EchoGen: Generating Visual Echoes in Any Scene via Feed-Forward Subject-Driven Auto-Regressive Model

2025-09-30

AI Summary

Existing subject-driven generation methods face a fundamental trade-off between efficiency and zero-shot capability: fine-tuning paradigms are computationally expensive, while diffusion-based approaches suffer from slow inference. This paper introduces the first feed-forward subject-driven generation framework based on Vision Autoregressive (VAR) modeling. Its core innovation is a dual-path injection mechanism that decouples high-level semantic identity from low-level detailed features, implemented via a semantic/content dual-encoder architecture. This design integrates decoupled cross-attention and multimodal attention to enable precise, disentangled control over identity and appearance. The method supports zero-shot transfer without retraining, achieving image quality and subject fidelity comparable to state-of-the-art diffusion models while drastically reducing sampling latency. Extensive quantitative and qualitative experiments demonstrate its superior efficiency, fine-grained controllability, and strong generalization across diverse subjects and poses.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Subject-driven generation is a critical task in creative AI; yet current state-of-the-art methods present a stark trade-off. They either rely on computationally expensive, per-subject fine-tuning, sacrificing efficiency and zero-shot capability, or employ feed-forward architectures built on diffusion models, which are inherently plagued by slow inference speeds. Visual Auto-Regressive (VAR) models are renowned for their rapid sampling speeds and strong generative quality, making them an ideal yet underexplored foundation for resolving this tension. To bridge this gap, we introduce EchoGen, a pioneering framework that empowers VAR models with subject-driven generation capabilities. The core design of EchoGen is an effective dual-path injection strategy that disentangles a subject's high-level semantic identity from its low-level fine-grained details, enabling enhanced controllability and fidelity. We employ a semantic encoder to extract the subject's abstract identity, which is injected through decoupled cross-attention to guide the overall composition. Concurrently, a content encoder captures intricate visual details, which are integrated via a multi-modal attention mechanism to ensure high-fidelity texture and structural preservation. To the best of our knowledge, EchoGen is the first feed-forward subject-driven framework built upon VAR models. Both quantitative and qualitative results substantiate our design, demonstrating that EchoGen achieves subject fidelity and image quality comparable to state-of-the-art diffusion-based methods with significantly lower sampling latency. Code and models will be released soon.

Problem

• Enables subject-driven generation without fine-tuning
• Resolves trade-off between generation quality and speed
• Disentangles semantic identity from visual details for control

Innovation

• Feed-forward subject-driven auto-regressive model for visual echoes
• Dual-path injection disentangles semantic identity and fine details
• Multi-modal attention mechanism ensures high-fidelity texture preservation