PixelSmile: Toward Fine-Grained Facial Expression Editing

TL;DR

PixelSmile introduces a diffusion-based framework for fine-grained facial expression editing, leveraging a new dataset and evaluation benchmarks to achieve precise, controllable, and identity-preserving expression manipulation.

Contribution

The paper presents PixelSmile, a novel diffusion model that disentangles expression semantics using symmetric training, combined with a new dataset and benchmarks for evaluation.

Findings

Achieves superior expression disentanglement and identity preservation.

Enables precise linear control of expressions via textual latent interpolation.

Supports smooth and controllable expression blending.

Abstract

Fine-grained facial expression editing has long been limited by intrinsic semantic overlap. To address this, we construct the Flex Facial Expression (FFE) dataset with continuous affective annotations and establish FFE-Bench to evaluate structural confusion, editing accuracy, linear controllability, and the trade-off between expression editing and identity preservation. We propose PixelSmile, a diffusion framework that disentangles expression semantics via fully symmetric joint training. PixelSmile combines intensity supervision with contrastive learning to produce stronger and more distinguishable expressions, achieving precise and stable linear expression control through textual latent interpolation. Extensive experiments demonstrate that PixelSmile achieves superior disentanglement and robust identity preservation, confirming its effectiveness for continuous, controllable, and fine-grained expression editing, while naturally supporting smooth expression blending.