Line Search-Based Feature Transformation for Fast, Stable, and Tunable Content-Style Control in Photorealistic Style Transfer

TL;DR

This paper introduces a novel line search-based feature transformation for photorealistic style transfer that enhances control, speed, and stability across various models, enabling better content-style balance.

Contribution

A new general-purpose transformation method that improves control, speed, and stability in photorealistic style transfer models.

Findings

Outperforms existing transformations in speed and stability

Provides consistent control over content and style balance

Demonstrates effectiveness across multiple style transfer models

Abstract

Photorealistic style transfer is the task of synthesizing a realistic-looking image when adapting the content from one image to appear in the style of another image. Modern models commonly embed a transformation that fuses features describing the content image and style image and then decodes the resulting feature into a stylized image. We introduce a general-purpose transformation that enables controlling the balance between how much content is preserved and the strength of the infused style. We offer the first experiments that demonstrate the performance of existing transformations across different style transfer models and demonstrate how our transformation performs better in its ability to simultaneously run fast, produce consistently reasonable results, and control the balance between content and style in different models. To support reproducing our method and models, we share the code at https://github.com/chiutaiyin/LS-FT.