Matthew Tancik

The model completed training correctly, we have an off-by-one logging error which is why you don't see the last log. The codes does not automatically quit as this would close...

Great find! It does seem to fix the NAN issue on the redwood2 scene. The only downside is that there is a non-negligible performance hit (I'm seeing that training is...

Tested on a fisheye dataset. Seems to work fine.

Fixed in #914 Will be included in next pip package. Note that the code is the same, now it just reports that training is finished and tells users to use...

I think that would be a great addition. Feel free to create a PR.

Yea, the documentation is a little out of date, we will update. For running the viewer locally, you can skip that section, once the dependencies are setup (node, yarn), you...

Any chance you have a firewall on port 7007?

Yea, colab unfortunately fails to propagate the logs. I'm going to just remove this option from the script to avoid this confusion in the future.

We use the dataset provided [here](https://press.liacs.nl/mirflickr/). However in practice, any natural image dataset will suffice. Ideally use an image dataset that has similar statistics to the images that you plan...

I'm sorry, I don't maintain the rights MIRFLICKR and therefore cannot host and share it. I would recommend contacting the authors of the dataset for download assistance. Practically speaking, any...