I have any questions of your paper

[soohyoen]

Thank you for your awesome work.

Ive read your paper, but Id want to know more about how this method is possible.

1. Diffusion models have randomness properties; however, your model have consistency of any view-point.
   How to get consistency of any view-point?

2. Your model is the SOTA of 3D editing models.
   I would like to know why your model has a high CLIP dimension score compared to other models.

[2hiTee]

For question one, I think this work is not concentrating on multi-view consistency, the multi-view consistency is guranteed by inputting multi-view images for editing, just like InstructNeRF2NeRF, but the multi-view consistency still faces a problem.