Search papers, labs, and topics across Lattice.
This paper introduces VIGOR-his, a novel decade-spanning cross-view dataset of 43,653 location-level quadruplets across 11 cities, designed to address gaps in historical street-view imagery. They propose CrossTimeEdit, a generative model that treats historical street-view synthesis as an editing task, leveraging recent street views for scene constraints and temporal satellite differences as change evidence. Trained with a multi-reward reinforcement learning objective (IA, BP, QP) optimized via Flow-GRPO and GDPO, CrossTimeEdit achieves a 17.12% improvement over baselines in scene consistency, realism, and perceptual quality.
Synthesizing historical street views is now possible with 17% better realism and consistency by treating it as a reward-guided editing problem.
Historical street-view imagery records urban evolution, but uneven coverage leaves substantial gaps in historical records. Generating plausible past appearances requires restoring changed structures while preserving persistent scene content. We construct VIGOR-his, a decade-spanning cross-view dataset containing 43,653 location-level quadruplets across 11 cities on three continents. Its automated pipeline performs spatial pairing, consistency screening, change classification, and the generation and validation of satellite-based change descriptions and local editing instructions. Based on VIGOR-his, we propose CrossTimeEdit, a model that reformulates historical street-view generation as editing, using recent street views to constrain viewpoint and unchanged appearance and temporal satellite differences as change evidence. Starting from FLUX.2 [Klein] 4B, we train CrossTimeEdit through supervised fine-tuning (SFT) followed by online reinforcement learning (RL). We design three street-view editing criteria, namely Instruction Alignment (IA), Background Preservation (BP), and Quality and Physical Plausibility (QP), as both RL reward dimensions and evaluation metrics. We optimize this multi-reward objective using Within Group Relative Policy Optimization for flow-matching models (Flow-GRPO) with Group reward-Decoupled Normalization Policy Optimization (GDPO), which normalizes each reward dimension before aggregation. CrossTimeEdit improves overall performance across the three editing criteria by 17.12\% over the pretrained baseline and outperforms cross-view generation models in scene consistency, visual realism, and perceptual quality. The implementation code, dataset, and model weights are available at https://luhanwen67.github.io/CrossTimeEdit-release/.
