Arizona Sunshine 2 Launch Trailer
How can it take two whole work days to capture only one shot for a game trailer?
The last trailer I finished in 2023 was the Arizona Sunshine 2 Launch Trailer, which had some of the most challenging VR capture I've ever done. I'm going to break down the 14 second opening shot which took me multiple days to capture across at least 60 takes. This was captured entirely in-game with no help from any scripting or programming help.
The ideas I’m communicating in this shot are (in this order)
This is a VR game with a voiced protagonist
You play with a dog companion
You shoot zombies with a variety of guns
The game has a light sense of humor
Here's the sequence of events in this shot:
The protagonist looks off to the distance
Buddy (wearing two guns on his harness) runs ahead of the player
The protagonist throws a tennis ball for Buddy to fetch
Buddy chases the ball as the protagonist points to the distance
The protagonist turns to the left where three zombies are already shambling towards him
The protagonist headshots the first zombie with a pistol, the second with a Desert Eagle
The protagonist pulls out a machine gun and shoots the third zombie to pieces
The protagonist turns to the right to wave to Buddy who is returning with the tennis ball.
Bad dog, Buddy! You’re not supposed to catch it here! But also, who’s a good boy, Buddy!?
To get an idea of how difficult this shot was to capture, look at this list and think of every moment as a potential point of failure for the whole shot. Before I could even start the shot there were a number of things I had to set up:
Load the level
Use debug options to spawn all the weapons needed
Give Buddy two of the guns (I wanted to show that you can use Buddy as a means to store two additional weapons)
Load the other weapons into my holsters
Pick up the tennis ball at the start of the level
Walk to the location where the shot starts
Set the game to play at half speed
Shoot one of the zombies to make them attack me
Even this initial setup had its own point of failure in the part where I have to make the zombies attack. In this level, those three zombies are lying on the ground until you get close or shoot one of them. To make it easier start the shot from the exact same spot, I chose to shoot one of them because it would be more difficult to walk to the zombies, then walk back to the exact mark at the start of the shot. I ruined several takes before they started by accidentally one-shot killing the zombie by shooting them in the head rather than the arm or leg.
Did I mention I had to do all of this at half speed? Running the game at half or quarter speed is common in game capture because it gives the capture artist more time to react to the randomness of the game or the finicky-ness of the controls. The problem in VR is you have to move your head and hands at half speed too so the motion looks "normal" when you speed it back up in the editing software. If you move too quickly, the focus of the shot can be missed or the motion will look cartoonishly fast.
Sometimes I timed things poorly which meant at the start of the shot you’d see me commanding Buddy to hit his mark.
If I successfully angered the zombies I had to look out into the distance for a bit before throwing the ball. But I couldn’t just throw the ball and have Buddy chase it because in the game takes too long to react to the ball after you throw it. So I had to manually command him to move forward so he’d be in a good position for the shot and THEN throw the ball.
This part of the shot could be ruined in multiple ways:
I forgot to command Buddy to move forward
I performed the sequence of actions too quickly, which meant you could see the VR UI when I commanded Buddy
I moved my hands too quickly, which made the movement unnaturally fast
Buddy caught the ball in mid-air
I threw the ball in a direction which made Buddy run off to the left or right instead of ahead
The zombies got too close too soon, which made Buddy attack the zombies instead of run after the ball
Everyone, you’re too close! You’re supposed to be more evently distributed in the frame!
If I got this part of the shot correct, the next thing I had to do was wait 14 seconds and then turn to my left to see the zombies attacking me. The ways this part of the shot could fail:
I didn't wait long enough before turning left, so there wasn't a long enough pause for the dialogue to play out
I waited too long, so the dialogue had some dead air before I turned to the left
The zombies were too far away when I turned left
The zombies were too close when I turned left
Bad dog, Buddy! Go chase the ball, not the zombies! But also, good dog, Buddy!
Assuming I got everything right up to this point I then had to headshot the first two zombies and then use the machine gun on the third. Here are the ways this part of the shot could fail:
I missed the head of any of the first two zombies
Any of the first two zombies spawned with "fungus" on their head which meant they needed two headshots to be killed
Any of the first two zombies spawned wearing a hat, which also meant they took two shots to kill
I didn't shoot with the right cadence I wanted for the shot (too fast, or too slow)
When pulling out the machine gun my hand was sometimes in the right position which resulted in me pulling out a gun magazine rather than the machine gun from over my shoulder
I was looking too high, so everything was framed too low
Everyone you’re framed too low! Oh wait, that’s a camera operator mistake…
I think there were some more things that went wrong across many takes, but those were the main ones. Assuming I made it this far without any mistakes the next thing was looking to the right to see Buddy returning with the ball. Here are the ways this could fail:
Buddy was nowhere to be seen
Where I was positioned meant there was a big bush blocking my view of Buddy
Buddy didn't have the ball and was just wandering around
Buddy returned too soon without the ball and attacked the zombies
After about 30 takes I had a version of this shot I was very pleased with! I think the best one was actually take 24 or so. Right now, some of you are remembering that I said this shot took about 60 takes.
So where do the other 30 come in?
Well, after seeing this shot, the client felt the level art looked better at the start of the level than it did in this section several meters ahead of the start. Long story short, I had to figure out a way to do this exact sequence of events but at the start of the level instead. This had a handful of extra steps where I had to use debug to fly over to these zombies, anger them, then fly back to the start of the level, and wait for them to get into a good position so they'd be where I wanted them when I turned left.
This new scenario had all the same points of failure plus a few others, mostly around figuring out how much time to wait before starting a new take.
This debug text also occasionally popped up, but fortunately it never ruined any good takes.
This is also when I should mention that at this point in production the sound was already being mixed. This meant I had to match the exact timing of the original shot so I could swap out this shot and have all the bullet hits still match (my mixer probably could've done it with new timing but this was such a layered mix I wanted to give him as much time as possible).
About 30 takes later I had a version of the shot I was satisfied with. I sent this alternate version of the shot to the client and it was in. BUT! At the 11th hour they decided my original of version of the shot had better timing, composition and choreography. Which I’m REALLY happy about because I much preferred this version of the shot. This was by far the most complicated shot of the trailer, but there were many other shots which were very difficult to get right for their own reasons.
Also, credit where credit is due, I would never have conceived this shot if the debug version of the game didn't have motion smoothing on the Desktop view of the game! I had to upgrade my capture rig to be able to run the game with both the headset and Desktop views running at the same time, but the results speak for themselves. So if you're making a VR game do your best to make sure the game is capable of running a Desktop view with a stabilized camera so your trailer can have MUCH better shots that make your game look good!
Fun footnote to this is I inadvertently made a trailer which perfectly fit into the ESRB’s new rule which allows you to have a trailer for an M-Rated game which uses the ESRB icon (with full descriptors) instead of the full screen ESRB rating provided there is no extreme violence within the first four seconds of the trailer.
