Taming Diffusion Transformer for Efficient Mobile Video Generation in Seconds

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Wu, Yushu, Li, Yanyu, Kag, Anil, Skorokhodov, Ivan, Menapace, Willi, Ma, Ke, Sahni, Arpit, Hu, Ju, Siarohin, Aliaksandr, Sagar, Dhritiman, Wang, Yanzhi, Tulyakov, Sergey
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908567800381440
author Wu, Yushu
Li, Yanyu
Kag, Anil
Skorokhodov, Ivan
Menapace, Willi
Ma, Ke
Sahni, Arpit
Hu, Ju
Siarohin, Aliaksandr
Sagar, Dhritiman
Wang, Yanzhi
Tulyakov, Sergey
author_facet Wu, Yushu
Li, Yanyu
Kag, Anil
Skorokhodov, Ivan
Menapace, Willi
Ma, Ke
Sahni, Arpit
Hu, Ju
Siarohin, Aliaksandr
Sagar, Dhritiman
Wang, Yanzhi
Tulyakov, Sergey
contents Diffusion Transformers (DiT) have shown strong performance in video generation tasks, but their high computational cost makes them impractical for resource-constrained devices like smartphones, and practical on-device generation is even more challenging. In this work, we propose a series of novel optimizations to significantly accelerate video generation and enable practical deployment on mobile platforms. First, we employ a highly compressed variational autoencoder (VAE) to reduce the dimensionality of the input data without sacrificing visual quality. Second, we introduce a KD-guided, sensitivity-aware tri-level pruning strategy to shrink the model size to suit mobile platforms while preserving critical performance characteristics. Third, we develop an adversarial step distillation technique tailored for DiT, which allows us to reduce the number of inference steps to four. Combined, these optimizations enable our model to achieve approximately 15 frames per second (FPS) generation speed on an iPhone 16 Pro Max, demonstrating the feasibility of efficient, high-quality video generation on mobile devices.
format Preprint
id arxiv_https___arxiv_org_abs_2507_13343
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Taming Diffusion Transformer for Efficient Mobile Video Generation in Seconds
Wu, Yushu
Li, Yanyu
Kag, Anil
Skorokhodov, Ivan
Menapace, Willi
Ma, Ke
Sahni, Arpit
Hu, Ju
Siarohin, Aliaksandr
Sagar, Dhritiman
Wang, Yanzhi
Tulyakov, Sergey
Computer Vision and Pattern Recognition
Image and Video Processing
Diffusion Transformers (DiT) have shown strong performance in video generation tasks, but their high computational cost makes them impractical for resource-constrained devices like smartphones, and practical on-device generation is even more challenging. In this work, we propose a series of novel optimizations to significantly accelerate video generation and enable practical deployment on mobile platforms. First, we employ a highly compressed variational autoencoder (VAE) to reduce the dimensionality of the input data without sacrificing visual quality. Second, we introduce a KD-guided, sensitivity-aware tri-level pruning strategy to shrink the model size to suit mobile platforms while preserving critical performance characteristics. Third, we develop an adversarial step distillation technique tailored for DiT, which allows us to reduce the number of inference steps to four. Combined, these optimizations enable our model to achieve approximately 15 frames per second (FPS) generation speed on an iPhone 16 Pro Max, demonstrating the feasibility of efficient, high-quality video generation on mobile devices.
title Taming Diffusion Transformer for Efficient Mobile Video Generation in Seconds
topic Computer Vision and Pattern Recognition
Image and Video Processing
url https://arxiv.org/abs/2507.13343