This production log has been pretty neglected over the last year, with most of my posts appearing on Patreon, Facebook, Google+, or Tumblr. For 2016, though, I’m going to start using this for short posts about the project status.Although we have not yet completed an episode, we did accomplish a lot of work in 2015:
- All the sets for Part 1 have been modeled to a rough state (most need materials and textures work).
- A few sets, such as the train passenger car and the “aquarium room” where the press conference takes place, and the park around the “Glory to the Conquerors of Space” monument are at or near completion.
- We have even got several of the sets needed for Part 2, including the locations for the “media montage” and the interior of the Lunar Transportation System lander.
- We have begun modeling sets for Part 3, including creating hatches and lunar concrete elements that will make up most of the ISF-1 Colony interiors.
- The three main characters in Part 1 of the Pilot (Georgiana, Hiromi, and Sergei) have been completely modeled and textured.
- Basic models and rigs are complete for the other main characters in the pilot (Parts 2 & 3), namely Tim, Josh, Anya, and Rob, although they still need NPR materials and texture work.
- All of the “Walk-On” characters (that is to say, characters who have at most, a few lines, and then leave the story, but who appear fully on-screen) have been completed.
- A few of the “Extras” (characters who only appear in the back of scenes and have no lines) have been finished.
- We have integrated Papagayo lip-sync into our workflow, so we can make characters speak easily and quickly (of course, it’s still necessary to animate emphasis and emotion by hand, but synchronizing to the dialog is extremely helpful).
- We have animated significant parts of the passenger train sequence and the press conference sequence — these are the two largest character animation scenes in part 1.
Of course, I (Terry Hancock) have worked very hard on this. But I feel I must mention the people who have done so much of this work over the past year, especially:
- Keneisha Perry, who has done just about everything for characters except for creating the basic models for the primary characters, from clothes, to creating secondary characters, to rigging and character animation.
- Bela Szabo, who is still finishing the last few of the main character models (we have the six characters from the two families, with only Allen and Sarah remaining of the colonists).
- Travis Souza, who not only created numerous sets for Part 1, but also created the Mooncrete elements for the colony.
- Johnnie Wilson, who made almost all of the sets needed for the media montage at the beginning of Part 2.
- Chris Kuhn, whose brilliant mechanical modeling work has provide the flagship look for this show since he joined the project in 2013.
- Sathish Kumar, who has come back to the project multiple times since the sprint in Summer of 2012, most recently to complete the aquarium room set.
- Rosalyn Hunter, who handed me the script for episode 3, “Cyborg” just this Christmas.
What remains for Part 1 is, I think, a lot less, but still substantial:
- A great deal of tedious “polishing” of models — drawing and painting textures, animating 2D displays, adjusting lighting and shading models and so on. And in a few cases, actual fixing of modeling errors (There’s a problem with the launch gantry that still hasn’t been edited into the production set, although the correction has already been made. And there are geometrical problems remaining with the interior of the Soyuz module).
- Finishing the creation of lip-sync data from Papagayo.
- Most of the mechanical animation.
- Fire, smoke, volumetric, and particle effects.
- More character animation.
- Camera and lighting animation for all scenes.
- Rendering — a substantial job which requires more preparation, and probably money for more computer equipment to run it on (if we are to do it in a reasonable time, rather than it taking months to do a single episode).
- Editing of scenes in Kdenlive.
- Editing of foley sound and sound-mixing in Ardour and/or Audacity.
Parts 2 and 3, of course, require even more work.
In the process, we’ve pushed the technology to its limits, and occasionally beyond. It has become clear to me that in the areas of workflow, version control, and distributed collaboration on a longer-format film (or series!) with many characters and sets, the free-software infrastructure around Blender is not adequate. We need more and better tools, with some particular problems to solve:
- Partial download capability, so that contributors don’t have to copy the entire 12+ gigabyte codex to alter a single file. But it must account for link dependencies, to avoid breakage and data loss.
- The ability to correctly handle the “caching” of expensive intermediates — such as rendered frames from Blender source files — which it may be more practical to download than to regenerate (at least, for those without a rendering cluster).
- The ability to MERGE changes from two or more different artists on a single Blend file. Lock-based version control requires too much overhead and reduces the advantages of collaboration.
- Greater robustness with inter-file linking in Blender so that data is not so easily lost if a link happens to get broken by moving files or renaming objects. Having to work around the brittle linking system is also a big time-waster.
- Tools to maintain many characters with identical or similar rigs, and to ensure that rigs remain similar (i.e. automatically catch errors).
This, combined with the ongoing holding-pattern with the Lib-Ray project (yep, still pending) for distributing films, leaves us with a substantial amount of infrastructure to build. The biggest question here is, “Who’s going to do it?” Do I slow down production on “No Children in Space” so that I can work on the software? Do I try to convince some amiable programmer with Python and Blender API experience to write the code? Do I try to raise the money to simply pay a professional programmer to do the work?
None of the remaining work is substantially outside my own skill set, but I’m now facing the simple problem that there’s not enough hours in the day to do all of this work. There’s simply too much of it. Which means I’m going to have to find some help one way or another. Or I’m going to have to accept that it will take much longer. I’ll be working both angles in January.