Skip to content

Sketch2NeRF

Guide text-to-3D generation with multi-view sketches

Visit official site

What Sketch2NeRF is

The framework is described as guiding text-to-3D generation with multi-view sketches while optimising a NeRF scene.

What you can do with it

  • Control 3D generation with sketches
  • Synthesise high-fidelity 3D content
  • Optimise a neural radiance field

Method notes

  • Pretrained 2D diffusion models include Stable Diffusion and ControlNet
  • A simultaneous generation and reconstruction method is proposed
  • Evaluation uses two collected multi-view sketch datasets

Pros & cons

✓ What we like

  • Two multi-view sketch datasets are used in evaluation
  • Sketch control is applied alongside text prompts

! What to watch out for

  • The method depends on pretrained 2D diffusion models
  • Outputs are tied to the supplied multi-view sketches

FAQ

Which 2D models are named?

Stable Diffusion and ControlNet are named as pretrained 2D diffusion models.

What is optimised?

A 3D scene represented by a neural radiance field, or NeRF.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review