NeRF visualization library under construction

Hello there. This is a very interesting code. I wonder if it can also visualize the plenoctree format for a 3D model. For instance, I have a tree.npy file from https://github.com/sxyu/plenoctree. How can I visualize it?

I have successfully visualized the NeRF model in the browser. However, how do I save the scene in npz.file? For instance,

scene.set_nerf(nerf_func, center=[0.0, 0.0, 1.0], radius=2, use_dirs=True)
scene.display(open_browser=False, port=args.port)
scene.write(path=args.npz_path) # how to save the scene?

Hello,

Are there any recommendations on what the sh_proj_sample_count and chunk parameters mean, and how to set them when calling the Scene.set_nerf() method? The latter doesn't seem to be included in the docstring for it.

The reason I ask is mainly because I want to better predict what the dimensions of the grid variable will be, based on the parameters passed to the function. I think this might help prevent errors in places like in sh.project_function_sparse() function, where it does the transpose:

func_value = func_value.transpose(0, 1).reshape(
           sample_count, batch_size * C) # [sample_count, batch_size * C]

For research purposes, I would like to visualize a Plenoxel.

I added a method for this:

See https://github.com/pietertolsma/nerfvis/blob/master/nerfvis/scene.py

Hello, I am getting this error:

File "/home/me/anaconda3/envs/ngp_pl/lib/python3.9/site-packages/nerfvis/scene.py", line 976, in add_nerf assert grid.shape[0] > 0, "This NeRF is completely empty! Make sure you set the bounds reasonably" AssertionError: This NeRF is completely empty! Make sure you set the bounds reasonably

im using a ngp_pl trained nerf instead of nerf_pl. I believe the structure is the same unless i'm way off. Could you please shed some light onto this issue?

Thanks in advance.

P.S: I can render the silica nerf from nerf_pl with no issues.

Hi, when using NerfVis, there is a time-slider in the top-right corner, but there is nothing about it in the documentation. How can i utilize it in the visualization? Can I train a NeRF on a video and then use the time-slider to display said video from an arbitrary viewpoint?

Thanks a lot!

Hello there, I tried your code on "chair" example from blender dataset and used the nerfvis library for visualization. It totally worked. However, when I tried to visualize other examples, like "ship" from blender dataset, it stops at loading 100% and it never shows the 3D model like

Can you tell me what is the problem? Thank you in advance.

First of all, thank you for your great work!

How to slice any cross-sectional views I want rather than just three direction views?

Thank you in advance.

import nerfvis scene = nerfvis.Scene("My title") scene.add_cube("Cube1", color=[1.0, 0.0, 0.0], translation=[-1.0, -1.0, 0.0]) scene.display(url='10.255.26.23',port=8020)

【jupyter can accessed it,but nerfvis cannot access it:(】 then there is error: Serving localhost:8020

Exception happened during processing of request from ('172.24.142.99', 64624) Traceback (most recent call last): File "/opt/conda/envs/py36/lib/python3.6/socketserver.py", line 320, in _handle_request_noblock self.process_request(request, client_address) File "/opt/conda/envs/py36/lib/python3.6/socketserver.py", line 351, in process_request self.finish_request(request, client_address) File "/opt/conda/envs/py36/lib/python3.6/socketserver.py", line 364, in finish_request self.RequestHandlerClass(request, client_address, self) File "/opt/conda/envs/py36/lib/python3.6/site-packages/nerfvis/scene.py", line 840, in init super().init(*args, directory=dirname, **kwargs) TypeError: init() got an unexpected keyword argument 'directory'

please show me how to solve it