Fast high-resolution image synthesis with latent adversarial diffusion distillation

A Sauer, F Boesel, T Dockhorn, A Blattmann… - SIGGRAPH Asia 2024 …, 2024 - dl.acm.org
Diffusion models are the main driver of progress in image and video synthesis, but suffer
from slow inference speed. Distillation methods, like the recently introduced adversarial …

Diffusion models and representation learning: A survey

M Fuest, P Ma, M Gui, JS Fischer, VT Hu… - arxiv preprint arxiv …, 2024 - arxiv.org
Diffusion Models are popular generative modeling methods in various vision tasks, attracting
significant attention. They can be considered a unique instance of self-supervised learning …

Osv: One step is enough for high-quality image to video generation

X Mao, Z Jiang, FY Wang, W Zhu, J Zhang… - arxiv preprint arxiv …, 2024 - arxiv.org
Video diffusion models have shown great potential in generating high-quality videos,
making them an increasingly popular focus. However, their inherent iterative nature leads to …

Videoagent: Self-improving video generation

A Soni, S Venkataraman, A Chandra… - arxiv preprint arxiv …, 2024 - arxiv.org
Video generation has been used to generate visual plans for controlling robotic systems.
Given an image observation and a language instruction, previous work has generated video …

Flow generator matching

Z Huang, Z Geng, W Luo, G Qi - arxiv preprint arxiv:2410.19310, 2024 - arxiv.org
In the realm of Artificial Intelligence Generated Content (AIGC), flow-matching models have
emerged as a powerhouse, achieving success due to their robust theoretical underpinnings …

From slow bidirectional to fast causal video generators

T Yin, Q Zhang, R Zhang, WT Freeman… - arxiv preprint arxiv …, 2024 - arxiv.org
Current video diffusion models achieve impressive generation quality but struggle in
interactive applications due to bidirectional attention dependencies. The generation of a …

Adversarial diffusion compression for real-world image super-resolution

B Chen, G Li, R Wu, X Zhang, J Chen, J Zhang… - arxiv preprint arxiv …, 2024 - arxiv.org
Real-world image super-resolution (Real-ISR) aims to reconstruct high-resolution images
from low-resolution inputs degraded by complex, unknown processes. While many Stable …

Nitrofusion: High-fidelity single-step diffusion through dynamic adversarial training

DY Chen, H Bandyopadhyay, K Zou… - arxiv preprint arxiv …, 2024 - arxiv.org
We introduce NitroFusion, a fundamentally different approach to single-step diffusion that
achieves high-quality generation through a dynamic adversarial framework. While one-step …

Multistep Distillation of Diffusion Models via Moment Matching

T Salimans, T Mensink, J Heek… - arxiv preprint arxiv …, 2024 - arxiv.org
We present a new method for making diffusion models faster to sample. The method distills
many-step diffusion models into few-step models by matching conditional expectations of …

Stable Consistency Tuning: Understanding and Improving Consistency Models

FY Wang, Z Geng, H Li - arxiv preprint arxiv:2410.18958, 2024 - arxiv.org
Diffusion models achieve superior generation quality but suffer from slow generation speed
due to the iterative nature of denoising. In contrast, consistency models, a new generative …