ICE-NeRF: Interactive Color Editing of NeRFs via
Decomposition-Aware Weight Optimization

Jae-Hyeok Lee     Dae-Shik Kim
KAIST
ICCV 2023
Paper Code (available soon)

Color Editing on Real Scenes

Results were obtained by applying ICE-NeRF to a pre-trained vanilla NeRF.


Our Results on Horns Scene.

Our Results on Flower Scene.

Our Results on Fortress Scene.



Color Editing on Synthetic Scenes

Results were obtained by applying ICE-NeRF to a pre-trained TensoRF.


Our Results on Chair Scene.

Our Results on Ship Scene.

Our Results on Hotdog Scene.



Multi-Color Editing


Multi-Color Editing Result on Horns Scene.



Multi-View Mask Examples (SMR)

The proposed SMR utilizes the user's single mask input by rendering it into multiple views. Typically, in training, up to 5 additional views are rendered and used. Here, to demonstrate the well-generated multi-view masks, we show the rendered masks from 120 consecutive views.


SMR Output on Horns Scene.

SMR Output on Flower Scene.

BibTeX

@InProceedings{Lee_2023_ICCV,
      author    = {Lee, Jae-Hyeok and Kim, Dae-Shik},
      title     = {ICE-NeRF: Interactive Color Editing of NeRFs via Decomposition-Aware Weight Optimization},
      booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
      month     = {October},
      year      = {2023},
      pages     = {3491-3501}
  }