Questions regarding how to process data into correct data format

[Yushi-Du]

Hi, thank you very much for the wonderful work!

When I was using your method to train a nerf on my data, I found the image shown in the visualizer is empty in the rgb channel. However, there are contents in the depth and accumulation channels.

I have checked the camera pose using other methods and it is correct. I wonder if there are some mistakes in my data format?

Here is an example of my data. I am using blender to render the images. The width, height and focal parameters are retrieved using blender python API.

{
"fl_x": 262.5,
"fl_y": 262.5,
"k1": 0,
"k2": 0,
"p1": 0,
"p2": 0,
"cx": 128,
"cy": 128,
"w": 256,
"h": 256,
"aabb_scale": 16,
"frames": [
{
"file_path": "images/0001.png",
"transform_matrix": [
[
-0.8327592015266418,
-0.4579005837440491,
0.3111884593963623,
0.40454572439193726
],
[
-3.7252868878567824e-07,
0.5620833039283752,
0.827080488204956,
1.0752044916152954
],
[
-0.553634524345398,
0.6887590885162354,
-0.4680800437927246,
-0.6085042953491211
],
[
0.0,
0.0,
-0.0,
1.0
]
]
}
]
}

I would really appreciate you reply and assistance. Thank you once again.