Pith. sign in

REVIEW 3 cited by

On Aliased Resizing and Surprising Subtleties in GAN Evaluation

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2104.11222 v3 pith:PSGSJ7FF submitted 2021-04-22 cs.CV cs.GRcs.LG

classification cs.CVcs.GRcs.LG
keywords imagecompressiongenerativeimagesprocessingresizingvariationsaliasing
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Metrics for evaluating generative models aim to measure the discrepancy between real and generated images. The often-used Frechet Inception Distance (FID) metric, for example, extracts "high-level" features using a deep network from the two sets. However, we find that the differences in "low-level" preprocessing, specifically image resizing and compression, can induce large variations and have unforeseen consequences. For instance, when resizing an image, e.g., with a bilinear or bicubic kernel, signal processing principles mandate adjusting prefilter width depending on the downsampling factor, to antialias to the appropriate bandwidth. However, commonly-used implementations use a fixed-width prefilter, resulting in aliasing artifacts. Such aliasing leads to corruptions in the feature extraction downstream. Next, lossy compression, such as JPEG, is commonly used to reduce the file size of an image. Although designed to minimally degrade the perceptual quality of an image, the operation also produces variations downstream. Furthermore, we show that if compression is used on real training images, FID can actually improve if the generated images are also subsequently compressed. This paper shows that choices in low-level image processing have been an underappreciated aspect of generative modeling. We identify and characterize variations in generative modeling development pipelines, provide recommendations based on signal processing principles, and release a reference implementation to facilitate future comparisons.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Consistent and Controllable Image Animation with Motion Linear Diffusion Transformers

    cs.CV 2025-08 conditional novelty 6.0 of 10

    MiraMo turns a static image into a video using a linear-attention transformer that learns inter-frame motion residuals, with DCT-based noise refinement and a user-controllable dynamics knob.

  2. FEAT: Full-Dimensional Efficient Attention Transformer for Medical Video Generation

    cs.CV 2025-06 conditional novelty 6.0 of 10

    FEAT combines spatial, temporal, and channel attention in a diffusion transformer and reports improved medical video generation metrics with lower parameter counts than Endora.

  3. PyPotteryInk: One-Step Diffusion Model for Sketch to Publication-ready Archaeological Drawings

    cs.GR 2025-02 conditional novelty 5.0 of 10

    A one-step diffusion model converts archaeological pottery sketches into clean inked drawings, with expert ratings showing usable quality on Italian protohistoric vessels.

Pith tools