LookingGlass: Generative Anamorphoses via Laplacian Pyramid Warping

原始链接: https://studios.disneyresearch.com/2025/06/09/lookingglass-generative-anamorphoses-via-laplacian-pyramid-warping/

Anamorphosis images are distorted visuals that only reveal their true form when viewed from a specific angle or through a special device. While these illusions have a long history, they typically appear meaningless when viewed normally. This paper presents a novel approach using generative latent rectified flow models to create anamorphic images that maintain a discernible interpretation even when viewed directly. The key innovation is Laplacian Pyramid Warping, a frequency-aware image warping technique that ensures high-quality results. This research expands upon the concept of Visual Anagrams by applying it to latent space models and a broader range of spatial transformations. Ultimately, this method enables the generation of new and engaging perceptual illusions that are both anamorphic and visually coherent from multiple viewpoints.

这个黑客新闻线程讨论了一个迪士尼研究项目,该项目“通过拉普拉斯金字塔翘曲的生成变形,”,该项目从特定角度创建了扭曲的图像,这些图像从特定的角度降为一致。 用户与相关概念建立了连接,包括视频像素操纵(Marek Gibney的工作在视频帧中交换像素),带有多个解决方案(站立数学)的视觉难题以及隐肌的可能性。共享指向视觉词汇和扩散幻觉的链接。 讨论深入研究了变形技术的历史,质疑诸如达芬奇镜子(Da Vinci's Mirror)写作之类的早期例子是否有资格为“变形加密”。 一位评论者与迪斯尼高管分享了负面的经历,他驳回了其Genai创业公司的成就,从而对迪士尼文化进行了更广泛的思考。 该线程以这种研究的实际应用的多年生问题结束,基本研究可能导致意外突破。
相关文章

原文

Anamorphosis refers to a category of images that are intentionally distorted, making them unrecognizable when viewed directly. Their true form only reveals itself when seen from a specific viewpoint, which can be through some catadioptric device like a mirror or a lens. While the construction of these mathematical devices can be traced back to as early as the 17th century, they are only interpretable when viewed from a specific vantage point and tend to lose meaning when seen normally. In this paper, we revisit these famous optical illusions with a generative twist. With the help of latent rectified flow models, we propose a method to create anamorphic images that still retain a valid interpretation when viewed directly. To this end, we introduce Laplacian Pyramid Warping, a frequency-aware image warping technique key to generating high-quality visuals. Our work extends Visual Anagrams to latent space models and to a wider range of spatial transforms, enabling the creation of novel generative perceptual illusions.

联系我们 contact @ memedata.com