CoD-Lite: Real-Time Diffusion-Based Generative Image Compression

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jia, Zhaoyang, Xue, Naifu, Zheng, Zihan, Li, Jiahao, Li, Bin, Zhang, Xiaoyi, Guo, Zongyu, Zhang, Yuan, Li, Houqiang, Lu, Yan
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914473167552512
author Jia, Zhaoyang
Xue, Naifu
Zheng, Zihan
Li, Jiahao
Li, Bin
Zhang, Xiaoyi
Guo, Zongyu
Zhang, Yuan
Li, Houqiang
Lu, Yan
author_facet Jia, Zhaoyang
Xue, Naifu
Zheng, Zihan
Li, Jiahao
Li, Bin
Zhang, Xiaoyi
Guo, Zongyu
Zhang, Yuan
Li, Houqiang
Lu, Yan
contents Recent advanced diffusion methods typically derive strong generative priors by scaling diffusion transformers. However, scaling fails to generalize when adapted for real-time compression scenarios that demand lightweight models. In this paper, we explore the design of real-time and lightweight diffusion codecs by addressing two pivotal questions. First, does diffusion pre-training benefit lightweight diffusion codecs? Through systematic analysis, we find that generation-oriented pre-training is less effective at small model scales whereas compression-oriented pre-training yields consistently better performance. Second, are transformers essential? We find that while global attention is crucial for standard generation, lightweight convolutions suffice for compression-oriented diffusion when paired with distillation. Guided by these findings, we establish a one-step lightweight convolution diffusion codec that achieves real-time $60$~FPS encoding and $42$~FPS decoding at 1080p. Further enhanced by distillation and adversarial learning, the proposed codec reduces bitrate by 85\% at a comparable FID to MS-ILLM, bridging the gap between generative compression and practical real-time deployment. Codes are released at https://github.com/microsoft/GenCodec/tree/main/CoD_Lite
format Preprint
id arxiv_https___arxiv_org_abs_2604_12525
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle CoD-Lite: Real-Time Diffusion-Based Generative Image Compression
Jia, Zhaoyang
Xue, Naifu
Zheng, Zihan
Li, Jiahao
Li, Bin
Zhang, Xiaoyi
Guo, Zongyu
Zhang, Yuan
Li, Houqiang
Lu, Yan
Computer Vision and Pattern Recognition
Recent advanced diffusion methods typically derive strong generative priors by scaling diffusion transformers. However, scaling fails to generalize when adapted for real-time compression scenarios that demand lightweight models. In this paper, we explore the design of real-time and lightweight diffusion codecs by addressing two pivotal questions. First, does diffusion pre-training benefit lightweight diffusion codecs? Through systematic analysis, we find that generation-oriented pre-training is less effective at small model scales whereas compression-oriented pre-training yields consistently better performance. Second, are transformers essential? We find that while global attention is crucial for standard generation, lightweight convolutions suffice for compression-oriented diffusion when paired with distillation. Guided by these findings, we establish a one-step lightweight convolution diffusion codec that achieves real-time $60$~FPS encoding and $42$~FPS decoding at 1080p. Further enhanced by distillation and adversarial learning, the proposed codec reduces bitrate by 85\% at a comparable FID to MS-ILLM, bridging the gap between generative compression and practical real-time deployment. Codes are released at https://github.com/microsoft/GenCodec/tree/main/CoD_Lite
title CoD-Lite: Real-Time Diffusion-Based Generative Image Compression
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.12525