vsitzmann / deepvoxels Goto Github PK

How can we get intrinsics.txt for our prepared data?

The code also reads an "intrinsics.txt" file from the dataset directory. This file is expected to be structured as follows:

f cx cy
origin_x origin_y origin_z
near_plane (if 0, defaults to sqrt(3)/2)
scale
img_height img_width

When running the code, I found that each training can only train one object. Can I train on more than one object?

Lifting invalid
error: nothing in frustum bounds
Lifting invalid
error: nothing in frustum bounds
Lifting invalid
error: nothing in frustum bounds
Lifting invalid
error: nothing in frustum bounds
Lifting invalid
error: nothing in frustum bounds
Lifting invalid
error: nothing in frustum bounds
Lifting invalid

Thanks for your outstanding work!
I'm trying to run this model with my dataset. But I can't obtain the camera parameters. I've tried to use sparse bundle adjustment but it didn't work.
Could u please briefly explain which tools you use and the steps?

Hello, I was trying to use your wrapper function to generate intrinsics and extrinsics and the code crashed with this error.

Extracting poses
Traceback (most recent call last):
File "colmap_wrapper.py", line 245, in
images = read_poses(reconst_dir)
File "colmap_wrapper.py", line 222, in read_poses
images = read_images_binary(os.path.join(colmap_workspace, 'sparse', '0', "images.bin"))
File "colmap_wrapper.py", line 160, in read_images_binary
with open(path_to_model_file, "rb") as fid:
FileNotFoundError: [Errno 2] No such file or directory: './output/reconstruction/sparse/0/images.bin'

I will try to debug this, but any help is appreciated. Regards.

The dataset contains depth maps in png format. I can read them via cv2.imread(depth_path, cv2.IMREAD_ANYDEPTH) and get values in unit16. How can I scale the values to get the correct depth?

Hi, I wonder what is the computer's graphic memory when training the data? I received the error msg:
RuntimeError: CUDA out of memory. Tried to allocate 9.00 MiB (GPU 0; 8.00 GiB total capacity; 641.54 MiB already allocated; 8.70 MiB free; 474.50 KiB cached)

I keep printing nvidia-smi and found that it reached ~7GB and finally reported the error because I only have 8 GB memory.

I wonder how much memory is required? If it does require more than 8 GB memory, any suggestion for me to change the code? Thanks a lot.

First of all, thank you for your outstanding work.

There are a few issues that I would like to clear up, as I am having trouble getting good results with new data:

• Why is the translation of the frustum bounds to the barycenter switched off while lifting?

When computing the frustum bounds while lifting, in 'compute_frustum_bounds()', I find the following code:

# Transform to grid coordinates (grid at origin)
pl = torch.round(torch.bmm(world_to_grid.repeat(8, 1, 1), torch.floor(p)))
pu = torch.round(torch.bmm(world_to_grid.repeat(8, 1, 1), torch.ceil(p)))
pl = torch.round(torch.floor(p))
pu = torch.round(torch.ceil(p))

The grid coordinates with which we intersect later are translated but not the boundaries.

• What is the role of the near_plane parameter in the intrinsics file and how should it be chosen?

It seems that it is only used for projection, while otherwise a separate opt.near_plane parameter is used. Moreover, it is not used to clip the visible field (as the concept of near plane is) but to translate the points. It is unclear why this is necessary for projection and how to select a value of this parameter and opt.near_plane.

Hello,
Thank you for sharing code.
Which GPU did you use for training this model?
Do you provide trained checkpoints?

I'm getting the following error:

img_dir: /content/drive/My Drive/nerf/orchid/images
trgt_dir: /content/drive/My Drive/nerf/orchid/
dense: False
Bundle Adjusting
F0611 13:36:02.462942  8239 automatic_reconstruction.cc:51] Check failed: ExistsDir(options_.workspace_path) 
*** Check failure stack trace: ***
    @     0x7f2a28ed70cd  google::LogMessage::Fail()
    @     0x7f2a28ed8f33  google::LogMessage::SendToLog()
    @     0x7f2a28ed6c28  google::LogMessage::Flush()
    @     0x7f2a28ed9999  google::LogMessageFatal::~LogMessageFatal()
    @     0x5598b03d6350  (unknown)
    @     0x5598b0301f8d  (unknown)
    @     0x5598b02e4e3e  (unknown)
    @     0x7f2a23377b97  __libc_start_main
    @     0x5598b02eea7a  (unknown)
Extracting poses
Traceback (most recent call last):
  File "colmap_wrapper.py", line 245, in <module>
    images = read_poses(reconst_dir)
  File "colmap_wrapper.py", line 222, in read_poses
    images = read_images_binary(os.path.join(colmap_workspace, 'sparse', '0', "images.bin"))
  File "colmap_wrapper.py", line 160, in read_images_binary
    with open(path_to_model_file, "rb") as fid:
FileNotFoundError: [Errno 2] No such file or directory: '/content/drive/My Drive/nerf/orchid/reconstruction/sparse/0/images.bin'

What did I do wrong?