Video temporal grounding (VTG) aims to locate precise segments in videos based on language queries, which is a fundamental challenge in video understanding. While recent Multimodal Large Language Models (MLLMs) have shown promise in tackling VTG through reinforcement learning (RL), they overlook the challenges arising from both the quality and difficulty of training samples. (1) Partially annotated samples. Many manually annotated samples contain relevant segments beyond the annotated interval, introducing ambiguous supervision. (2) Hard-to-ground samples. Samples with poor zero-shot performance produce consistently low and indistinguishable rewards during RL training, exhibiting no clear preference among multiple outputs and thus hindering learning efficiency. To address these challenges, we propose VideoTG-R1, a novel curriculum RL framework with reflected boundary annotations, enabling data-efficient training. Specifically, we propose a Boundary Reflection Agent that utilizes MLLMs to predict query-relevant timestamps outside the annotated intervals, allowing us to identify and filter out partially annotated samples, thereby reducing ambiguity. Furthermore, we introduce a Difficulty Estimation Agent to assess the training difficulty of each sample and design a curriculum RL strategy that dynamically masks the videos of hard-to-ground samples according to the training steps, easing the training difficulty and providing clearer preference. Experiments on the VTG and grounded VideoQA tasks demonstrate the effectiveness of our method. Remarkably, with only 10% of the training samples and 21% of the computational budget, VideoTG-R1 outperforms full-data counterparts under both group relative policy optimization (GRPO) and supervised fine-tuning (SFT). The code is available at https://github.com/ldong1111/VideoTG-R1.

Text-to-3D (T23D) generation has transformed digital content creation, yet remains bottlenecked by blind trial-and-error prompting processes that yield unpredictable results. While visual prompt engineering has advanced in text-to-image domains, its application to 3D generation presents unique challenges requiring multi-view consistency evaluation and spatial understanding. We present Sel3DCraft, a visual prompt engineering system for T23D that transforms unstructured exploration into a guided visual process. Our approach introduces three key innovations: a dual-branch structure combining retrieval and generation for diverse candidate exploration; a multi-view hybrid scoring approach that leverages MLLMs with innovative high-level metrics to assess 3D models with human-expert consistency; and a prompt-driven visual analytics suite that enables intuitive defect identification and refinement. Extensive testing and a user study demonstrate that Sel3DCraft surpasses other T23D systems in supporting creativity for designers.

Text-to-image (T2I) generation has made remarkable progress in producing high-quality images, but a fundamental challenge remains: creating backgrounds that naturally accommodate text placement without compromising image quality. This capability is non-trivial for real-world applications like graphic design, where clear visual hierarchy between content and text is essential. Prior work has primarily focused on arranging layouts within existing static images, leaving unexplored the potential of T2I models for generating text-friendly backgrounds. We present TextCenGen, a training-free approach that actively relocates objects before optimizing text regions, rather than directly reducing cross-attention which degrades image quality. Our method introduces: (1) a force-directed graph approach that detects conflicting objects and guides them relocation using cross-attention maps, and (2) a spatial attention constraint that ensures smooth background generation in text regions. Our method is plug-and-play, requiring no additional training while well balancing both semantic fidelity and visual quality. Evaluated on our proposed text-friendly T2I benchmark of 27,000 images across three seed datasets, TextCenGen outperforms existing methods by achieving 23\% lower saliency overlap in text regions while maintaining 98\% of the original semantic fidelity measured by CLIP score and our proposed Visual-Textual Concordance Metric (VTCM).

Color has been widely used to encode data in all types of visualizations. Effective color palettes contain discriminable and harmonious colors, which allow information from visualizations to be accurately and aesthetically conveyed. However, predefined color palettes not only lack the flexibility of custom color palette generation but also ignore the context in which the visualizations are used. Designing an effective color palette is a time-consuming and challenging process for users, even experts. In this work, we propose the generation of an image-based visualization color palette to exploit the human perception of visually appealing images while considering visualization cognition. By analyzing color palette constraints, including harmony, discrimination, and context, we propose an image-driven color generation method. We design a color clustering method in the saliency-hue plane based on visual importance detection and then select the palette based on the visualization color constraints. In addition, we design two color optimization and assignment strategies for visualizations of different data types. Evaluations through numeric indicators and user experiments demonstrate that the palettes predicted by our method are visually related to the original images and are aesthetically pleasing, supporting diverse visualization contexts and data types in practical applications.

PNG-8 images are commonly used on the web due to their small size, but their limited color palette often leads to dithering artifacts. Unfortunately, restoring these images using a conventional convolutional neural network (CNN) often results in suboptimal performance since the spatial distribution of dithering is not uniform across the image. This is because the convolutional operator is spatially consistent, meaning it applies the same kernel to all pixels, which we refer to as a global transformation. To address this issue, we propose PNG8IRNet, one approach that combines global and self-conditioned transformations to remove dithering artifacts. Our method incorporates a multilayer perceptron (MLP) to generate diverse kernels for each pixel, taking into account the spatial non-uniformity of dithering, which we define as a self-conditioned transformation. PNG8IRNet demonstrates its performance on multiple datasets, substantially enhancing visual comprehension through a comprehensive set of experiments.

Reference-based color transfer between images has been a fundamental function in image editing. However, existing approaches pay less attention to high-dynamic-range (HDR) images. It is worth noting that designing an appropriate representation for HDR images to achieve satisfying color transfer is challenging. In this paper, we propose an innovative high-dynamic-range image color transfer generative adversarial network (HDRCTGAN) to encode the original image into fine representations that allow transfer of the color of the reference image to the target image. We propose to learn fine representations through a generative adversarial network (GAN) in a self-supervised way. Particularly, the proposed method is self-supervised learning that requires only unlabeled HDR images instead of supervised learning that requires lots of ground truth pairs. HDRCTGAN consists of a generator to transfer the color of the reference image to the target image over the feature domain and a discriminator to suppress the artifacts caused by the generator. We also design a loss function to ensure that HDRCTGAN possesses two required properties: (a) high fidelity and (b) self-identity. The proposed approach yields a pleasing visual result. We have carried out HDR specific evaluations including both objective quantitative experiments with HDR metrics and subjective user studies operated on HDR display devices to demonstrate the effectiveness of our method. Furthermore, we have verified the applicability of the proposed approach to several applications, such as color transfer of HDR images captured by smartphones, color transfer of fabric images, and reference-based grayscale image colorization.

Recently, generative adversarial networks (GAN) have been widely used to solve image-to-image translation problems such as edges to photos, labels to scenes, and colorizing grayscale images. However, how to recover details of smoothed images is still unexplored. Naively training a GAN like pix2pix causes insufficiently perfect results due to the fact that we ignore two main characteristics including spatial variability and spatial correlation as for this problem. In this work, we propose DeSmoothGAN to utilize both characteristics specifically. The spatial variability indicates that the details of different areas of smoothed images are distinct and they are supposed to be recovered differently. Therefore, we propose to perform spatial feature-wise transformation to recover individual areas differently. The spatial correlation represents that the details of different areas are related to each other. Thus, we propose to apply full attention to consider the relations between them. The proposed method generates satisfying results on several real-world datasets. We have conducted quantitative experiments including smooth consistency and image similarity to demonstrate the effectiveness of DeSmoothGAN. Furthermore, ablation studies are performed to illustrate the usefulness of our proposed feature-wise transformation and full attention.

Abstract Image color editing techniques are of great significance for users who wish to adjust the image color. However, previous works paid less attention to the translucent images. In this paper, we propose a new method to recolor the translucent images while preserving detailed information and color relationships of the source image. We consider the recolor problem as a location transformation problem and solve it in two steps: automatic palette extraction and homography estimation. First, we propose the Hmeans method to extract the dominant colors of the source image based on histogram statistics and clustering. Then, we propose homography estimation to map the source colors to desired colors in the CIE‐LAB color space. Further, we adopt a non‐linear optimization approach to improve the result generated by the last step. The proposed method maintains high fidelity of the source image. Experiments have shown that our method generates a state‐of‐the‐art visual result, in particular in the shadow areas. The source images with ground truth generated by a ray tracer further verify the effectiveness of our method.