Adaptive Fused Prior Transfer for Controllable Generative Image Compression
AFP-GIC transfers adaptive fused priors from a frozen pretrained model to guide generative compression without transmitting them, reducing decoder latency and parameters while improving very-low-bitrate naturalness.
Published May 16, 2026▲ 4 on Hugging FaceCode ★ 4arXiv ↗
Only vote on papers you've read. Sign in with GitHub to vote.
AFP-GIC delivers a real theorem, 20% fewer parameters, and 18% lower latency over DC-VIC via adaptive fused priors, yet its competitive PSNR/SSIM and missing MS-SSIM leave very-low-bitrate gains unproven, while unfrozen AdaCode domain-shift risks and hidden…
Abstract
At very low bitrates, image compression discards fine textures and local structures, while distortion-oriented reconstruction often produces over-smoothed images. Generative codecs synthesize missing details, but existing codebook-based controllable designs generally rely on single-codebook reconstruction priors. We propose Adaptive Fused Prior Transfer for Controllable Generative Image Compression (AFP-GIC), which transfers an image-adaptive fused prior from a frozen pretrained AdaCode model. Encoder-side prior features guide latent formation, while the decoder predicts a compatible fused prior from the compressed representation and control variables, without transmitting the prior itself. A motivating analysis shows that better decoder-side prior alignment tightens a reconstruction-error upper bound and that the fused-prior family includes single-codebook choices as special cases. A single pretrained model supports five evaluated bitrate operating points. Under the unified benchmark, AFP-GIC achieves 18.1% lower decoder latency and uses 31.10 million (20.5%) fewer inference parameters than DC-VIC. Experiments on Kodak, CLIC2020, and DIV2K show competitive PSNR and SSIM, with the clearest naturalness gains in NIQE scores and very-low-bitrate visual comparisons. Code: https://github.com/yifeipet/AFP_GIC.