CVFeb 13, 2024

Learning Continuous 3D Words for Text-to-Image Generation

arXiv:2402.08654v121 citationsh-index: 40CVPR
Originality Incremental advance
AI Analysis

This provides incremental improvement for users of diffusion models by offering more precise control over specific 3D-aware attributes in image generation.

The paper tackles the problem of fine-grained control over abstract, continuous attributes like illumination direction in text-to-image generation by introducing Continuous 3D Words, which enable users to adjust attributes via sliders without adding overhead to the generative process.

Current controls over diffusion models (e.g., through text or ControlNet) for image generation fall short in recognizing abstract, continuous attributes like illumination direction or non-rigid shape change. In this paper, we present an approach for allowing users of text-to-image models to have fine-grained control of several attributes in an image. We do this by engineering special sets of input tokens that can be transformed in a continuous manner -- we call them Continuous 3D Words. These attributes can, for example, be represented as sliders and applied jointly with text prompts for fine-grained control over image generation. Given only a single mesh and a rendering engine, we show that our approach can be adopted to provide continuous user control over several 3D-aware attributes, including time-of-day illumination, bird wing orientation, dollyzoom effect, and object poses. Our method is capable of conditioning image creation with multiple Continuous 3D Words and text descriptions simultaneously while adding no overhead to the generative process. Project Page: https://ttchengab.github.io/continuous_3d_words

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes