We propose FluxC, a zero-shot codec that uses Flux as a compression prior. Although DiffC has shown that pretrained diffusion models can serve as zero-shot codecs, we ask how the choice of coding path and reconstruction method affects rate--distortion--perception behavior. We analyze FluxC along three axes: coding cost and bit allocation, reconstruction method, and low rate distortion. At matched signal-to-noise ratio (SNR), we show that the path choice does not affect the ideal information rate, but changes how bits are allocated across time steps. For reconstruction, we compare ancestral sampling, one-step ODE reconstruction, and many-step ODE reconstruction. Our analysis shows that ODE-based reconstruction can achieve lower distortion than ancestral sampling, especially in the low rate regime. Experiments on Gaussian sources and natural images support this prediction. Moreover, FluxC performs competitively with, and in some cases outperforms, recent neural image compression methods, suggesting that Flux can serve as a principled zero-shot codec.
Figures & tables
Figure 1
Figure 1: (Top) Rate as a function of the path time t on three datasets. (Bottom) Rate as a function of SNR on a logarithmic scale.
Figure 2: Incremental rate as a function of t . The incremental rate denotes the additional coding cost assigned to each coding step, i.e., the increase in rate between adjacent time points.
Figure 3: Signed MSE change relative to the corresponding zero rate reconstruction in the ultra low rate regime ( R≤10−3 bpp). More negative values indicate larger distortion reductions.
Figure 4: Rate–distortion and rate–perception curves on natural images. We report PSNR, LPIPS, and MS-SSIM as distortion and FID as perception on a logarithmic rate axis.
Figure 5: Qualitative reconstructions at approximately 0.01 bpp. The examples compare MS-ILLM ( Muckley et al., 2023 ) , DDCM ( Ohayon et al., 2025 ) , DiffC and FluxC variants
Appendix figures & tables8 assets
Supplementary material from the paper’s appendix.
Appendix
Figure 6: Experiments on isotropic Gaussian source. Lower D/d and lower P/d are better.
Figure 7: Experiments on Gaussian source across mild power-law, heavy-tailed power-law, and low rank spectra.
Dataset
Method
Enc steps
Encode
Decode
Kodak
DiffC SD1.5
50
39.41
72.82
Kodak
DiffC SD2.1
50
38.27
69.81
Kodak
DiffC Flux
46
36.00
517.08
Kodak
FluxC-1
46
27.35
47.91
Kodak
FluxC-ODE
46
35.05
574.51
DIV2K
DiffC SD1.5
50
40.10
73.14
Appendix
Table 1: Encode and decode time on 512 × 512 images across three datasets. Times are seconds per image. Encode and decode denote the average processing times per image. Bold indicates the lowest runtime for each dataset.
Figure 8: Estimation MSE of x0 -prediction as a function of SNR for Flux, SD 1.5, and SD 2.1 on each dataset.
Dataset
Method
Enc steps
Encode
Decode
Kodak
FluxC-ODE
16
11.79
99.14
Kodak
FluxC-ODE
6
4.57
26.59
Kodak
FluxC-ODE
4
5.09
9.65
Kodak
FluxC-1
16
11.67
18.48
Kodak
FluxC-1
6
4.45
7.69
Kodak
FluxC-1
4
5.08
5.09
Appendix
Table 2: FluxC encoding and decoding time (s/image) on average. Bold indicates the lowest runtime for each dataset.
Figure 9: Rate–distortion and rate–perception curves for FluxC with nominal encoding schedules of 50, 20, 10, and 5 steps.
Figure 10: Additional qualitative reconstructions at approximately 0.003 bpp compared with PerCo, DDCM, and DiffC variants and our FluxC variants.
Figure 11: Additional qualitative reconstructions at method-specific rates near the 0.02 bpp operating region.
Diffusion-based image compression has achieved strong perceptual quality at ultra-low bitrates. However, existing codecs are often tied to specific backbones and specialized components, making diverse, rapidly evolving generative models difficult to reuse. This raises a natural question: Can modern generative foundation models be connected to image compression through a simple and extensible interface? Two insights guide our design: stronger generative priors make a simpler codec interface viable, and generation and compression can be intrinsically linked through latent transport. We therefore propose Gen2-IC with two stages: (1) Latent Compression maps clean image latents to entropy-constrained latents; and (2) Latent Transport refines them with one near-terminal update based on the pretrained model. Gen2-IC requires neither auxiliary conditioning signals nor task-specific backbone modifications. With lightweight adaptation and no distillation, it supports fast encoding and one-step decoding across multiple bitrates. We validate Gen2-IC on SD-2.1, SANA-1.5, FLUX.1-dev, and Qwen-Image-2512, spanning U-Net and Transformer architectures as well as diffusion and flow-matching formulations. With stronger priors, Gen2-IC delivers gains below 0.05 bpp: the Qwen variant leads diffusion-based generative codecs in reconstruction fidelity (PSNR), perceptual similarity (LPIPS and DISTS), and recognizer-based semantic fidelity (OCR CER/WER and face-ROI similarity) across four benchmarks.
DiffC provides a principled way to reuse pre-trained diffusion models for lossy compression, but its encoding and decoding procedures remain slow because they require many discretized forward and reverse steps. We study whether few-step generative models -- Rectified Flow, Consistency Trajectory Models (CTM), and MeanFlow -- can be cast as codecs within the same reverse channel coding (RCC) framework. The main challenge is that RCC requires posterior and shared distribution parameters, whereas these models do not explicitly parameterize intermediate conditional distributions. For Rectified Flow and MeanFlow, we use the equivalence between velocity parameterization and diffusion-style denoising parameterization to derive the quantities required by RCC. For CTM, which is distilled from EDM, we adopt the EDM noise parameterization together with local Gaussian approximations of the sender and shared distributions at intermediate states. This yields a proof-of-concept probabilistic formulation that enables compression with pre-trained few-step generative models without retraining. On low-resolution benchmarks, the resulting codecs reduce encoding and decoding time and improve realism in the low-bit-rate regime.
Zero-shot diffusion image compression based on reverse channel coding achieves high perceptual quality at low bitrates without retraining, yet its bitstream remains vulnerable to bit flips. Prior work has shown that a joint combinatorial representation can amplify a single bit flip into substitutions of multiple atoms and that indexing each atom independently confines this error propagation. However, independent indexing reduces the number of atoms that fit within the bit budget and gives the decoder no indication of which atoms are corrupted. To address these issues, we propose BitFlat, a training-free error-localizing bitstream framework with two components. Blockwise Locality Alignment (BLA) decomposes the globally coupled atom selection at each diffusion step into independent selection units, each describing one signed atom from a dedicated codebook with a fixed-width field, so that any bit error affects only the unit in which it occurs. Reliability-Aware Diffusion Control (RADC) uses the bits saved by BLA to impose structured valid-codeword constraints within each field and converts the decoding state of each field into a binary reliability gate, which turns detected corrupted atoms into erasures at known positions and excludes them from noise synthesis. BitFlat requires neither target residuals nor additional network forward passes and leaves the diffusion model unchanged. Experiments on Kodak, DIV2K, and CLIC2020 show that BitFlat outperforms conventional, learned, and diffusion-based codecs at bit error rates from 10−4 to 10−2 and, even with 26.7% fewer bits, surpasses Robust Turbo-DDCM.
Xinfeng Deng, Li Zhou, Haitao Chen +5
College of Electronic Science and Technology, National University of Defense Technology, Changsha, 410073, China.