CubeDiff: Repurposing Diffusion-Based Image Models for Panorama Generation

A row of six windows onto six generated worlds, signed by six cartoon faces.

cubediff.github.ioCubeDiff: Repurposing Diffusion-Based Image Models for Panorama Generation
Screenshot of https://cubediff.github.io/, © the authors. Shown for reference. Request removal
Venue
ICLR 2025
Year
2025
First author
Nikolai Kalischek
Field
image generation
Template
hand-written CSS (no framework)
Theme
Light

Checked against the rendered page. Only Roboto 400 is requested, so all bold headings are browser-synthesized; body text is justified at normal line height.

Why it is here

Author avatar portraits and a full-bleed video strip make a lively hero. The tl;dr pairs a 3D cube render with text, and a flowing pipeline graphic leads into rounded result tiles. Body text is slightly cramped and justified.

Overview

Color Palette

Typography

Roboto
CubeDiff: Repurposing Diffusion-Based Image Models for Panorama Genera Abstract · Method · Results · BibTeX — 0123456789

Spacing & Shape

Page Modules

Guidelines

More like this