Language-Grounded Multi-Domain Image Translation via Semantic Difference Guidance

TL;DR

LACE is a novel framework that enables precise, multi-attribute, language-guided image translation across multiple domains by explicitly grounding semantic differences into visual transformations.

Contribution

LACE introduces a dual-component system combining semantic-structural fusion and explicit semantic delta grounding for improved multi-domain image translation.

Findings

Achieves high visual fidelity and structural preservation.

Provides interpretable, attribute-specific control.

Surpasses prior methods on benchmark datasets.

Abstract

Multi-domain image-to-image translation re quires grounding semantic differences ex pressed in natural language prompts into corresponding visual transformations, while preserving unrelated structural and seman tic content. Existing methods struggle to maintain structural integrity and provide fine grained, attribute-specific control, especially when multiple domains are involved. We propose LACE (Language-grounded Attribute Controllable Translation), built on two compo nents: (1) a GLIP-Adapter that fuses global semantics with local structural features to pre serve consistency, and (2) a Multi-Domain Control Guidance mechanism that explicitly grounds the semantic delta between source and target prompts into per-attribute translation vec tors, aligning linguistic semantics with domain level visual changes. Together, these modules enable compositional multi-domain control with independent strength modulation for each attribute. Experiments on CelebA(Dialog) and BDD100K demonstrate that LACE achieves high visual fidelity, structural preservation, and interpretable domain-specific control, surpass ing prior baselines. This positions LACE as a cross-modal content generation framework bridging language semantics and controllable visual translation.