WEBVTT

00:00:01.000 --> 00:00:04.085
Why can a huge open-world game still feel empty?

00:00:04.085 --> 00:00:06.013
This lab plays with that idea:

00:00:06.013 --> 00:00:07.748
a storage bill on one side,

00:00:07.748 --> 00:00:10.383
and how well a world holds players on the

00:00:10.383 --> 00:00:10.769
other.

00:00:11.219 --> 00:00:14.353
First, the bill. The tool adds twelve gigabytes of

00:00:14.353 --> 00:00:15.921
code, a base of textures,

00:00:15.921 --> 00:00:19.369
about point one eight gigabytes for every scanned prop,

00:00:19.369 --> 00:00:22.628
and an audio package. Four hundred twenty props with

00:00:22.628 --> 00:00:26.014
uncompressed audio come to one hundred forty-two and a

00:00:26.014 --> 00:00:26.954
half gigabytes.

00:00:27.404 --> 00:00:29.628
Notice how small the code slice is.

00:00:29.628 --> 00:00:32.995
In the tool's breakdown, gameplay logic is under nine

00:00:32.995 --> 00:00:35.728
percent of the install. The rest is art and

00:00:35.728 --> 00:00:38.968
sound, which look great but don't change the rules.

00:00:39.418 --> 00:00:42.975
Now the focus-testing dial. Every point of smoothing adds

00:00:42.975 --> 00:00:45.284
point eight five to the bounce score,

00:00:45.284 --> 00:00:48.280
and it also sands down friction from hazards and

00:00:48.280 --> 00:00:51.525
faction camps. At the default seventy-eight percent,

00:00:51.525 --> 00:00:53.522
bounce tops eighty-five percent.

00:00:53.972 --> 00:00:57.161
That bounce number is the tool's own invented Teflon

00:00:57.161 --> 00:01:00.350
score, the share of players it imagines sliding off.

00:01:00.350 --> 00:01:03.661
It's a thought experiment, not measured industry data.

00:01:04.111 --> 00:01:06.106
Try the Arrowhead Niche preset.

00:01:06.106 --> 00:01:09.904
Smoothing drops to eighteen, hazards and factions multiply,

00:01:09.904 --> 00:01:12.736
the install shrinks to ninety-one gigabytes,

00:01:12.736 --> 00:01:15.375
and bounce falls to about thirty percent.
