Reusable personalized color editing method adapts photo look quickly
PrefLUT: Reusable and Refinable Personalized Color Editing from Pairwise Preferences
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
People often like different color edits on the same photo, but usual editing tools target just one fixed look. The authors created PrefLUT, which learns a user’s color preferences from pairs of images they like or dislike, stores these in a tiny profile, and reuses or updates it easily to edit new photos. This approach works fast and efficiently on powerful GPUs and can create standard color lookup tables used in photo editing software. The authors also designed a test to make sure the edits truly reflect user preferences and the pictures involved.
What this means in practice
- •For photo app developers: Provide personalized color filters that update efficiently based on user feedback without retraining models for each person.$Commercial implications: Enables customizable photo editing products that adapt quickly to users’ changing color preferences, increasing user satisfaction and product differentiation.
- •For video editors: Apply consistent personalized color grading across multiple clips using reusable LUTs derived from user preferences.
Authors
Chuanzhi Xu, Langyi Chen, Chengkun Yue, Xuanhua Yin, Boyu Wei, Qingwen Zeng, Zihan Deng, Weidong Cai
Abstract
Photographic color editing is inherently personal: the same image can appear too warm, too muted, or already satisfactory to different users. Most lookup table (LUT) and reference-guided methods target a specified appearance rather than model persistent preferences from repeated user choices. To address this gap, we introduce PrefLUT, a reusable and refinable user-preference modeling framework for deployable 3D LUTs, encoding ordered preferred/non-preferred image pairs into a lightweight Reusable User Profile that is reused across queries and refined using additional user preference pairs, without per-user optimization. A Query-Conditioned LUT Predictor combines this profile with each image to predict a LUT latent vector and edit strength. An Identity-Residual LUT Decoder and Edit-Strength Controller then produce an exportable 3D LUT. Experiments on three datasets demonstrate effective personalized editing and general-purpose enhancement. Each quantized profile requires only 260 bytes, and editing takes 1.365 ms/image on an RTX 5090 GPU. We also introduce the Preference-Conditioning Verification Protocol (PCVP), an evaluation protocol to verify whether personalized image edits depend on user preferences and the query image through controlled changes to user profiles, preference orders, pair correspondences, and query images.