Beyond Prompts: Unconditional 3D Inversion for Out-of-Distribution Shapes

TL;DR

This paper reveals that state-of-the-art text-to-3D models often become insensitive to prompts due to latent sink traps, and proposes a method leveraging unconditional priors for robust out-of-distribution shape editing.

Contribution

It introduces a novel framework that decouples geometric expressivity from linguistic sensitivity, enabling reliable out-of-distribution 3D shape manipulation.

Findings

Models can generate diverse shapes despite prompt insensitivity.

Leveraging unconditional priors allows for high-fidelity shape editing.

The approach bypasses latent sink traps for robust out-of-distribution editing.

Abstract

Text-driven inversion of generative models is a core paradigm for manipulating 2D or 3D content, unlocking numerous applications such as text-based editing, style transfer, or inverse problems. However, it relies on the assumption that generative models remain sensitive to natural language prompts. We demonstrate that for state-of-the-art native text-to-3D generative models, this assumption often collapses. We identify a critical failure mode where generation trajectories are drawn into latent ``sink traps'': regions where the model becomes insensitive to prompt modifications. In these regimes, changes to the input text fail to alter internal representations in a way that alters the output geometry. Crucially, we observe that this is not a limitation of the model's \textit{geometric} expressivity; the same generative models possess the ability to produce a vast diversity of shapes but, as we demonstrate, become insensitive to out-of-distribution \textit{text} guidance. We investigate this behavior by analyzing the sampling trajectories of the generative model, and find that complex geometries can still be represented and produced by leveraging the model's unconditional generative prior. This leads to a more robust framework for text-based 3D shape editing that bypasses latent sinks by decoupling a model's geometric representation power from its linguistic sensitivity. Our approach addresses the limitations of current 3D pipelines and enables high-fidelity semantic manipulation of out-of-distribution 3D shapes. Project webpage: https://daidedou.sorpi.fr/publication/beyondprompts