Making a Mass Effect Special – It’s Laborious
I have made quite a few Mass Effect Specials already, both short and long ones. Most are about romancing the different characters in the Trilogy, but I have also planned to add Lore Specials about weapons and biotics. Maybe I will do some Specials about crazy decisions, epic fights or engineering tricks later, but for now I am concentrating on romances.
The reasons why I create Mass Effect Specials are:
- They are fun to watch,
- show specific aspects of the Trilogy without having to do a full play-through and
- allow me to show scenes that I do not include in my Let’s Play series.
But do you have any idea how much work is involved in making a Special, any Special?
The Main Difference Between Episode and Special
Before I can shoot an episode for LoSI, there might be updates to be installed, but that’s it, mainly. With booting, updating, a little planning, playing, and organizing the clips afterwards, an episode is shot in under three hours.
But for a Special I need many shots that are scattered throughout a game, two games, or even the whole Trilogy. This means I have to find the nearest Save Game to a scene, get there, do the shot, next. Additionally I am very strict on my character choice. This means, that all the shots must be made with the same character that I played throughout the Trilogy. You can imagine that this is more complex than just filming a Let’s Play episode, can’t you?
Managing Save Games
In September 2014 I got really fed up with the Origin cloud saves of Mass Effect 2. I messed up a game, but lost the Save Game before I messed up, because the cloud did not have it either. Don’t get me wrong, the EA cloud is a great backup, so if you mess up your computer, you get your current state back. But I was only keeping the last four saves and did not realize that deletions would also be mirrored in the cloud. So this was absolutely my fault, as I overestimated the possibilities of the cloud backup.
To fix that situation I created git repositories for my ME1 and ME2 saves. Mass Effect 3 followed in Mai 2015. Since then I commit all Save Game deletions and creations to a git repository, alongside with configuration changes and my notes about each individual character and career. I even created a branch one time to test something out that could be merged to master.
Using a git repository for each game gives me the power to “time travel”. If I need to film a specific scene, I can browse my commits, find the one where I played about the time this scene happens, checkout the commit, copy the relevant Save Games elsewhere, checkout the current state, and move the Save Games back in. I do not even have to exit any of the Mass Effect titles for that, this can be done on-the-fly.
Making a Mass Effect Special
A Special is basically one topic or motto that is presented in a concentrated way. Therefore the first thing to do is being clear about what I want to present. When I am clear what the result should be, I make a short plan of what has to be in there. After the list is ready, I can start shooting the clips I need.
But most of these clips can not be shot straight away. Although I can get any Save Game back that went into the git repository, it is rare that a Save Game was made right before one of the scenes I made. So basically I have to load the nearest Save Game of the character featured in the Special, and then play through the game to the point were I have to start recording. To speed this up I usually set the difficulty level to Normal or Casual, so combat would not cost too much time. Those sessions are about reaching specific scenes, and not about playing the game. Hopefully you can imagine that this feels a lot more like work and a lot less like fun.
If you wonder how much is really involved, then please know that the “Tali’Zorah Love Story” needed 67 clips to be shot throughout all three titles. A regular Let’s Play episode normally has 10 to 16 clips.
And while shooting that episode is done in one session, a vast Special needs many session. The story about Tali is very lengthy, so it took two weeks to film all clips.
My latest Special that I have just finished filming, is a lot shorter, because the character is introduced in Mass Effect 2 and a lot scarcer than Tali, but still needed 38 clips. These clips have been taken in three sessions with a total of 14 hours of work. The resulting Special has a length of about 76 minutes, while Talis story has a length of 6 hours and 3 minutes in total.
Cleanup and Convert
What Episodes and Specials do have in common, is, that all clips get cleaned up and converted to a lossless format. Cleaning up means, that duplicate frames get removed and substituted by a calculated frame between the previous and the next. This way FPS drops aren’t that visible any more, and any “stutter” of the clips get reduced. Normally you would not notice this substitution, which is the goal, but sometimes it is obvious.
The conversion to the lossless format “UT Video” is done, because it is far more predictable and reliable when editing. Also the quality of the rendered result is higher when going from there.
But I can not edit these huge video clips, my hardware just isn’t powerful enough. The editing software I use, Shotcut, allows to render low quality “proxies” to edit with. These are small representations in low resolution and quality with which editing is very fast and smooth. But creating these proxies also costs time.
The 38 clips of the Jack Special I mentioned above, have been cleaned up and converted in just under 20 hours. Any clip that is finished converting, is added to my Shotcut playlist and the proxy is made. The creation of the proxies took another five to six hours.
Editing an Episode is relatively easy. The clips are already in the correct order, there is not much need to shift things around. Also they are linear, so besides editing some stuff out, like shortening elevator rides, there is no real challenge.
Creating the intro from the clips of the previous episode is quite some work, and getting and adding the frames for the credit roll is, too. But beside that, making a regular episode is fairly easy.
A Special, however, has many gaps between the clips. So the transition from one scene to another might not be possible without adding extra stuff as a kind of a bridge. That’s the reason why the full Tali’Zorah Love Story has seven episodes.
If you take a look at the special episode 8 of the Tali’Zorah Love Story, which is the whole story without the bridges and fillers, you will see that this condenses the story so much, that it might not make much sense to someone, who is not familiar with the Trilogy.
However, this condensed episode of the Tali’Zorah Love Story took about three times as long to edit than a regular Let’s Play episode. So even without the extra story elements, it is more work to achieve a result.
Before I can render a video, I have to make sure that the audio is okay.
A timeline like you can see above, is not rendered directly. With shotcut, every project is stored in .mlt files. These can be added to a timeline, too. So for every project I add a “render project”, which just concatenates my ept mini-intro with the .mlt file of the video. On the latter I put a two-pass-filter. This filter analyses the overall audio level, and can tell me how high the whole video spikes, what the average loudness (in LUFS aka LKFS) and what the range is.
The goal is to have a final result that averages about -18 to -20 LUFS, and spikes around +1.0. This way the average audio is clear on most viewers, even on cell phones, while loud scenes do not distort. Distortion would start if we went higher, so that is where my limit comes from.
Making a Mass Effect Special – Not So Fast!
A Special consists of scenes from various places and times throughout the Trilogy. Even if it is one that only takes place in one part of one game, it may still consists of scenes from various chapters.
Unfortunately this means that clips get combined, which play in totally different settings. This can result in a final video, that averages below -21 LUFS while still spiking to +1 LUFS. But a final result of averaging way under my target of -18 to -20 LUFS will be quite annoying to watch. You would have to view it with a rather high volume setting, so spiking action scenes will blast you.
In this case there is only one chance. I have to add a two-pass-filter to every clip to near their averaging a bit, and to lower the spikes.
Example: The Special I am currently work on, the Full Jack Love Story, averages at -23.65 LUFS, while spiking at 0.99 LUFS. This is not acceptable, because I can only raise it by roughly 2-3 LUFS, which will result in a rather quiet video. So I had to add two-pass-filters run analysis on 98 clips and 4 flacs. This way I could see where the highest and lowest spiking clips average per scene, and calculated a target LUFS for each filter so that the whole video results in silent scenes being reduced away from their low spikes a little, while loud scenes being reduced more to cut their massive spikes. This way the final filter can then achieve my target of something around -18 to -20 LUFS. In the video mentioned, the final target was -19 LUFS, and the new spikes just do not distort, and the quieter scenes are fine to watch without having to raise the volume on your speakers.
Why not target the goal on the single clips?
This was one of the first mistakes I made. When adding two-pass-filters to the individual clips, these belong to groups that play in the same setting. All these single filters have to be consistent over the group in what amount of LUFS they raise or lower. Otherwise the volume awkwardly changes within a single scene, just because it was made of several clips.
There is some freedom between scene changes, but still you do not want a significant change here either.
The only solution is to use single clip filters to even out oddities to get consistent scenes, while leaving the overall raise to the render target, which applies the filter over the whole of the video.
The Bottom Line
It is absolutely okay to be impatient while waiting for the next Episode or Special to be released. I am kind of impatient myself. But I do hope that you now understand what keeps me so long. Yes, I like to do this and will continue. And no, I am not complaining, only explaining. With more spare time and better hardware this would be quicker, only that both are no option here. Therefore, if I ask you for some patience, please don’t be angry with me, now that you know what it takes to make a Special.