The joys of using world space / procedural UVs for Unity3D

NEW, 9 July 2013: I've detailed a different implementation of the same effect, better for texturing smooth bumpy surface / terrain, in another post -- "A Smoother Triplanar Shader." I still think this way is good for some things too, though.

One of the greatest benefits of old Quake-lineage BSP systems is managed UVs for world geometry; move your polygons wherever you want, and let shaders texture and tile them properly. It lets level designers focus on building, and frees environment modelers to focus on geometry rather than the mundane work of UVing and texturing yet another concrete wall.

If you're an indie developer doing the work of both, well, any shortcuts are welcome. And if you're not thinking about these things, then just ask yourself whether the player's going to look at this thing you're modeling for more than a few seconds. Just use a block and let the engine worry about it.

If you're an actual graphics / shader programmer, you can do pretty nifty stuff with this technique: Tom Betts at Big Robot is working on a not-Minecraft, and talks about using "voxel skinning and virtual texturing" which sure sounds and looks rather pretty. It's kind of similar to what Valve used for the caves in Half-Life 2: Episode Two -- let the computers do math and walk away!

But I'm not Mr. Betts, so I'm using a much more pitiful and simplistic thing.

Here's the shader I've been using for my projects, slightly modified from something I found on Unity3D Answers a long time ago. I found it after a lot of fruitless digging through significantly worse implementations -- one editor script destructively re-UV'd all the meshes in your scenes, the other script did some weird thing with material offsets -- just leave that all at the door and use a shader-based solution, trust me.

Level With Me, a post-mortem / some unnecessary notes / dear players, it's no longer okay to not know how video games work.

To be clear, I think all readings of "Level With Me" are valid, even the ones that say it's pretentious (though I think it's a remarkably contentless thing to say about experimental work) and it's indulgent (which is like berating a biography for focusing on an individual). But at least it implies a player's willingness to read the levels, even if they don't like what they read -- assuming they even played it.

Game design relies on a theory of mind for players. By that measure, many mainstream commercial games think players are utter simpletons and strive to explain every single thing -- Arkham City will have the Penguin frequently tell you how upset he is and how many enemies are left in the room; tool-tips will remind you that, yes, that glowing electric plate is electrified -- if you prefer your games to talk down to you, to patronize you, then I'm sorry you're going to be disappointed with this mod.

I was shocked, then, by the most common line of criticism I saw: a refusal to read, an insistence that a level without a puzzle-y Portal puzzle is a bad level. It's like the rhetorical equivalent of donkeyspace. I literally can't go through the mental gymnastics required to conclude that challenge is the only interesting thing about first person single player games. Comments like that make me miss all the people who said it was pretentious; I want a higher level of criticism.

Then I watched a "Let's Play" of Level With Me, even the grueling hour or two where he's stuck at the end of chapter one -- and at the end of the whole playlist, he says he doesn't think he "got it" and wants an explanation. Well, whatever you took from it is what it meant. You don't need me to tell you what it means. (This, perhaps, is what the anti-intellectual "pretension police / gestapo" understand better than anyone else.)

Nonetheless, given his struggle and triumph, I'll honor his request. If you can't bother playing the mod, check out his Let's Play Level With Me playlist on YouTube. Now, here's an explanation of my intent and one possible reading of the mod. There are MASSIVE SPOILERS. You were warned:

Tales from Zobeide

The biggest bottleneck for city generation right now is time; it takes 80 seconds for 2800 buildings. I guess I can just delay the player and make them watch a lot of loading screens? I'm also hoping a "complete" city will have at least 4000, which will probably be about 1.5 million polys -- but they're among the cheapest type of polygon to render, so it'll hopefully be okay. Draw calls seem under control too; I'm merging entire city blocks as combined meshes. (And I'm testing on an integrated card, so it should be fine.)

Postcard from Antiquity

Currently finishing up a geocentric space RTS, another collab with Eddie Cameron, for the Super Friendship Club "Universe" pageant. (Newer interface demo is playable here.) Come make games with us!

Postcard from Yorda

Started working on an Ico-ish game on a whim, now I have a half-finished monster -- all because I have to prove to Unity's Flash export that I'm not going to let it beat me with it's half-implemented functions and dozens upon dozens of undocumented bugs.

Right now I'm just improvising a giant castle thing, and then I'm going to sprinkle some gameplay in there or something -- I've left some empty spaces for puzzles, whatever those will be. Try out the v0.3 prototype here, which'll probably be the last public prototype since so much of this one is about novelty:

http://dl.dropbox.com/u/19887116/yorda/03/yorda3.html

(click to walk, double-click to run, or just sit back and let the pathfinding work its magic)

Postcard from Zobeide

Zobeide is going to be a hypertext FPS about the danger of desire. I have a decent interface with Twine working, but the seamless level-sharing has been kind of a technical roadblock, with the Playtomic API randomly deciding to fail sometimes. Now I'm thinking I need to have a generic backup city to load in case it can't reach the servers. Next, I need to model some more buildings and think a bit more about the interaction / aesthetic.

About "Territory"

Territory will be up in about a month or two -- it's an upcoming project that will pair critical essays about video games with short browser games that make arguments.

Part of it stems from my longstanding distaste of games criticism / writing that doesn't readily offer a translation from theory into practice, an opinion that probably isn't popular with most writers (the idea that criticism must be "useful" otherwise it's just whining) so I won't dwell on it.

Perhaps the hypothesis here is that words alone aren't the best way to discuss games. Discussing games is maybe a job best left for games: instead of writing 750 or so words on a game, a reviewer should make a short game about the game. Readers' comments would take the form of mods and new levels, mutations of existing rulesets and graphics to make a new point. (Eventually, some enterprising online pharmacy will code a spambot that will make spam levels and spam games.)

But for that glorious future to happen, I guess we need to think more about how we "read" games. As game designers, we're always deeply concerned with how players behave and think. Territory might help with that.

Truly, some pieces of games writing really shine with an additional layer of interactivity, although there's probably a tendency among those who call themselves gamers to disregard these interactions as "gimmicks," as if every digital interaction must require repetition, time, and mastery. Without those elements, these aren't games...

... Because games have mechanics! "Games aren't X!" But every time you say that, it doubles as an open invitation to the entire world to make a game that is exactly X. Maybe even 2X.

(A related tangent: these days, toddlers learn how to swipe, pinch, and click, before they ever learn how to spell or type words. What do we make of that?)

Anyway. If Territory is successful, I'll open it up to wider participation and start playing matchmaker for games writers / game developers. For now, I've talked to a few people and I'm planning on making small games based on their essays. It's a test run. We'll see how it goes. If it ends up failing, well, let this post be a testament to my completely benevolent and well-meaning intentions.

Level With Me, a Portal 2 mod

STARING EYES! It's unofficially Radiator vol. 3, and the only mod I'll have put out in the last two years. Watch this space as release is pretty imminent, and thanks for reading. We'll see if people hate it or love it.

EDIT: It's out now. Try it out. A typical run-through for me is about 30 minutes long, so that means it'll take the average Portal 2 player about 1-2 hours. If you get stuck on the first chapter, then just skip to the second, it's really not a big deal.