How to make an indie game trailer. An Ice-Pick Lodge Studio experience

How the trailer for Know By Heart was created.

The game Know By Heart was born as a pet-project of the technical director of Ice-Pick Lodge Airat Zakirov. Started a few years ago, the game was first developed by him alone in his spare time, but gradually gained a small team and grew to a stage of announcement and a Steam page, where you can add the game to your Wishlist. We'd be happy if you check it!

So, what is Know by heart?

Know By Heart is a game about accepting the loss, in it creates the illusion of returning to childhood and the subsequent сoming of age, the realization of the short-lived time. The player will be immersed in the life of a provincial Russian town, where the echoes of the Soviet era are still heard.

A short but crucial episode in the life of the town is intertwined with the story of the newly reunited childhood friends. They are about to plunge into memories, reconnect and resurrect their old friendship. However, the joy of the meeting soon turns out to be overshadowed by the fact that the guys are beyond their control.

We hope to write a short series of articles about the project’s backstage that can be useful to start-up developers. There are quite a few interesting topics to discuss:

Is it possible to make a budget motion capture for an indie game?
How to implement voice overs for a game?
What is the relationship between marketing and the number of add-ons?
How do I release the game on Steam?

But we’re going to start with the making of the announcement video.

Although the studio has been in the industry for a long time, we've not previously created video game trailers by ourselves. That’s why the production of the video was learned the hard way.

Perhaps some of the readers might want to look at this process from the inside. In our opinion, the trailer turned out to be interesting and was received quite warmly during the recent announcement.

Chicken or egg?

First of all we had to solve the question that worried us most. What is the primary: music or script? In other words, we make a video with a particular plan, and then we make a soundtrack, or we write the music first, and then we adjust the video series to it.

We chose the first option. The script, then the music. I wanted to give a hint about the history of the game, and without the script, it wouldn’t have worked. In addition, the composer of the project said he would write a soundtrack for any video series. Still, he’d have to be moved a little bit to get into the music.

It says it’s done. The script for trailer was written, but there was a problem in the process. Due to limited resources, we could not do any complex staged scenes. So the entire scenario ended up with fragments of the game itself with minimal changes, which did not allow to seamlessly link individual scenes into history.

That is, the words «to make an indie game trailer cheap» in the title in reality means that the clip almost entirely consists of a demonstration of moments of the game without storyboards and specially made animations and scenes.

Technical details of filming

For the draft version of the trailer, footage was shot with low bitrate and resolution. To give the editor a choice, alternatives were also recorded.

For these purposes, we initially used the Unity Recorder package built into the Unity Engine. There you can record video from the editor with an ideal frame rate (in our case 60) in high resolution (we have 3840x2160).

However, there are also problems. Unity Recorder shoots videos in obscenely poor quality due to the very low bitrate even at the High setting. Therefore, for the final shooting, we integrated a plugin into the project that allows Unity Recorder to record video via FFmpeg.

In addition, we slightly changed the arguments in the integration so that the encoding goes with an even higher bitrate.

Basic arguments used for FFmpeg process (no video size or framerate):

-y -f rawvideo -vcodec rawvideo -pixel_format rgba -colorspace bt709
-color_trc iec61966-2-1 -loglevel error -i — -c:v h264_nvenc
-b:v 70M -maxrate:v 70M -bufsize:v 50M -rc vbr -preset fast

Thus, 30 seconds of 4K video came out on average in about 275 megabytes. The quality turned out to be adequate to survive the compression when recording, editing and uploading to Steam, YouTube.

First draft

The editor put together the first draft of the script and immediately concluded that, despite the good video series, the parts are not connected in the sense of each other and therefore the roller needs a voice-over as a binding element.

This upset us, because the editing’s belief in the necessity of voice-over meant that there was nothing in the ad that worked to transmit the story. We really wanted the trailer to broadcast some simple message with the minimum media, without saying too much.

In search of a solution we remembered the early trailer of the same game, made for internal use. The video was assembled from almost static frames, but unlike the current version worked to transmit the message.

This was largely due to the photographic frames in which the faces disappeared. This image resonates well with the main themes of the game, and we realized that we need to put it in the basis of the video.

After discussion, we chose as the binding element the mirror, which assembles at the beginning of the trailer and breaks at the end, thus representing the stages of the relationship between friends.

The mirror was taken from a slightly modified video game mechanics. As a result, these scenes could be removed at minimal cost.

Music selection

More experienced colleagues who have already dealt with trailers have whispered to us that the decision to go from the script may not always be the most correct way for projects with limited ability to shoot unique footage. So, in addition to adding a binding element, we decided to write music immediately.

This is a key element, 'cause music can:

convey the atmosphere of the game;
outline the general "video drama";
conveys a message by itself, especially if written by the person who composes it for the rest of the game.

Our composer quickly gave out nine sketches. Inside the studio and among folks we voted and chose this track.

https://soundcloud.com/nikolaykashevnik/know-by-heart-soundtrack-announce-trailer-music

Second draft

A new montage was made with a mirror and music. In addition, the editor pointed out which scenes he liked and which should be re-shot: change the angle, turn off the headlights on the machines or otherwise correct. The other things we asked you to replace yourself. So it was decided to move all the dark and rainy scenes to the second half of the trailer to create a general feeling of a bright start.

The result already looked much more complete and gave a basic view of the game.

Timing of the video

Total duration

As the pace of life increases, the length of the trailers decreases. Today, a two-minute trailer is considered too long. We initially aimed in the time of one minute 10 seconds (the PR manager at all offered to shorten the video to 40 seconds, saving only the most interesting parts). In the end, we couldn’t get into that frame, and the length of the final clip was one minute, 22 seconds.

First 7 seconds

There is a rule of the first seven seconds. If the viewer's attention is not captured during this time, the video will be skipped. This is about a person who accidentally stumbled upon a video during Web surfing.

Often the problem is solved quite mechanically: at the very beginning aggressive cutting of frames under the cheerful soundtrack, and then the roller slows down.

We had arguments about the delayed start of the trailer, where the train arrived and the girl came out on the platform. Nothing bright happened, and the scene lasted more than seven seconds.

In the end, we did not radically change anything and confined ourselves to adding a locomotive whistle to the mentioned frames. The reasons for this are:

We didn’t think the trailer would have too many random viewers from the outside and it would be able to turn. The game is not called a «game for all», but its audience is rather narrow.
What we wanted was not so much to make a commercial that lured everyone into the game, but to help people figure out whether they liked the game or not.

Real data

And here is the real view data. Indeed, viewers fall off rather abruptly at the beginning. In the first 7 seconds, approximately 27% of viewers leave. Then there is a rather long plateau, where the interest hardly changes then the credits.

27% seems like a pretty big value. These viewers did not form their opinion, because they did not get to the scenes that give an idea of the game. Therefore, in the release and gameplay videos, we may be solving this problem.

We'd like to add a few words about the shortened version of the video, which is 40 seconds long. The current graph shows that interest is fairly even before the credits, meaning viewers hardly stop watching. In addition, if you perceive the video as a filter that allows you to decide whether to wait for the game or not, a shorter duration is impractical.

The trailer matches the game and its pace in an atmospheric way, and if a person can't watch it to the end, it is most likely not his product. It seems to us that attracting such a user will rather create the opposite effect. Especially after the release, when expectations from the game are very different from reality.

In subsequent videos, gameplay and release, we will most likely try to keep the viewer in the first 7 seconds, probably by adding intriguing footage from the game. But we will not be led by the overall duration, leaving the trailers at the pace they need.

The final version

For the final clip, all scenes were reshot in 4K resolution, 60 frames per second with a good bitrate. A couple of places in the trailer had to be collected in two languages - this is a typewriter and titles with the name of the game.

Both videos (Russian and English) were sent for noise dubbing, which was performed in 2-3 iterations. Only then did the difference become visible, how much the sound complemented it.

It remained to make two covers for YouTube (1920x1080) with the name of the game and the inscription "Announcement Trailer", and then the announcement itself followed.

By the way, regarding the page on Steam. For your game to receive the “coming soon” status, you do not need to have uploaded builds to the platform.

We emphasize that the Steam page is very important at the time of the announcement. It contains a button for adding a game to wishlists. The wishlist is a pretty good indicator of future sales. However, keep in mind that it takes some time to get the page ready and styled and requires approval from the Steam team, and each iteration can take several workdays. In our case, the page was rejected twice. So this task should not be postponed until the last moment before the announcement.

Conclusions

We do not pretend to be universal. Whether to apply them or not is up to you in each specific case. There are some hints for a small indie game with limited resources for filming a video:

the video should start with music;
use a soundtrack similar in spirit to the game, so that the viewer immediately feels the atmosphere of the whole work;
the script may not work the way you thought, because the possibilities for shooting footage are limited;
you can add an atmospheric binding element that conveys a simple message and builds the video around itself;
use noise dubbing, it raises the final quality;
The first 7 seconds are really important. Do not include logos here unless you are Rockstar Games.