Although we may replace some of the voices later on, I have now completed assembling the voices for the pilot episode, and I have now started working on the full sound mix with effects and sound. I have also spent a considerable amount of time refactoring the Blender source files for the project — there is a lot of scaling, matching, and linking work that needs to be done. I also recently found a problem with my system that may explain some of my frustrations with Kdenlive.
Refactoring Blender Source Files
We had a valuable learning experience working the “Teaser Trailer” this last Summer, but in fact, we found a lot of things that need closer attention. In particular, the linking of files was not entirely consistent, and there are a number of scale errors between the different files (not just inconsistent scale, but also just elements that haven’t been matched to the plan). There’s also a lot of unnamed objects.
In addition, the Subversion repository is cluttered with unnecessarily duplicated Blender files, owing to my misguided attempt to preserve attribution information in filenames. In the future, we’ll be relying more heavily on Subversion itself for information about who has contributed, and the really definitive information will be kept in text blocks in each Blender file.
This all obliges me to do a lot of tedious clerical work to straighten out the files. This has been very educational, as I’ve learned a lot more about the practical aspects of collaboration in Blender, the interaction of the files with a portable Subversion repository, and the means of editing and fixing links and data blocks (mainly textures).
I also took the time to document this in an article for Free Software Magazine on re-factoring a multimedia project (I’ll post a link later when it gets published).
Now that I’ve figured out what my goal is with the refactoring, I should be done in a few days. I plan to reassemble the shot files for the Teaser when I’m done, and then we’ll (finally!) be able to animate.
Also, I decided to start doing daily modeling exercises in Blender to get myself more up-to-speed with this stuff. I did my first daily project and uploaded it to BlendSwap yesterday. So we’ll see how that goes. Not every exercise will produce a sharable model, but a lot of them will. At least some of these will be prop and set development for Lunatics. So that should be fun.
Sound and Animatic Editing Progress
I got the last remaining pieces of dialog for the pilot episode, and these have been cut together into a voice track in Audacity. So we now have complete “Original Language” and “English-Only” voices. I have combined this with my earlier full-sound edit for the opening of the pilot (from the beginning voice-over to the arrival in Baikonur).
There are a still several voices I’m doing myself in this track. I may want to cut-in replacements later, but I’ve decided to cut this off until I’ve finished the full sound edit for the episode.
These have to be made in combination with the animatics, especially for the pilot where a lot of the story is told visually. We need to match effects to the images and events.
So, I’ve also started editing the “story reel” animatic in Blender’s VSE — after a very frustrating attempt to do it in Kdenlive.
However, I also recently discovered that I have not had OpenGL running on my production computer! This really surprised me, because in the past, that would have been completely unusable. There is a substantial improvement after fixing it, though, as you would expect. This definitely improves the performance of Blender and it may have a substantial impact on Kdenlive (which I believe does use GL rendering).
So that was embarrassing, but a least it’s working now!
