Random Gooseshittery - Update #4 (Patreon)

Good evening everyone, hope you're doing well. So I've had 6 weeks on Untitled Goose Game bullshittery and here's a breakdown on what I'm doing:

• Week 1 & 2 - Cutting together the whole 36 minute video. Taking extra time to make sure the cut is final. As I was worried about unnecessary parts getting pulled into keyframing.
• Week 3, 4 and 5 - Animating (keyframing) up to the halfway mark of 17 minutes. Making just over 5 minutes a week.
• Week 6 - QA polishing of work complete so far to release at least the first half asap.

On that front, I'm still very much in the polishing. So there's not a huge amount to report I'm afraid.

This basically involves rendering the video, watching it through, making a laundry list of obvious mistakes, then fixing them. Rinse repeat each morning until the video seems ready. Below is an example of the types of mistakes being spotted from a screenshot a couple of days ago.

I'm hoping I can get the whole thing cracked this week. And some of those renders start looking final.

Additionally, I mentioned that I was looking for a slightly rounder font to replace Dillenia UPC Bold for this specific bullshittery. I'm pretty sure I've found it on the form of Vagabond Bold. Similar looking, but less pointy, to fit the soft aesthetic of this particular game:

And I don't even need to faff changing every single line of dialogue. Since more streamlined editing practises have gotten me into a habit of creating as few objects as possible. Therefore a simple expression of .setFont("Vagabond-Bold") snaps everything to the new style (with some additional properties for styling).

To be clear, this isn't a change of font for all bullshitteries. Merely for this one with its overtly colourful and rounded art style.

So with quality assurance ongoing, let me describe what precisely I'm doing when it comes to the animation work. Mostly for those who are new to the channel. How am I spending my time?

To state something I'll repeat again for emphasis, what I'm doing as an editor isn't especially complicated. In fact, by the standards of visual effects, what I do is fairly straightforward. It's primarily a time sink. As it goes with all editing.

So I've already mentioned in Update #3 how I like to cut down everything to remove the "dead air". And present a timeline that's relatively fast moving and comedically energetic. What do I do after that?

The Adobe suite of editing programs come with the ability to select portions of your timeline and automatically export them into Adobe After Effects if you'd like to do any visual effects work. As you can no doubt expect - that's all of my timeline - for my particular style as a Youtuber.

Conveniently, doing this action will automatically replace that part of your timeline with an After Effects (.aep) file. Which is handy. As you don't need to do what you would in other programs, such as churn out video files and import them in. And any changes you make in After Effects update on your timeline automatically.

Before long my entire timeline consists of small pink "comps", now controlled by After Effects. Usually no more than 5 to 10 seconds long. And I'm normally able to animate about 8-10 of those daily. Which collectively is often about 60 seconds on average.

Those who remember The Forest updates can probably appreciate just how damaging any miscommunication between After Effects and Adobe Premiere would be. To somebody with my particular process. The interaction between these two programs is integral to everything I do.

Once everything exists in those After Effects comps, the next thing I do is import a "template" called 00-00. This is something I've gotten into the habit of making at the start of each project. And simply contains some examples of the text I plan to use. All of them in the correct font, in roughly the correct positions, and pre-parented to a null object with anchor points set in the right places.

From that point onwards it's a question of making choices with the scene you've got. For you'd open up the first composition of the day, and ask yourself - okay, how can I make this interesting to watch?

Questions to ask then would be things like:

• Who is talking in the scene? Should the viewer focus on them?
• What's moving in the scene? Is it going to be distracting? Shall I crop it out or leave it in?
• What's being said? How much text? How energetically is it moving?
• If it's too much then would it be smoother to simply highlight an object being talked about? 
• Are there any fun opportunities for making text move behind objects?
• Are there any fun opportunities for particle effects?
• Can I make the text bounce, move, wiggle, or do anything to extenuate a particular speaking voice?
• If I do, does it become distracting/annoying?

From there, one takes the text layer from the template and modifies the "Source Text" property.

Rather than do what I did years ago and make a new piece of text for each paragraph, it's far more efficient to make a single block and then modify "Source Text". Timing each keyframe for when each sentence is spoken.

I find that the trick here is to keep each block small and manageable. People will attempt to read it. So I should limit long unbroken sentences. And keep short double or triple stacked blocks, changing the Source Text with natural breaks for breathing.

Several passes will be necessary to make sure everything is timed perfectly. That no line is spoken out of sync with the audio. Something that is absolutely critical - be it a Youtube video with text - or a Christmas carol sing-along with a bouncing icon. Tale as old as "talkies" - the audio and the visuals must be in sync! Failure here is not even an option.

From there, it's a question of deciding motion.

I've talked a lot in the past about "Null Objects". These are simply invisible squares that have their own position/scale properties, upon which I can parent any text (or anything really). If I move the squares around (manual or automated, it doesn't matter) then simply parenting stuff onto them will share that motion, whilst remaining invisible.

This means that, time permitting, I might have 2-3 Null Objects tracking different parts of the screen. A good example for that might be multiple different soldiers in a shooter game and it's unclear precisely which one spoke when something funny was said. I can crudely keyframe 2-3 squares to see which one looks the smoothest. Then finalise that one.

But on the "how does one make the squares follow stuff", that's the golden ticket isn't it? It's extremely dependent on the game being played, how the joke was recorded. And the myriad of shot choices list above that determine what's going to be tracked.

Sometimes (as is the case with this project) the built in auto-tracker can handle it fine. And you merely need to define a nice and obvious different in colour or contrast. Then perhaps delete a few keyframes to stop it looking too...mechanical? Like the text has the jitters.

But a lot of the time you just need to manually do it. Moving the invisible square around every dozen frames until it's back over the target. Or every 3-4 frames. Or every frame. It's really a case of compensating for movement and is highly dependent on how much is going on.

Often its a case of grabbing a coffee, putting on some chill music, and keyframing it until it looks smooth enough.

Once that's done, its then a case of looking for any "masking" opportunities. Cutting shapes out of the text to make it look like it's moving behind foreground objects. To really sell the illusion that it's in the scene.

One could put down a mask on the text itself. That's perfectly fine. But if you make any scaling (size) changes to that text layer, it'll expand that mask. Potentially ruining it.

A better way to do it, I find, is to copy the scene, move it to the topmost layer (literally the foreground), called it the "foreground mask". And then do all masking jobs on that one layer. Even locking it, so it's at the very forefront and nothing can override it if you make a mistake hours later.

Doing it that way means you can swap out the text. And the job has already been done previously. That hedge will remain in the foreground come hell or high water.

And that's essentially it. This is the basic cycle of editing that needs to happen for each individual scene. Each one averaging about 5 to 10 seconds, building up the whole video in little bitesized chunks of After Effects work. For you're not really dealing with one video, but more than a hundred tiny videos that collectively render together to make one bullshittery.

So my process is essentially:

• Cut the jokes together in Premiere, as described in Update #2.
• Export them in little chunks to After Effects for the addition of text.
• Render, find mistakes, fix them <== we are here

Again, it's nothing overtly complicated when broken down into its components. And I'm not including any particularly fancy sequences that might be dotted around the timeline. But if you've done it right, the basic viewing experience should be much more interesting to view.