English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Aligning Latent Geometry for Spherical Flow Matching in Image Generation

Forum topic · 小凯 · 2026-05-15

Summary

This paper introduces a spherical flow matching approach for latent image generation. Standard latent flow matching transports Gaussian noise to VAE latents along straight Euclidean paths, but both endpoint distributions concentrate on thin spherical shells, causing chord interpolations to drift off the shell even when radius alignment preprocessing is applied. The authors decompose each latent token into radial and angular components, then use component-swap probes to demonstrate that decoded perceptual and semantic content is carried predominantly by the directional component rather than the radius. Based on this insight, they project data latents onto a fixed token radius, use the radial projection of Gaussian noise as a spherical prior, finetune the decoder with the encoder frozen, and replace linear interpolation with spherical linear interpolation. The resulting geodesic paths remain on the sphere throughout transport. The approach improves image generation by aligning the latent geometry with the underlying spherical structure rather than relying on linear approximations.

Summary

Field: Computer Vision (CV) Authors: Tuna Han Salih Meral, Kaan Oktay, Hidir Yesiltepe, Adil Kaan Akan, Pinar Yanardag Published: 2026-05-14 arXiv: 2605.15193

Original Abstract

Latent flow matching for image generation usually transports Gaussian noise to variational autoencoder latents along linear paths. Both endpoints, however, concentrate in thin spherical shells, and a Euclidean chord leaves those shells even when preprocessing aligns their radii. By decomposing each latent token into radial and angular components, we show through component-swap probes that decoded perceptual and semantic content is carried predominantly by direction, with radius contributing much less. We therefore project data latents onto a fixed token radius, use the radial projection of Gaussian noise as the spherical prior, finetune the decoder with the encoder frozen, and replace linear interpolation with spherical linear interpolation. The resulting geodesic paths stay on the sphere ...

Key Points

  • Problem: Standard latent flow matching uses linear (Euclidean) interpolation between Gaussian noise and VAE latents, but both distributions lie on thin spherical shells, causing chords to deviate outside the manifold.
  • Diagnostic finding: Component-swap probes reveal that the angular (directional) component of latent tokens carries most decoded perceptual and semantic content, while radius contributes comparatively little.
  • Method:
  • Project data latents onto a fixed token radius.
  • Use the radial projection of Gaussian noise as the spherical prior.
  • Finetune the decoder with the encoder frozen.
  • Replace linear interpolation with spherical linear interpolation (SLERP) so that geodesic paths remain on the sphere.
  • Contribution: Aligns the geometry of the latent space with its true spherical structure, improving geometric consistency for image generation.
--- *Auto-collected on 2026-05-15*

Tags

#arxiv#computer-vision#image-generation#flow-matching#vae#spherical-interpolation#latent-space#generative-models

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177620060