Few-Step Generative Models Adapted for Fast Lossy Compression Without Retraining
A new study proposes using few-step generative models — Rectified Flow, Consistency Trajectory Models, and MeanFlow — as image codecs within the reverse channel coding (RCC) framework, without requiring retraining. The work addresses a key bottleneck in DiffC, an existing diffusion-based compression method, which is slow due to its many forward and reverse steps. The approach reduces encoding and decoding time while improving image realism at low bit rates, offering a more practical path toward generative-model-based compression.
A preprint posted to arXiv presents a method for repurposing pre-trained few-step generative models as lossy image codecs using the reverse channel coding (RCC) framework. The study targets a known limitation of DiffC, a diffusion-based compression system whose encoding and decoding are computationally expensive due to the large number of discretized steps required. The authors address the core technical challenge that RCC requires explicit parameterization of intermediate conditional distributions — something few-step models do not natively provide. For Rectified Flow and MeanFlow, they exploit an equivalence between velocity parameterization and diffusion-style denoising to derive the necessary quantities; for Consistency Trajectory Models (CTM), they use EDM noise parameterization combined with local Gaussian approximations. The resulting proof-of-concept codecs are evaluated on low-resolution benchmarks, where they demonstrate faster encoding and decoding and improved perceptual quality in the low-bit-rate regime. No retraining of the underlying generative models is required, making the approach relatively accessible for practitioners with existing pre-trained models.
What's missing
The study is a proof-of-concept evaluated only on low-resolution benchmarks, leaving open questions about scalability to high-resolution images and real-world compression pipelines. Quantitative comparisons against established non-generative codecs (e.g., JPEG, WebP, VVC) are not described in the abstract, making it unclear how competitive the approach is in absolute terms. The local Gaussian approximations used for CTM may introduce modeling errors whose impact is not fully characterized.
What different sources said
- arXiv cs.LGCenter
Few-step Generative Models as Lossy Compression
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.