This is mostly for me, but anyone interested in details of the video is welcome to come along with me! I’m not going to do a breakdown of “the good” or “the bad”, only lessons learned and what I’d intended vs. what was delivered.
The video followed the overall outline I’d set for it. I decided early on how many sections I wanted, what they would cover, and how long they’d be in seconds. More important sections got 30 seconds, shorter ones got 20.
I wrote narration for each section. I read some of it aloud to measure the time, aiming for blocks of 10-second audio. Similarly, I searched Incompetech for a soundtrack that fit both the length and tone I wanted.
What I didn’t do on the audio is any kind of volume correction. I feel strongly that audio is critical to making any video good. On the other hand, recording with the lavalier and shotgun mics means I minimized echo and noise, and I feel like the audio quality itself is comparatively good. DaVinci Resolve comes with Fairlight, but I don’t know how to use it very well. I did use Audacity to denoise the audio, at least.
The green-screen bit at the start was done literally the night of the release. I’d been putting it off for assorted reasons. That said, I learned my lessons from previous setups and I think the capture itself came out great. The footage was comped in Blender. There’s a big bug where light would literally shine through me. If I do this again, I need to find ways to block the bloom from doing that.
The stargate map at 0:20 is super duper janky. Originally it was meant to be just a backdrop image to scenes of social unrest and natural disaster, but I had trouble finding appropriate footage that was both generic enough to fit into a sci-fi narrative, and clearly licensed for commercial re-use. In the end I gave up, stretched a 10-second clip into 20 seconds, and moved on. This is why the animation on the lines looks stilted, it’s running at 12 fps.
For the crew pictures at 1:10, I found CC or CC0 photos and used Affinity Photo to shop them onto Blender-rendered sci-fi scenes. I also applied a LUT to each picture, choosing from a LUT pack I’d recently gotten thanks to Humble Bundle. The edge detection wasn’t perfect, but it was minutes of work rather than hours. You can most clearly see this problem on Earl Baxter, who really should have been backlit much more dramatically.
Not only did I need to put the characters in an appropriate setting, but I also needed to sort of tell a mini-story with each scene, and I think I did okay in each case.
I’m most proud of the final shot at 1:40, the stargate jump. I had the benefit of some early feedback from folks on that, and took a suggestion about “showing the stars on the other side” and tried to make it work. The implementation of the effect wasn’t the best, but I was still learning the visibility modifiers you can attach to meshes, and I think one of those (plus subdividing some of the meshes further) would have made it look better.
What would have pushed this production further toward great, in my opinion?
- Attention to detail. I could have added some subtle text effects in Resolve to the space scenes, giving more of an impression that we’re flying around in a little shuttle observing this stuff. The asteroid ship schematic could have used this too.
- Audio mixing and composition. Instead of just dropping blocks of narration in, I feel I could have built a spine out of words and assembled footage around that.
- Learning Fairlight to enhance the audio experience.
- Smoother transitions. Some of the jumps were a little abrupt, and I think having a consistent visual language for the transitions would have sold the piece as a more coherent whole.
Overall, I think it’s good for what it is – an early “student film” that ties together elements that I’ve learned. It’s taught me a lot about the process. It’s easy to say “I could have done better”. Instead, I will say “I will be able to do better, because I did this first”.
Clocktropolis 