how to perform training with our own dataset ? Will it produce same results with drone captured images ?

[piyushsingh2k7]

No description provided.

[jmunkberg]

We need accurate camera poses and good foreground segmentation masks for high quality results. Avoid motion blur, defocus blur and changing lighting conditions (we assume a constant lighting and optimize a single environment map from the images).

To prepare your own dataset, you can mostly follow https://github.com/bmild/nerf#generating-poses-for-your-own-scenes but please note that we need masks in the alpha channel.