Results not as good on custom dataset

[anna-debug]

Hello!
I have been training the network on my own data and the rendering results are way worse than the ones from training on the provided datasets. I trained on a dataset that has 53 images. I kept the same parameters for the config files, changing the image size and name, of course.

One of the input images:

And these are the result images taken from the logs folder:


Visualizing the dense output from colmap:

What could be the possible reasons for such results?

[zfkuang]

Hey anna,

I can't tell what's going on based on the images you're sending me. Can you send me your log files? (it would be better if you could also send your training images, you can send them to my email: zhengfei@stanford.edu)

Also, please make sure you are running your code under ubuntu. Windows is not supported for now.

[pfirmino]

Any plans to have a working windows release?