Direct Ascent Synthesis: Revealing Hidden Generative Capabilities in Discriminative Models

TL;DR

This paper reveals that discriminative models like CLIP have hidden generative abilities, demonstrated through a novel multi-resolution optimization method called DAS that synthesizes high-quality images without extra training.

Contribution

The paper introduces Direct Ascent Synthesis (DAS), a method to extract and utilize the latent generative capabilities of discriminative models, challenging the traditional separation between discriminative and generative architectures.

Findings

DAS produces high-quality, diverse images from discriminative models.

Discriminative models encode richer generative knowledge than previously thought.

Generated images maintain natural statistics and avoid adversarial patterns.

Abstract

We demonstrate that discriminative models inherently contain powerful generative capabilities, challenging the fundamental distinction between discriminative and generative architectures. Our method, Direct Ascent Synthesis (DAS), reveals these latent capabilities through multi-resolution optimization of CLIP model representations. While traditional inversion attempts produce adversarial patterns, DAS achieves high-quality image synthesis by decomposing optimization across multiple spatial scales (1x1 to 224x224), requiring no additional training. This approach not only enables diverse applications -- from text-to-image generation to style transfer -- but maintains natural image statistics ($1/f^2$ spectrum) and guides the generation away from non-robust adversarial patterns. Our results demonstrate that standard discriminative models encode substantially richer generative knowledge than previously recognized, providing new perspectives on model interpretability and the relationship between adversarial examples and natural image synthesis.