The Unreasonable Effectiveness of Text Embedding Interpolation for Continuous Image Steering

TL;DR

This paper introduces a training-free, continuous image editing method using text embedding interpolation, enabling smooth control over generated images without additional training or manual intervention.

Contribution

It proposes a novel, training-free framework that leverages text embedding interpolation for continuous, controllable image editing in text-conditioned generative models.

Findings

Method achieves smooth, continuous edits comparable to training-based approaches.

It outperforms other training-free methods in semantic control quality.

The approach generalizes across image and video generation modalities.

Abstract

We present a training-free framework for continuous and controllable image editing at test time for text-conditioned generative models. In contrast to prior approaches that rely on additional training or manual user intervention, we find that a simple steering in the text-embedding space is sufficient to produce smooth edit control. Given a target concept (e.g., enhancing photorealism or changing facial expression), we use a large language model to automatically construct a small set of debiased contrastive prompt pairs, from which we compute a steering vector in the generator's text-encoder space. We then add this vector directly to the input prompt representation to control generation along the desired semantic axis. To obtain a continuous control, we propose an elastic range search procedure that automatically identifies an effective interval of steering magnitudes, avoiding both under-steering (no-edit) and over-steering (changing other attributes). Adding the scaled versions of the same vector within this interval yields smooth and continuous edits. Since our method modifies only textual representations, it naturally generalizes across text-conditioned modalities, including image and video generation. To quantify the steering continuity, we introduce a new evaluation metric that measures the uniformity of semantic change across edit strengths. We compare the continuous editing behavior across methods and find that, despite its simplicity and lightweight design, our approach is comparable to training-based alternatives, outperforming other training-free methods.