Good evening everybody, hope you're doing well.
It's been 2 weeks since I was able to deliver the 20 minute PEAK bullshittery. I come now with the final chunk. The last 5 minutes. Including the whole thing polished, in what I hope will be the final version that goes public:
I present Random PEAK Bullshittery - https://youtu.be/pfrIlbYiSaU
So, a project retrospective. It wasn't quite as easy as I was expecting.
Whilst I was initially hopeful that the bright colour palette would result in an easy tracking job. I eventually had to walk back that assessment. There was quite a lot that the automated tracking couldn't do. Necessitating a lot of manual keyframing work. Probably more than 75%.
Interestingly, I took also took several attempts at using a competing form of tracking called Mocha AE (which I've either purchased previously and cannot remember or comes free with After Effects). And whilst it was occasionally seem more capable than the default tracker, I'm pretty sure it wasn't built with game footage in mind. It quickly floundered under relatively simple in-game curveballs. Such as excessive cameras movement and in game lighting changes/effects.
It also had this nasty habit of keyframe dumping. Where it would stack vast amounts of motion when it really wasn't necessary. Being a tracker for planes, rather than points. Meaning I often had to walk back and run corrections in a manner that...probably would have just been faster if I'd manually keyframed it from the get go.
That old editors conundrum.
On the tail end of the video.
I figured I'd leave with something a bit more memorable. Showing the comedic contrast of a high profile moment on stream where I went back to save ZF Digi. Only for him to leave me to die a short while later on the same level.
I asked two creators to help me here. One was the animator "Pupudimpim" who has also been helping me on the Dungeons & Dragons bullshittery. And the other was ZF Bavon, who gave me several different renditions of the Terminator 2 theme on a sarcastic sounding kazoo.
The result was a short parody.
Throughout the video, you'll notice that we used the in game voice chat. Which is a common fixture of 4 player coop games on steam.
And whilst it certainly helps with audio that sounds in universe, it did make the editing quite tricky at times. For I'm rather used to be able to separate each of the game inputs.
This is typically very handy as it lets me dial additional voices back when they're not needed. Dial the game up or down, etc. But naturally, since the voice lines were baked into the game sounds, I found my hands tied in numerous little ways.
Sometimes the scene would need to end sooner than it would normally. Because the in game sounds would play something awkward.
The rain on TROPICS or the wind on MESA would badly expose any heavy audio cutting of voices.
Or someone said something funny, but they were physically distant or quiet. Meaning I couldn't simply boost their volume in isolation, but would suddenly have to make the wind howl.
I'd also like to draw your attention to these animated intros for each of the biomes we visited. For whilst a couple of these were genuine, I found myself tempted to simply recreate them, to ensure that they were all present for consistency.
In the instances where I made them myself, making any text appear behind them was simple. Just a case of moving the text layer down.
But if I wanted to do it for the existing text then I needed to MASK it. Cutting the shapes out around the existing title, in a manner that hopefully nobody will even notice
Interestingly, on doing so I spotted an exceedingly minor game bug. For the foreground text does not block out the player names above heads, despite being in the background. Leading to a distracting series of shapes moving around the titles, in some cases.
But cutting out the affected letters, and then freeze-framing them free of the text, fixed the issue. At least in the instances where they were annoying. Mainly The Kiln.
On that note, "eyeline" draw.
I've mentioned it many times in these Patreon posts. But I consider it a bad habit for any Youtube video to have something wiggling in the middle of the screen. But then to also have motion/text to be wiggling in other parts of the screen.
The viewer, myself included, only has one set of eyes. And the eyeline being continually drawn around the video gets annoying, in my opinion. Unless chaos is the whole point, it should be smoothed and controlled.
Therefore I discovered in the quality assurance that the SABER effect (fire) got very irritating if there was more than one example on the screen. For the constant motion was very distracting. Making me tone it down a bit when it came to the final implementation.
This scene with Quebec was "playing the tissue" is an example of one of After Effect's useful basics - compositions. For the whole screen is a "composition". But there's effectively few limits when it comes to shoving other compositions inside your compositions.
The tissue then, is simply a 500 by 500 pixel composition, with a pair of traced hands. Parented to a null object to float around and follow Quebec. With the tissue being a little white shape layer, made to flutter briefly via keyframes and a rotation wiggle. With the anchor point set up with the fingers.
Better to keep things nice and simple.
On the lava statue segments, originally that whole segment was VERY text heavy. Basically everything that everybody said was represented in the text. My voice especially.
But I quickly realised that it was a suckers move. For everything that's happening is in the middle of the screen. The way in which the character carries the idol meant that there's a ton of motion in the middle of the frame.
This meant that I decided to keyframe the idol around with the text "immune". Auto tracking just couldn't handle it. So it had to be a manual job. But I think it worked out better in the end. And since I thought it worked well, I went back to do it whenever the idol was being held.
Of course, that also had to setup the idol a couple of times to ensure that it didn't come out of nowhere. Showing me learning of its sun blocking properties. And later retrieving it from the end of the level. Introduction, setup, reminder, pay off.
This "Cactus Fight" was a similar example of the text being dialled back to the only moving element. Adding my voice or Quebec's was most a distraction.
And this shot here of the Cactus rolling was a text layer made 3D. But then it's anchor point was set, so that the text was closer to the viewer. Meaning that when rotated, it appeared to be orbiting a spherical point. Giving it a pretend roll.
Add a little circle shaped mask every time it looked like it was on the other side of the cactus, and voila - text rolls away.
This plastic bag of the trail mix is a good example of multiple masks working on concert. For whilst you can cut shapes out of text, that's often spoiled if you come around later and change it.
Therefore what I like to do is duplicate the existing footage. And pin it to the top of the composition in such a manner that nothing can go over it. Then I MASK multiple shapes out of it. Even imitating transparency with a simple opacity fade and Gaussian blur, for the transparent plastic bag.
This shot with Social was a headache. For the joke is that he's blocking my view. In order to mess with me.
But it required three masks for his ghost shape. One for each short clip. And whilst they were not geometrically complex, they moved a great deal.
Furthermore, there's no consistent and unobstructed shot of the path for the sake of tracking. And you can't pick something too far to the side because it gives this weird parallax-looking movement in the final result.
My solution was to do it by hand. Doing the mask element first, so I'd know which part of the word would be unobstructed. Picking a very short word "path", so I could hide the periodic switching of the layers. And then alternating between right or left manual tracking depending on which direction Social floated. Selecting some blob in the background, as close as I could get.
The Percy bark was caught a couple of times on the livestreams. Initially being just text. But then Ilse was kind enough to knock together a quick bit of art one afternoon:
And Digi's shitty mic effect. Holy heck, was a tricky thing to try and manage.
What I wanted to do, was to make text that would warp massively when his mic volume spikes.
After Effects lets you churn out keyframes based on the audio amplitude. And so I generated those, parented them to the effect. Multiplied them by 2 or 3. And then parented all effects to it.
Unfortunately though, I discovered that less-is-more when it comes to these warp effects. The harder it was to read, the more annoying it became.
So ultimately I stuck with a simple combination of Venitian blind effects, static, a bit of blur and only a little bit of warping.
So that's a deep dive on this project. Thank you as always for your extreme patience with me. My plan now is to take a short break. Maybe go fishing. Certainly do some bungalow repairs. And when my editing batteries are recharged, I'll be back for the next project.
Thank you all :)