Instruct 3D-to-3D
High-fidelity text-guided conversion and editing of 3D scenes using iterative diffusion updates.

Edit 3D scenes with text instructions using Iterative Dataset Updates and Diffusion Models.
InstructNeRF2NeRF is a state-of-the-art framework for editing Neural Radiance Fields (NeRFs) using text-based instructions. Unlike traditional 3D editing which requires manual geometry or texture manipulation, this tool utilizes a 2D diffusion model (InstructPix2Pix) to iteratively refine a 3D scene. The architecture employs a unique 'Iterative Dataset Update' (IDU) method, which alternates between updating the underlying dataset of images using the diffusion model and retraining the NeRF to maintain multi-view consistency. This ensures that edits—such as changing a person's clothes, turning a landscape from summer to winter, or stylizing a room—remain spatially coherent from any camera angle. As of 2026, it remains the industry standard for researchers and VFX artists looking to bridge the gap between text-to-image generative AI and consistent 3D world-building. It is primarily built upon the Nerfstudio ecosystem, offering high modularity and support for various NeRF backbones like Nerfacto.
The system replaces original training images with diffusion-edited versions incrementally during NeRF training.
High-fidelity text-guided conversion and editing of 3D scenes using iterative diffusion updates.
High-Fidelity Shading-Guided 3D Asset Generation from Sparse 2D Inputs
High-Quality Single Image to 3D Generation using 2D and 3D Diffusion Priors
Edit 3D scenes and NeRFs with natural language instructions while maintaining multi-view consistency.
Verified feedback from the global deployment network.
Post queries, share implementation strategies, and help other users.
Leverages a diffusion model specifically trained to follow human instructions for image editing.
Uses camera pose information to ensure the diffusion model applies edits consistently relative to the viewer.
Fully integrated as a plug-in for Nerfstudio, utilizing its optimized rendering backends.
Adjusts the strength of the diffusion guidance dynamically throughout the training process.
Changing a sunny 3D capture of a building to a snowy winter scene without re-modeling.
Registry Updated:2/7/2026
Render fly-through video
Modifying a 3D avatar's clothing texture and material for fashion design.
Turning photo-real assets into stylized, artistic, or 'cartoon' versions for indie games.