Helldivers 2 bullshittery - Update #4

Good evening all. I'm back from a family BBQ. Lulu got lots of cuddles from various family members as she was passed around from lap to lap. And she came back thoroughly full of chicken and cheese.

She's currently flopped to my left, utterly tuckered out from all the fun.

Now I have a bit more time to talk about the development of HELLDIVERS 2 bullshittery. To go into the details of how the editing went.

As a project, that turned out to be quite refreshing. For regular Patrons know that I often struggle with some video games that are not amenable to easy keyframing.

HELLDIVERS 2 however was not that at all.

In fact, the presence of very distinct helmets with unbroken shapes - and even sometimes coloured patterns such as yellow stripes - meant that the autotracking was easily able to handle most of it. I'd say a full 70% was automated keyframing.

Note how easily I was able to auto track both the smooth shape of my helmet, and the text above Harry's head. The contrast between the target and the background is often excellent.

Of course, sometimes autotracking can come with distinct disadvantages. For many of the scenes featured quite aggressive head bobbing, as the soldier on the screen slogged around in battle, performed dives, or generally reacted to being shot.

As a result, it was sometimes wiser to step in and do things manually. Placing down keyframes every 4-5 frames, and then generally smoothing out the motion by hand eye coordination. You can see the comparison here, in that scene where we "have the element of surprise". The left being how it looked originally. And the right being my rework during quality assurance.

Often it just looks nicer when you step in and do it manually.

On the extreme end of the spectrum, sometimes the tracking became nauseating. A huge amount of motion that became very difficult to ready on a glace. In which cases I decided instead to just shimmy the visuals to show something else. Then leave the text entirely static. Sometimes the footage is just too frantic. And with livestreams, there are no reshoots. I can only work with what I have.

As with all projects, I mentioned the "masking" method. This involves placing a copy of the footage at the highest level in the hierarchy (the foreground, essentially). Then cutting shapes out of it using the Pen tool. This sells the illusion of the text being in the game itself, by making some foreground objects stick in the foreground.

Of course, this can't be done for everything. It's definitely a case of picking ones' battles. For geometrically complex shapes, bodies especially, take ages and look imperfect if you scrutinise it in slow motion. Anything with weird shapes or lots of holes. Often its a case of doing your best and trying to keep it limited to a few seconds.

When you don't need to be precise however, the masks do let you put in a "feather" effect. Softening the edges. I used this on the comp with Tom blowing up the mines. His name text disappears anyway from that point. But a simple masks lets me transition it away naturally behind dust.

Also you'll notice the fire and lightning effects on some of the text. That's a downloadable effect called "SABER". You've likely seen me use it in Divinity II bullshitteries.

Saber lets you create a series of different glowing effects. I assume it's meant to be for the creation of lightsabers for Star Wars related videos.

But it also lets you setup a custom "core". And that core can be text you've typed. Letting you wrap the same effects around it and cover it in various distortions.

A few adjustments later and you've got moving fire, or snaking lightning. Which I put on the text tracked to the head of the tesla towers and trying to mimic the overall look.

And I added little flare. I would adjust the power of the glow effects to match the intensity of the charge in game. So it looks as though the text is surging in power along with the tesla tower.

Also in a prior update I mentioned the use of Pixel Polly. This is where the text can be made to explode into fragments and fly off in a specified direction. Helldivers has lots of opportunities for it. Lots of things exploding

But I would also try to stagger it in such a way that you could read the text first, before the polly took effect.

For example, the first explosion would strike...sending the text flying. Before the blast wave of a second exploded the text into smithereens.

Also making a regularly reappearance as the PEW text effects all the way back from CSGO.

Those are, mechanically speaking, extremely straightforward. Consisting of just 2 or 3 keyframes to make them fly past the screen.

Where I like to make things fancy is when they stick into objects. To do that, I use a combination of a Null Objects (invisible squares). Which I overlay onto the impact position and manually keyframe. This process might take a few goes to get right. Bodies can be fickle.

Then I make the text layer 3D, opening up orientation options in X, Y Z. And then try to guess where it would be facing. After that, a bit of masking to block out bits of the PEW that might be hidden by the motion, you get this.

On the graph gag about the accountant, special thanks to regular viewer MaddJon, who was kind enough to appear in this bullshittery.

MaddJon is an actor in real life. And on being sent my commission request with instructions, was quick to return with a great many takes. Both serious and comedic . In two different clothing styles for me to sift through and pick the most appropriate one to the music. Many thanks MaddJon.

From there it was a simple case of keying out the greenscreen.

The falling notes in the background came from a simple shot of a single cartoon dollar bill, spinning in the void.

This note was then copied about 10 times, given the "CC Cylinder" effect. Which wraps it around a curved shape, according to dimensions you specify. And some wind motion was added by changing its falling keyframes from linear to bezier. Making nice wide curves.

And lastly, there was that incendiary mine scene. That was fun. And you might actually smile when I tell you how deceptively easy that was. You might have already guessed how.

This composition was made of several stages.

The first was a standard null object, manually keyframed to follow the lid of the cannister. The words "incendiary mines" is parented to follow it.

After that though I get tricksy. The word is replaced with 17 individual letters, marked by yellow null objects now. Each one remains on the cannister and following the motion. Waiting for the opportunity for a landmine to pass their area.

After which the null object changes to a green one. And I manually keyframe its path to where it lands on the field.

But then here's the fun part. What do the incendiary mines all have?

Glowing red lights. Against a dark blue/green background. I couldn't have asked for better tracking points.

From the moment the land mines stop moving, it all becomes automated. Dramatically simplifying the work, as the automated software is able to take over flawlessly. I tell it to track the red light in X square. And it does so.

So there we go. A fun project indeed. Now to focus on preparing the public release version with fixes. Should keep me busy until the middle of next week :)