Vibe Spaces for Creatively Connecting and Expressing Visual Concepts

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Yang, Huzheng, Xu, Katherine, Lu, Andrew, Grossberg, Michael D., Bai, Yutong, Shi, Jianbo
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915680820920320
author Yang, Huzheng
Xu, Katherine
Lu, Andrew
Grossberg, Michael D.
Bai, Yutong
Shi, Jianbo
author_facet Yang, Huzheng
Xu, Katherine
Lu, Andrew
Grossberg, Michael D.
Bai, Yutong
Shi, Jianbo
contents Creating new visual concepts often requires connecting distinct ideas through their most relevant shared attributes -- their vibe. We introduce Vibe Blending, a novel task for generating coherent and meaningful hybrids that reveals these shared attributes between images. Achieving such blends is challenging for current methods, which struggle to identify and traverse nonlinear paths linking distant concepts in latent space. We propose Vibe Space, a hierarchical graph manifold that learns low-dimensional geodesics in feature spaces like CLIP, enabling smooth and semantically consistent transitions between concepts. To evaluate creative quality, we design a cognitively inspired framework combining human judgments, LLM reasoning, and a geometric path-based difficulty score. We find that Vibe Space produces blends that humans consistently rate as more creative and coherent than current methods.
format Preprint
id arxiv_https___arxiv_org_abs_2512_14884
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Vibe Spaces for Creatively Connecting and Expressing Visual Concepts
Yang, Huzheng
Xu, Katherine
Lu, Andrew
Grossberg, Michael D.
Bai, Yutong
Shi, Jianbo
Computer Vision and Pattern Recognition
Creating new visual concepts often requires connecting distinct ideas through their most relevant shared attributes -- their vibe. We introduce Vibe Blending, a novel task for generating coherent and meaningful hybrids that reveals these shared attributes between images. Achieving such blends is challenging for current methods, which struggle to identify and traverse nonlinear paths linking distant concepts in latent space. We propose Vibe Space, a hierarchical graph manifold that learns low-dimensional geodesics in feature spaces like CLIP, enabling smooth and semantically consistent transitions between concepts. To evaluate creative quality, we design a cognitively inspired framework combining human judgments, LLM reasoning, and a geometric path-based difficulty score. We find that Vibe Space produces blends that humans consistently rate as more creative and coherent than current methods.
title Vibe Spaces for Creatively Connecting and Expressing Visual Concepts
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.14884