SnapGen++: Unleashing Diffusion Transformers for Efficient High-Fidelity Image Generation on Edge Devices

TL;DR

This paper introduces SnapGen++, a diffusion transformer framework optimized for edge devices that maintains high image quality while significantly reducing computational and memory requirements through novel architecture, training, and distillation techniques.

Contribution

The paper presents a compact DiT architecture with sparse attention, an elastic training framework for multi-capacity models, and a knowledge-guided distillation method for efficient high-fidelity image generation.

Findings

Achieves transformer-level quality with minimal resource usage.

Enables real-time image generation on mobile devices.

Outperforms existing models in efficiency and quality trade-offs.

Abstract

Recent advances in diffusion transformers (DiTs) have set new standards in image generation, yet remain impractical for on-device deployment due to their high computational and memory costs. In this work, we present an efficient DiT framework tailored for mobile and edge devices that achieves transformer-level generation quality under strict resource constraints. Our design combines three key components. First, we propose a compact DiT architecture with an adaptive global-local sparse attention mechanism that balances global context modeling and local detail preservation. Second, we propose an elastic training framework that jointly optimizes sub-DiTs of varying capacities within a unified supernetwork, allowing a single model to dynamically adjust for efficient inference across different hardware. Finally, we develop Knowledge-Guided Distribution Matching Distillation, a step-distillation pipeline that integrates the DMD objective with knowledge transfer from few-step teacher models, producing high-fidelity and low-latency generation (e.g., 4-step) suitable for real-time on-device use. Together, these contributions enable scalable, efficient, and high-quality diffusion models for deployment on diverse hardware.