Papers for

image generation engineers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Flow matching technique improves image generation with adaptive guidance control

Classifier-Free Guidance in Flow Matching: Non-Autonomous Potentials, Overshoot, and Posterior-Mean Control

Abstract: Classifier-free guidance (CFG) improves conditional generation in Flow Matching, but strong guidance can distort the generated distribution and reduce diversity. We provide a geometric account of this behavior by viewing Flow Matching as a time-varying gradient flow and characterizing how CFG reshapes its underlying potential. This view explains how stronger alignment can be accompanied by mean displacement and trajectory concentration, and motivates controlling guidance through the model-implied terminal posterior mean. We therefore propose Posterior-Mean-Capped CFG (PMC-CFG), a training-free, per-sample method that adaptively retains the strongest feasible guidance without additional network evaluations. Experiments on synthetic and large-scale image-generation benchmarks show that PMC-CFG limits guidance-induced distortion and concentration while improving the alignment--diversity trade-off, with particularly strong benefits when nominal guidance is large.

Mon 21 SeptComputer Vision and Pattern Recognition
The gist
Generating images with AI can be tricky when pushing the system too hard, causing the results to look less diverse or realistic. The authors study why stronger guidance in a method called flow matching can distort outputs. They suggest a way to keep guidance strong but balanced by controlling how the model changes the image generation paths. Their new method, called Posterior-Mean-Capped CFG, adapts its strength per image without extra computing steps. Tests show it keeps images varied and well-aligned when using high guidance.
Open 2609.24287v1

Visual autoregressive models improve image quality with new dependency approach

Logit Refiner: Improving Visual Autoregressive Models via Intra-Scale Dependency Modeling

Abstract: Visual Autoregressive Models (VAR) generate images through next-scale prediction, producing all tokens within each scale in parallel. We show that this parallel decoding constitutes a mean-field-style approximation that discards spatial dependencies among same-scale tokens, causing locally incoherent samples regardless of backbone capacity -- a limitation of the decoding rule. Addressing this limitation, we introduce the Logit Refiner, a lightweight autoregressive module that restores intra-scale dependencies by sequentially sampling tokens conditioned on frozen backbone features. Adding only ~10% parameters and less than 5% of the base model's training compute, it plugs into any pretrained VAR checkpoint without retraining. Controlled ablations isolate joint intra-scale sampling -- rather than additional capacity or training -- as the critical ingredient. Across backbones from 310M to 2B parameters on class-conditional ImageNet 256x256, the refiner consistently improves generation quality, enabling a 1.1B-parameter model to surpass one twice its size. The approach further generalizes to text-to-image generation, confirming that the mean-field bottleneck persists across VAR variants and is effectively alleviated by our method. Project page: https://compvis.github.io/logit-refiner/

Thu 10 SeptComputer Vision and Pattern RecognitionArtificial IntelligenceMachine Learning
The gist
Many image generation models create pictures piece by piece but treat all parts at the same detail level independently, which can cause local visual glitches. The authors found this happens because the standard step in these models ignores connections between nearby pixels at the same scale. They introduced a small add-on called the Logit Refiner that samples image tokens one by one in order, capturing these local connections without retraining the original model. This improvement leads to clearer, more coherent images across various model sizes and works for both class-based and text-based image generation.
Open 2609.11804v1