arXiv subjects

Tianxiang Du

Publications and source records attributed to Tianxiang Du.

4 recordsLinked to original sources

arXiv2026

AesFormer: Transform Everyday Photos into Beautiful Memories

In everyday photography, aesthetically appealing moments are often captured with structural flaws (e.g., composition, camera viewpoint, or pose) that existing retouching and portrait enhancement methods cannot fix. We formulate Aesthetic Photo Reconstruction (APR) as improving a photo's aesthetic quality via structural reconstruction while preserving subject identity and scene semantics. Although recent advances in image editing models make APR feasible, they often lack aesthetic understanding, yielding edits that are semantically plausible yet aesthetically weak. To address this, we propose AesFormer, a two-stage framework that decouples aesthetic planning from image editing. In Stage 1, an aesthetic action model (AesThinker) analyzes the input along seven progressive photographic dimensions and outputs executable editing actions; we further apply GRPO-A to encourage broad exploration over diverse action plans beyond SFT. In Stage 2, an action-conditioned editor (AesEditor) performs structural edits guided by these actions. To support APR, we build a video-based corpus-mining pipeline (VCMP) and construct AesRecon, a benchmark of 9,071 strictly aligned (poor, good) image pairs. Experiments show that AesFormer substantially improves APR performance and is competitive with Nano Banana Pro. Code is available at https://github.com/PKU-ICST-MIPL/AesFormer_ICML2026.

cs.CV↗

arXiv2026

Venus: Benchmarking and Empowering Multimodal Large Language Models for Aesthetic Guidance and Cropping

The widespread use of smartphones has made photography ubiquitous, yet a clear gap remains between ordinary users and professional photographers, who can identify aesthetic issues and provide actionable shooting guidance during capture. We define this capability as aesthetic guidance (AG) -- an essential but largely underexplored domain in computational aesthetics. Existing multimodal large language models (MLLMs) primarily offer overly positive feedback, failing to identify issues or provide actionable guidance. Without AG capability, they cannot effectively identify distracting regions or optimize compositional balance, thus also struggling in aesthetic cropping, which aims to refine photo composition through reframing after capture. To address this, we introduce AesGuide, the first large-scale AG dataset and benchmark with 10,748 photos annotated with aesthetic scores, analyses, and guidance. Building upon it, we propose Venus, a two-stage framework that first empowers MLLMs with AG capability through progressively complex aesthetic questions and then activates their aesthetic cropping power via CoT-based rationales. Extensive experiments show that Venus substantially improves AG capability and achieves state-of-the-art (SOTA) performance in aesthetic cropping, enabling interpretable and interactive aesthetic refinement across both stages of photo creation. Code is available at https://github.com/PKU-ICST-MIPL/Venus_CVPR2026.

cs.CV↗

arXiv2024

A Novel Design of Capacitive Plasmonic Near Field Transducer

A near field transducer (NFT) is a key photonics component in heat assisted magnetic recording (HAMR) for the localized heating of the magnetic medium. In this work, we present a novel NFT design through capacitive coupling. In our design, tapered metal bars separated by thin dielectric materials with gap distance G are used to create the plasmonic resonance and focus the electromagnetic field. The design is motivated by the intention to improve thermal stability, which can be achieved through segmentation using thermally stable dielectric material between the plasmonic metal bars. Using COMSOL Multiphysics software, the performance of this capacitive-coupled NFT is systematically modeled. It is shown that the electromagnetic field could gradually be focused through the tapering towards the air bearing surface (ABS). In addition, the focusing effect could be enhanced with a smaller NFT peg size at the resonant wavelength. The material selection for the NFT tip material will be discussed to further address the thermal stability of the device. In conclusion, this capacitive-coupled NFT with dielectric separation gaps and tapering yields an enhanced |E| field intensity at the tip with the potential for an enhanced material thermal stability. Such a design can also exhibit applications in other energy delivery systems as well as plasmonic waveguides and sensors.

physics.optics↗

arXiv2022

Understanding the growth of high-aspect-ratio grains in granular L10-FePt thin-film magnetic media

A systematic investigation has been performed to optimize the microstructure of $\mathrm{L1_0-FePt-SiO_x}$ granular thin film as recording media for heat-assisted magnetic recording. The FePt-BN nucleation layer, which is stable even at $700^\circ \text{C}$, is used to control the grain sizes and microstructure during high-temperature processing. The study finds that films of high-aspect-ratio FePt grains with well-formed silicon oxide (SiOx) grain boundaries require the grading of the deposition temperature during film growth as well as the grading of the silicon oxide concentration. Well-isolated columnar grains of $\mathrm{L1_0-FePt}$ with an average height greater than 11 nm and diameters less than 7 nm have been achieved. Transmission electron microscopy (TEM) analysis of the microstructures of samples produced under a variety of non-optimal conditions is presented to show how the microstructure of the films depends on each of the sputtering parameters.

cond-mat.mtrl-sci↗