AcFlow: Controlling Text-to-Image Diffusion Transformers via Learned Conditional Activation Flow

Nove1yst, Author2, Author3, Author4, Author5

multimodalfine-tuningimage generation

2609.10723

Builder Relevance

80%

4h ago

Abstract

AcFlow introduces an inference-time controller for text-to-image diffusion transformers that enhances control over style intensity and suppresses unwanted concepts.

Reality Card

Core Claim

AcFlow achieves the best style-content trade-off among evaluated baselines, significantly improving style alignment while allowing for fine-grained control over image generation.

Method / Result

AcFlow attains a style-content alignment score of 0.5365/0.2860, outperforming the best baseline score of 0.4397/0.2684.

Limitations

The method may require extensive training data for optimal performance and may not generalize well to all unseen concepts.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.

← Back to all papers