How to Make a Trailer for The Legend of Zelda: Tears of the Kingdom
How the hell do you make a trailer for a game as massive as The Legend of Zelda: Tears of the Kingdom? In this article I broke down the key factors that make a game capture job difficult and Tears of the Kingdom has ALL OF THEM:
High volume of bespoke content
Lots of room for player expression
Versatile camera angle options
High randomness and intersecting systems
As far as I know, everything in the world is completely made by hand from the curve of each hill down to the puzzles, dungeons, and boss battles. The new mechanics that allow you to build machines like LEGOs mean the sky is the limit for player expression via things like this Trojan horse. There are loads of intersecting systems carried over from the previous game. For example, you can set grass on fire, which creates an updraft, which you can use to glide through the air, but the fire can then spread to other things which might result in burning the raft you made to cross a river. From the trailer it looked like they had pretty versatile camera tools to get cinematic angles.
Just. HOW!?
Finding the “best” spots in a game this big is virtually impossible, but creating constraints helps guide you to where you need to go.
The requirements of this shot are a beautiful sunrise (can therefore only face one general direction) and has a nice formation of sky islands and ground below.
Good Debug Options
Step one for me would be to ask for at least 2-3 months of time with a team of capture artists to work with. I would also want debug options to make it dirt easy to:
Display level locations on screen
Skip instantly to: shrines, villages, dungeons, cutscenes, boss battles, in-game cinematic moments, and major landmarks.
Fly around the world at great speed
Control the game speed (either for slow motion to facilitate easier capture, or to speed things up for time sensitive things like waiting for an enemy to walk into position)
Control time of day, weather, etc.
Have multiple save games
Reset puzzles and respawn enemies
Make the player invisible to enemies
Hook up a second controller, one for camera control, and one for controlling Link
Record a camera movement and play it back while I control Link
Displaying the level location is so everyone knows where a shot was captured. This makes it easy to skip back there by using debug tools, or flying there quickly. Fun fact: Link's Ascend ability where he shoots upward through a ceiling to what's above started as a debug option to save time. Increasing the speed of the game is great for if say, an enemy has a patrol route and you want to capture in a very specific location. Speeding up the game saves you a lot of waiting time in those circumstances. Weather is of course so you don't literally have to wait on the weather if you want to show a scene with rain or other weather.
How do you pick where to capture this shot!? In this case, this is an early game area which has a very large “hero” structure in it.
In the retail version of the game there are multiple quicksaves per user, but it would be essential to have multiple manual saves to make it easy to reload a scenario and try again. The ability to reset an encounter or respawn enemies would also help save time. For example, in shots where Link is fighting, enemies will eventually die after multiple hits. To do another take requires either loading a new save, or spawning a new enemy. For example, when I worked on Dead Cells, I could click one of my thumbsticks to bring up a menu of enemies and items, select one, and spawn it instantly. This was faster than completely reloading an old save game.
Making the player invisible to enemies or a "tourist mode" is good for scouting out locations with enemies in them. This lets you find camera angles without having to deal with being attacked by enemies.
4, 9 and 10 are all about getting the shots just right. Capture artists will often play the game in slow motion so there's more time to get the camera moves right and coordinate with whomever is playing the character. The ability to hook up a second controller or customize the camera while playing by yourself is essential to getting cinematic camera angles (other than working in-engine)
How do you pick which village to capture in? This one is a bit easier because there aren’t a ton of villages, but I’m sure they all look great in their own right.
Fix it in Pre
So how the heck does one create a plan for this?
For starters, I would do a thorough inventory of the game's systems related to combat, puzzles, and interacting with the world; we want to know all the verbs available to the player (especially the new ones). This is so we know what is new, interesting, and eye catching. For a game this high profile, not a single frame will go unscrutinized, so great care has to be taken to make sure it has the right mix of new and old for the sake of the returning fans and people who might be new to the game.
During this initial stage of production I would also do a thorough breakdown of all the story cutscenes so we know the overall story, and what parts to focus on. There's a ton of spoken dialogue in the game too, but that's not as interesting for trailers (so that's an entire swath of the game we could ignore). While story is a big part of The Legend of Zelda games, it's usually better to paint with a pretty broad brush. This trailer pretty much boils down to: Zelda and Link got separated and Ganondorf has returned.
I'd also want to know about any dynamic scenes where the player is fully in control, but they're in the middle of a story moment happening in realtime. Sort of like Naughty Dog's "Active Cinematic Experiences.". For example, the scene in the trailer where Link rides a horse twirling a spear while NPC characters fight enemies reads like a big story moment where the player is in control.
This shot looks very difficult to capture because of the mix of friendly and unfriendly NPCs.
I also typically like to see as much bespoke content as physically possible such as the aforementioned: shrines, villages, dungeons, boss battles, and other major landmarks. This is important for a trailer which will receive so much scrutiny. Why show a shot of Link just climbing a mountain when you can show that same thing combined with several background elements which tell another story?
For games I work on I'll frequently capture every instance of something like say, a lever to open a door, and then pick the most interesting one for the trailer. This is BY FAR the part of production I can easily imagine time limitations forcing the team to just pick a spot they know about rather than agonize over which is "best" for one moment. There is almost never ONE objectively "good" place for each particular shot (assuming the shot isn't tied to a unique landmark). That said, of course it's nice to be able to create a shot where story, composition, camera movement, NPCs, and weather all align. But when a game is THIS big, you can drive yourself crazy trying thinking there's ONE Korok seed sidequest which will look best for the trailer.
There are tons of places you could capture Link climbing a mountain, but this one is great because of the multiple human-made structures in the background which add visual interest. A relatively easy choice if many alternatives are backgrounds which are barren.
Plan of Attack
This time around they decided to do another epic 3+ minute trailer, which I think makes sense after the massive success of the Breath of the Wild trailer (which people often cite as a favorite). Because they have the brand strength of Nintendo and this is a massive game, a trailer longer than a typical 90 second trailer can signal how massive the game is. Kind of like how a lot of the latter Fast and the Furious trailers balk at the typical 2-2.5 minute trailer and clock in at 3 minutes or even more than four.
After the inventory and catalog it's time to start planning what goes into the trailer based on conversations about the pillars of the game's story and gameplay. These will be a mix of cutscenes, game mechanics, environmental storytelling, in-game story scenes, boss battles, and anything else which differentiates this game from Breath of the Wild.
What I'd typically start with are the unchangeable things like story cutscenes, shots which will go straight from the game into the trailer with no custom capture work at all. This is to just fill the timeline with ANYTHING which I know will make it into the final version in some form. For example, Ganondorf is a big reveal, so there have to be some sections carved out for him in the timeline. Story cutscenes for Zelda and the NPCs can also be slotted in.
There are a bunch of these cranks in the game, but this place is unique and looks very cool. It’s possible this is “raw” gameplay but I feel like Link would be larger in the frame if it was captured from regular play rather than a marketing debug camera.
Next up I'll go through my visual selects and start pulling the most visually impressive things and the newest game mechanics. Here are some things from the final trailer that feel like mandates:
Link using machines he built
Link fighting with fused weapons
Recall ability reversing time for objects
Link riding/gliding on a stone bird
Fighting with friendly NPCs
Link climbing
Link diving/gliding
Epic boss battles
The new abilities are of course the most important because these hooks fundamentally shape the gameplay and differentiate it from Breath of the Wild. Also, things like Link gliding on the stone bird and fighting with NPCs are new hooky mechanics. Things like Link climbing and gliding are the Anchors which tell people this sits on the solid foundation of the first game. Epic boss battles are typically the flashiest parts of these games, so of course they need to go in. You might be wondering why Link's Ascend ability isn't in the trailer. I think it's because that ability takes too long to be featured in a quick shot (though it did appear in an earlier trailer)
In a shot like this, background elements aren’t as important since the characters are taking up most of the attention. This angle makes it clear Link is using a sword fused with a shield.
Timeline of Events
Now it's time to create a timeline with some roughly blocked out ideas. Based on my previous review, this is a very broad timeline:
Slow, world establishing intro (and how Link fits into it)
Link exploring
Uh oh, a bad thing happens
Link springs into action, we see new stuff
Story break
Link springs back into action, we see even cooler new stuff
Story break
Final climax
With a game as enormous as Tears of the Kingdom there are still infinite possibilities within this outline. But to just start SOMEWHERE, I'd go back to my list of mandates and story cutscenes and place text title cards in my editing timeline to roughly carve out time for each of them in the section of the trailer which feels more appropriate. For example, the epic boss battles would come later in the trailer, and things like Link climbing would be better for the intro sections since they're foundational to the game, but not exciting in this sequel. It's more difficult to find a place for the things which fall in between fundamental and SUPER EPIC. But at least now the timeline has less room in it, which will force tough decisions and debates on what should go in and what can be left out.
Capturing Lightning in a Bottle
In the final trailer, out of the roughly 80 shots, here's what I see:
40 story cutscene shots
12 "raw" gameplay shots
26 gameplay shots with custom camera work
Of course, cutscenes are just captured as-is, but the other 38 shots represent a mountain of work as tall as the highest peak in the game. Even looking at the trailer now, I suspect several of those "raw" gameplay shots I'm looking at actually have custom camera work. The clearest sign a shot is "pure" gameplay is when the player character is perfectly centered and the camera moves are perfectly in sync with the player's movements. Especially after playing the game, I can see very few shots that look like they could be captured from regular play (with the HUD turned off).
This is one of the few shots that looks unequivocally “raw” gameplay (with HUD off), because Link is centered low in the frame and the camera tracks him perfectly.
To get the ball rolling, I would first ask the developers if there were any points of interest they were particularly interested in seeing in the trailer (but I wouldn't expect to hear anything since they're busy finishing the game). So I would divide the capture team to explore different parts of the game and create their own select reels. This is to get a variety of gameplay to fill in the montages between the story bits. I'd tell people to favor areas with lots of visual interest and environmental storytelling. For example, critical path story missions, ruins, unique manmade structures, and background elements that can fill the frame without distracting from the primary action of the shot.
This could easily take weeks of time even with multiple people, but would gather together a good variety of gameplay moments. At this stage we're just looking for what should happen in each shot, not worrying too much about cinematic framing (unless thinking about how Link is composed amongst background elements). Where each shot goes in the trailer will be the source of endless debate, but after selecting favorites I usually move shots to where they "feel" like they should go into the trailer based off my outline. A more objective way to do it is to rate the shots from 1-10 based on how new and interesting they are, and then generally put more high numbers later in the trailer. This isn't to say you should front load the least interesting shots (I like to occasionally inject the first half of the trailer with high interest shots) but the "rating" of each shot should trend upward.
Don’t want to show shots like this until very late in the trailer, otherwise where do you go from there?
Crafting Capture
As if just capturing the game wasn't enough, the building mechanics and weapon fusing offer infinite possibilities on top of already infinite possibilities. How the heck do you decide what weapons to make, and what things to build and show in the trailer!?
The trailer mercifully only has a handful of shots with things Link has built, and they're mostly conservative. Link built a carriage attached to a horse, he rides a mine cart, a blocky tank, and he made a structure which spins and shoots lasers. There's a shot of Link wielding a sword fused with a shield, and the arrow he nocks has some element fused with it. Previous trailers which showed things Link built were also fairly simple floating platforms and vehicles.
I would certainly dedicate some time to making devices to showcase in the trailer, but I think it's smart to keep the builds simple, functional, and recognizable (with one over-the-top-laser-firing structure for an injection of visual interest). Better to leave the goofier structures to the community. Several weeks could easily be spent brainstorming stuff for this too, but I don't think it would make sense to go TOO far down this rabbit hole.
Even though the building is THE hook of the game, I think it makes sense to not make it a huge focus of the trailer because it's potentially the most confusing and/or alienating part for people with uneducated eyes. It's a lot to try to cover in one trailer, and people might not recognize that they're player created structures. They might think they're actual vehicles you find in the game. Also, not everyone will be a master builder, and they wouldn't want to make people feel like the game is unapproachable if they don't like building. The previously released extended gameplay video which walks through Link's abilities was a better way to showcase the versatility of these unique powers. The game already has so much broad appeal, it can afford to put this massive system on the backburner in this trailer.
This build is pretty simple and functional, but it gets the job done.
Final Polish
As much work as it is to decide what Link will do in each shot, it's so much more to polish those shots up with final choreography, composition, and camera movements. Some shots will be tied to landmarks, but a lot of them will just be something like: "Link nocks an arrow into his bow while on horseback" (which can be captured virtually anywhere). The variables in a more generic shot like this include: location, weather conditions, time of day, and lighting.
With a game this big there's only so much it makes sense to explore every crevice to find the ideal place to capture a moment. For the sanity of the capture team, I'd suggest test capturing each of these shots in a variety of different places and conditions to see how they feel and look. The closer the camera is to Link (like in the combat shots), the less I'd worry about the background elements. In those cases it might be less ideal to have busy backgrounds which might distract from the primary action.
The wide shots are the ones where it would be most important to explore the game to find good composition and lots of visual interest and balance. These are the shots of Link gliding through the sky, walking through the world, etc. Again, I'd favor the scenes with the most environmental storytelling and history. After all the things are decided it's "just" a matter of executing on each shot with the right composition and camera movement.
For a wide shot like this, the background elements really define the composition and depth of the frame.
The Home Stretch
It's essential to make a shot list for the developers if there are areas which they want to polish up for the trailer. This also helps the capture artists and editor keep track of what needs more attention and what is finished, and of course this helps the producer keep everyone on schedule.
I don't know how anyone else would tackle a project this big, and frankly, I've never worked on a game this big before, so this could very well be a terrible workflow for something of this size and scope. But this would at least be a way to start (and hopefully finish) a trailer for a game this big.
To sum up:
Make good debug tools for saving time and getting cool shots
Take inventory of player verbs and bespoke content
Breakdown visuals and story cutscenes
Consider core pillars of the game
Find the mandatory hooky shots and anchor shots
Create timeline of the very broad story beats
Talk to developers about favorite scenes
Capture brainstorm and exploration
Pitching and rating favorite shots
Sorting shots roughly into their appropriate acts
Explore final choreography, composition, camera movement for each shot
Make a final shot list with every shot
Iterate, polish, tweak until there's no time left
This shot could’ve been captured virtually anywhere because it’s mostly about Link shooting this arrow, but the sunset in the background is very nice
I didn't even talk about music, but of course that needs to fit somewhere into the process. I have no idea if this was custom scored or several pieces edited together. For something with a full orchestra I feel like it's less likely it's custom scored because that's such a massive amount of work. If it's several cues edited together, massive kudos because the music flows so well from one moment to the next.
There's likely so much more that went into making this trailer that I'm neglecting. Either way, my hats off to the team that made this. When Nintendo puts all their trailer making muscle behind a game it really shows. There's a lot to learn from every frame in this trailer, but hopefully this helped you grasp what can be done behind the scenes to make sense of it all without getting super overwhelmed.
