models are based on the concatenation of linear and non-linear operations. The similarity of this structure with the operations present in Convolutional Neural Networks (CNNs) has motivated us to study if CNNs trained for low-level visual tasks are deceived by visual illusions. In particular, we show that CNNs trained for image denoising, image deblurring, and computational color constancy are able to replicate the human response to visual illusions, and that the extent of this replication varies with respect to variation in architecture and spatial pattern size. These results suggest that in order to obtain CNNs that better replicate human behaviour, we may need to start aiming for them to better replicate visual illusions.

Multiple-exposure approaches for high dynamic range (HDR) image generation share a set of building assumptions: that color channels are independent and that the camera response function (CRF) remains constant while changing the exposure. The first contribution of this paper is to highlight how these assumptions, which were correct for film photography, do not hold in general for digital cameras. As a consequence, results of multiexposure HDR methods are less accurate, and when tone-mapped they often present problems like hue shifts and color artifacts. The second contribution is to propose a method to stabilize the CRF while coupling all color channels, which can be applied to both static and dynamic scenes, and yield artifact-free results that are more accurate than those obtained with state-of-the-art methods according to several image metrics.

Image dehazing deals with the removal of undesired loss of visibility in outdoor images due to the presence of fog. Retinex is a color vision model mimicking the ability of the Human Visual System to robustly discount varying illuminations when observing a scene under different spectral lighting conditions. Retinex has been widely explored in the computer vision literature for image enhancement and other related tasks. While these two problems are apparently unrelated, the goal of this work is to show that they can be connected by a simple linear relationship. Specifically, most Retinex-based algorithms have the characteristic feature of always increasing image brightness, which turns them into ideal candidates for effective image dehazing by directly applying Retinex to a hazy image whose intensities have been inverted. In this paper, we give theoretical proof that Retinex on inverted intensities is a solution to the image dehazing problem. Comprehensive qualitative and quantitative results indicate that several classical and modern implementations of Retinex can be transformed into competing image dehazing algorithms performing on pair with more complex fog removal methods, and can overcome some of the main challenges associated with this problem.

Image dehazing tries to solve an undesired loss of visibility in outdoor images due to the presence of fog. Recently, machine-learning techniques have shown great dehazing ability. However, in order to be trained, they require training sets with pairs of foggy images and their clean counterparts, or a depth-map. In this paper, we propose to learn the appearance of fog from weakly-labeled data. Specifically, we only require a single label per-image stating if it contains fog or not. Based on the Multiple-Instance Learning framework, we propose a model that can learn from image-level labels to predict if an image contains haze reasoning at a local level. Fog detection performance of the proposed method compares favorably with two popular techniques, and the attention maps generated by the model demonstrate that it effectively learns to disregard sky regions as indicative of the presence of fog, a common pitfall of current image dehazing techniques.

There is not a large research on how to use color information for improving results in image denoising. Currently, most of the methods modify the color space from standard red green blue (sRGB) to an opponent-like one as better results are obtained, but out of this conversion, color is mostly ignored in the image denoising pipelines. In this letter, we propose a color decomposition to preprocess an image before applying a typical denoising. Our decomposition consists in obtaining a set of images in the spherical coordinate system, each of them with the origin of the spherical transformation in a different color value. These color values, that we call color centers, are defined so as to be far away from the dominant colors of the image. Once in the spherical coordinate system, we perform a mild denoising operation with some state-of-the-art method in the angular components. Then, we convert these images back to sRGB, and we merge them depending on the distance between the color of each pixel and the color centers. Finally, we denoise the preprocessed image with the same state-of-the-art method used in our preprocessing. Experiments show that our method outperforms the results of directly applying the denoising method on the input image for different state-of-the-art denoising methods.

We propose a novel image-dehazing technique based on the minimization of two energy functionals and a fusion scheme to combine the output of both optimizations. The proposed fusion-based variational image-dehazing (FVID) method is a spatially varying image enhancement process that first minimizes a previously proposed variational formulation that maximizes contrast and saturation on the hazy input. The iterates produced by this minimization are kept, and a second energy that shrinks faster intensity values of well-contrasted regions is minimized, allowing to generate a set of difference-of-saturation (DiffSat) maps by observing the shrinking rate. The iterates produced in the first minimization are then fused with these DiffSat maps to produce a haze-free version of the degraded input. The FVID method does not rely on a physical model from which to estimate a depth map, nor it needs a training stage on a database of human-labeled examples. Experimental results on a wide set of hazy images demonstrate that FVID better preserves the image structure on nearby regions that are less affected by fog, and it is successfully compared with other current methods in the task of removing haze degradation from faraway regions.

Emerging display technologies are able to produce images with a much wider color gamut than those of conventional distribution gamuts for cinema and TV, creating an opportunity for the development of gamut extension algorithms (GEAs) that exploit the full color potential of these new systems. In this paper, we present a novel GEA, implemented as a PDE-based optimization procedure related to visual perception models, that performs gamut extension (GE) by taking into account the analysis of distortions in hue, chroma, and saturation. User studies performed using a digital cinema projector under cinematic (low ambient light, large screen) conditions show that the proposed algorithm outperforms the state of the art, producing gamut extended images that are perceptually more faithful to the wide-gamut ground truth, as well as free of color artifacts and hue shifts. We also show how currently available image quality metrics, when applied to the GE problem, provide results that do not correlate with users' choices.

We propose a method for the color stabilization of cinema shots coming from different cameras that use unknown logarithmic encoding curves. The log-encoding curves are approximated by a concatenation of gamma-curves, whose values are accurately computed using image matches. The color stabilization procedure, based on the generic color processing pipeline of a digital camera, can be performed after the estimation of the encoding curves, and it also requires the existence of image matches. Our work can be applied in different scenarios such as multi-camera shoots, native-3D cinema, or color grading in post-production.

In this paper we present a novel approach to the problem of exposure fusion of a stack of pictures for the generation of high dynamic range (HDR) radiance maps. All exposure fusion approaches, when applied on 8-bit non-RAW pictures, perform photometric/ncalibration by estimating and inverting the camera response function, which is assumed to be a channelwise-independent function which does not change with the exposure. Our experiments show that these assumptions do not always hold and that the camera may automatically introduce changes (in gain, white balance, gamma correction value) from one exposure to the next when performing the non-linear operations involved in recording pictures in non-RAW formats such as JPEG. The net result is that HDR radiance maps obtained from exposure fusion of non-linear data may have substantially more error than if computed directly from the linear, RAW data. Our proposed method overcomes this problem and compensates for the changes introduced by the camera by matching the color correction and gamma correction transforms of all pictures to those of a reference picture in the stack, providing a clear improvement in terms of PSNR with respect to the classical method of Debevec and Malik.

Images obtained under adverse weather conditions, such as haze or fog, typically exhibit low contrast and faded colors, which may severely limit the visibility within the scene. Unveiling the image structure under the haze layer and recovering vivid colors out of a single image remains a challenging task, since the degradation is depth-dependent and conventional methods are unable to overcome this problem. In this work, we extend a well-known perception-inspired variational framework for single image dehazing. Two main improvements are proposed. First, we replace the value used by the framework for the gray-world hypothesis by an estimation of the mean of the clean image. Second, we add a set of new terms to the energy functional for maximizing the interchannel contrast. Experimental results show that the proposed enhanced variational image dehazing (EVID) method outperforms other state-of-the-art methods both qualitatively and quantitatively. In particular, when the illuminant is uneven, our EVID method is the only one that recovers realistic colors, avoiding the appearance of strong chromatic artifacts.

Abstract—Blind gamma estimation is the problem of estimating the gamma function that is applied to a linear image both for perceptual reasons and for the compensation of the non-linear behavior of displays. Gamma values change both inter- and intra-camera. In the latter case, the change comes from the use of different scene settings. In this paper we propose a new approach that relies on the use of more than a single image from the same scene. We estimate the gammas for all the different images at the same time with a method based on exploiting the structure of the standard in-camera processing pipeline. Our results improve over the state-of-the-art. Index Terms—Blind gamma estimation, image enhancement I.

We propose a method for color stabilization of shots of the same scene, taken under the same illumination, where one image is chosen as reference and one or several other images are modified so that their colors match those of the reference. We make use of two crucial but often overlooked observations: first, that the core of the color correction chain in a digital camera is simply a multiplication by a 3×3 matrix; second, that to color-match a source image to a reference image we do not need to compute their two color correction matrices, it is enough to compute the operation that transforms one matrix into the other. This operation is a 3×3 matrix as well, which we call H. Once we have H, we just multiply by it each pixel value of the source and obtain an image which matches in color the reference. To compute H we only require a set of pixel correspondences, we do not need any information about the cameras used, neither models nor specifications or parameter values. We propose an implementation of our framework which is very simple and fast, and show how it can be successfully employed in a number of situations, comparing favorably with the state of the art. There is a wide range of applications of our technique, both for amateur and professional photography and video: color matching for multicamera TV broadcasts, color matching for 3D cinema, color stabilization for amateur video, etc.

Finding color representations that are stable to illuminant changes is still an open problem in computer vision. Until now, most approaches have been based on physical constraints or statistical assumptions derived from the scene, whereas very little attention has been paid to the effects that selected illuminants have on the final color image representation. The novelty of this paper is to propose perceptual constraints that are computed on the corrected images. We define the category hypothesis, which weights the set of feasible illuminants according to their ability to map the corrected image onto specific colors. Here, we choose these colors as the universal color categories related to basic linguistic terms, which have been psychophysically measured. These color categories encode natural color statistics, and their relevance across different cultures is indicated by the fact that they have received a common color name. From this category hypothesis, we propose a fast implementation that allows the sampling of a large set of illuminants. Experiments prove that our method rivals current state-of-art performance without the need for training algorithmic parameters. Additionally, the method can be used as a framework to insert top-down information from other sources, thus opening further research directions in solving for color constancy.