It took a few practice runs to get a get a good play-through video of Hanami, but I’ve got a pretty decent run to show. It took me 26 minutes to play from beginning to end, and the original compressed file size was 45GB! Even at this size, the capture lacked a lot of pixelated sharpness, and the smoothness of the light gradients. However, recording an uncompressed file was completely out of the question. I’ve reduced the file size to just 376MB, and the quality is more or less unchanged.

To accompany the video, I’ve worked out a subtitle system to provide a running commentary of my thoughts whilst watching the video back. This commentary contains technical info, inspiration and a few pointless facts about the game that pretty much sum up its development. The subtitles take the form of a typical JRPG dialogue. I’ve drawn several self-portraits of myself pulling various facial expressions to sit next to the subtitle text box:

These portraits sit in the bottom left corner of the video, as this seems to be the place where there is least important content. Text is then written in the box to the right of the portrait.

I’ve used the typewriter animation preset in After Effects to make the text scroll in letter by letter, as it tends to do in video game dialogue. This makes it easier and quicker to read than a huge block of text that appears all at once. I’ve added each portrait image to the video’s timeline individually, and used the opacity settings to define which image is shown at which point in the video.

The actual text is contained in just a few rows on the timeline. I downloaded a script which allows you to import large chunks of text into After Effects and break it up into single lines based on where the line has been dropped. This expression goes in the source text parameter of the text layer:

type = text.sourceText;

m = thisLayer.marker.nearestKey(time);
sentences = type.split(“\r”);
n = m.index;
if (time =sentences.length){n=sentences.length-1}

if (n<1){n=1}
line = sentences[n];

line

You can find details on how it works here. Layer markers act as keyframes to switch between each line. I’ve then set and reset the typewriter start value to type in each line.

I’ve currently added about 15 minutes of commentary, with an appropriate facial expression for each line.