[In this opinion piece, Canadian level/game designer Tynan Sylvester examines discussions with his father over Crysis to discover what the average person actually looks at when playing a game - and it may not entirely be what you expect.]
I built a new PC this Christmas. One of the first games I tried on it was Crysis
. As expected, the graphics were incredible. Beautiful, lush jungles, fully animated grass and leaves, dynamic shadows and time-of-day, strong HDR. Anyone who is familiar with my older levels will know that I love this stuff.
I showed it to my father one day. "Check it out!" I said gleefully, certain that his jaw would drop at this incredible display of computational and artistic awesomeness.
Naturally, the first thing he noticed was that there were plants popping into existence as I moved around and the LOD system recalculated their detail relevance.
Naturally, I was aghast. Doesn't he see? Look at the technological and artistic beauty of it all! Possibly the most beautiful real-time graphical simulation ever created was sitting in front of him, and all he noticed was the very minor LOD popping. I'm trained to deconstruct digital images and even I wasn't registering those little plants swooping in and out of existence.
Obviously there was something different about how my father, a non-gamer, was perceiving the scene and how I was. I realized what it was some days later when playing Assassin's Creed
I was walking out of Maysaf. I got on the horse and turned it around so we could start down the path. I thought I heard something special and interesting... was Al'tair clucking at the horse? I think he was! That's so cool!
Me noticing this clucking and getting an emotion out of it was my experience with the same phenomenon my father was feeling when he criticized the LOD popping. I call it the scaffolding phenomenon.
It is this: players perceive a game-space as a real space, and get emotions from it based on that criteria. Most game development is just scaffolding that allows a basic game-space to be created. The masonry - the stuff that matters and that people see - is that stuff which would be interesting if it were in a real-space
To my father, the view of Crysis
was not a computer-generated image. In his mind, there is no such thing as HDR, normal maps, or physics simulations. To him, the picture of Crysis
is a picture of a real place. Looking at it this way, we can get a whole new perspective on our games.
Just ask yourself - if this were real, what would stand out? What would improve it? All that technological and artistic awesomeness was, for the most part, just scaffolding that allowed the program to build a basic environment.
He wasn't amazed by the jungle because in real life, jungles are common. They appear all over the place. Seeing a jungle is nothing special. Seeing plants pop in and out of existence as they change distance to you is special.
This applies to many games, and explains a lot of puzzling successes and failures in the gaming industry. Consider GTA: San Andreas
. The game's scaffolding is pretty austere. The characters are low-poly and the textures are blurry. The environments are quite empty. On a modern PC platform, the game is not technologically interesting.
But if you think of them as real places, it's obvious that GTA: San Andreas
is more interesting than Crysis
. The places and characters are more familiar, varied, and meaningful. The things you can do are more fun and relevant.
Compare the Crysis
shot above with this shot of the hood in GTA: San Andreas
. The Crysis
shot is a generic jungle. The i>GTA: San Andreas shot shows a group of human beings in a recognizable environment. It suggests a certain cultural context (West Coast gangsta) and a certain type of relationship between these characters. It suggests things you can do in this environment, and what they mean in the established cultural context. It suggests fantasies you can fulfill.
GTA: San Andreas
did well because the developers focused on creating lots of good masonry and didn't spend all their time building scaffolding. Consider the much talked-about having-sex-with-hookers game mechanic. Technologically, there's almost nothing there. You stop near the girl, she does a distance check, and runs her get-in-car animation.
You drive somewhere secluded, the game does another distance check, then it periodically nudges the velocity of your car for a few seconds as you gain health. Then the hooker gets out and returns to her standard AI routines. No special animations, not much sound, minimal interaction.
But if this were real - that's a pretty cool event! It's a story that the player can feel. "I picked up a hooker, had sex with her, then killer her and took my money back" is a lot more relevant and interesting than "I drove through a jungle and killed some soldiers".
Players don't notice scaffolding, they only notice masonry. They're a lot more forgiving of simple or crappy scaffolding than they are of simple or crappy masonry.
So stop fretting about how perfect your scaffolding is. Normal people don't care. They don't even see it. To normal people, the stuff in the game is real, and will be judged as though it were real.
[Tynan Sylvester is a Canadian level designer and game theorist. He got his start in 1999, working on the popular Tactical Ops mod for Unreal Tournament. Since then, he has worked on levels for Groove Games, Digital Extremes, and Epic Games.]