First Shoot, pt 1 acquiring the wrong data

This weekend I attempted my first shoot! It was... a learning experience.


A good friend of mine moved into a new apartment this weekend-- I thought this would be a really well-timed opportunity to attempt shooting a full room since the rooms would be completely empty. It is also a very nice property in a cute old apartment building.

I activated my student license of 3DF Zephyr Aerial, and while reading the documentation I noticed a chapter on generating models from 360 cameras.

Here's where I got ahead of myself! I decided to rent out a Vuze  360 camera. The DSLR I wanted wasn't available, so I rented a Nikon D60 and quickly realized my Pixel phone had a better camera. I then made my focus on the 360 data acquisition and this was... not a good idea. I should have introduced 360 after the fact, instead of relying on it.

Anyways, I shot my friends large bathroom and small office. It was around 4PM and it was golden hour. In retrospect I should have started with the standard three ring capture (outwards instead of inwards) and used the 360 capture as backup, but I was rushing myself. I took the Vuze camera around in a path once with the tripod at about waist height, and then again holding the tripod over my head. I then took my phone and shot a couple hundred photos of detail spots.

When I came back to the lab and pulled the data off the Vuze card, I was disappointed to find that Zephyr would not accept the format.


Here's what I know:

•  Zephyr's documentation states that it wants an equirectangular render of the data-- this is panorama stitch of all the fisheye lenses together into a rectangular jpg. Vuze tells me that it renders an equirectangular image, but for whatever reason, these renders did not process through Zephyr. One concern I had was that even though I checked 'max resolution' in Vuze, it renders a square image.
  
  
• Zephyr doesn't actually work with the equirectangular image directly-- it chops it up into a cubemap and maps the pointcloud off of that (projecting outwards instead of inwards). Each side of the cube gets a name appended onto that file, ex "-top" 
• Vuze can also render a 'cubemap' but it compiles this into one image. One solution would be to manually chop up these renders and append the appropriate name, but that sounds like a nightmare. Also, it would not be a good long term solution.
  
  
• It makes the most sense to me to take the data directly out of the Vuze (8 fisheye images) and process in Zephyr.
  
  
• To do this, I need to manually enter the metadata associated with the camera. To do that, I need to understand what this menu is asking for.