Blending-NeRF: Text-Driven Localized Editing in Neural Radiance Fields
This addresses the challenge of precise 3D object editing for applications in computer graphics and vision, though it appears incremental as it builds on existing NeRF and CLIP methods.
The paper tackles the problem of text-driven localized editing of 3D objects in Neural Radiance Fields (NeRF) by proposing Blending-NeRF, which uses two NeRF networks and new blending operations guided by CLIP to edit target regions without distorting form, resulting in naturally and locally edited 3D objects from various text prompts.
Text-driven localized editing of 3D objects is particularly difficult as locally mixing the original 3D object with the intended new object and style effects without distorting the object's form is not a straightforward process. To address this issue, we propose a novel NeRF-based model, Blending-NeRF, which consists of two NeRF networks: pretrained NeRF and editable NeRF. Additionally, we introduce new blending operations that allow Blending-NeRF to properly edit target regions which are localized by text. By using a pretrained vision-language aligned model, CLIP, we guide Blending-NeRF to add new objects with varying colors and densities, modify textures, and remove parts of the original object. Our extensive experiments demonstrate that Blending-NeRF produces naturally and locally edited 3D objects from various text prompts. Our project page is available at https://seokhunchoi.github.io/Blending-NeRF/