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.

Everyday photos often capture meaningful moments, but the final image may still look ordinary because the camera angle, composition, or pose was not ideal when the shutter was pressed. Most photo-editing tools focus on surface-level changes, such as adjusting color, lighting, sharpness, or facial details. However, they usually cannot fix deeper photographic problems. We study how AI can turn such ordinary photos into more aesthetically pleasing images through content-level photo reconstruction. Our system, AesFormer, works in two steps. First, it analyzes the photo and plans useful photographic changes, such as improving the composition, camera viewpoint, or pose. Then, it uses this plan to guide an image editor to reconstruct the photo accordingly. To train and evaluate this system, we built a large collection of matched “before” and “after” examples from photography tutorial videos, where ordinary shots are improved into more appealing final photos. This work helps AI photo editing move beyond simple retouching toward reconstructing photos in a way that better reflects photographic aesthetics.