Right now I’m leaning toward Run + Gun featuring a commando squad of anthropomorphic(-ish) dinosaurs armed with machine guns fighting an invading hoard of killer robots. We’ll see if the dinosaurs work out (I’m concerned there body shapes will be too wide) but it sure looks cool in my mind’s eye.

After deciding that Farseer was not working out for Run + Gun I looked to the XNA Platformer Starter Kit for some ideas on how to proceed. What you see below is a modified version of that kit. The tile-based collision system from the PSK has been replaced with a hash grid, allowing for more flexibility. For example, moving objects like enemies and bullets can be treated in the same manner as stationary ones like walls and floors. The demo below shows off switch powered doors and a touch activated elevator.

You can also catch a glimpse of the eight direction aiming. Aiming is controlled, along with movement, by the left analog stick of an Xbox 360 game pad. Players can switch to aim mode by holding the left bumper. Aim mode prevents the user from moving and only allows them to change their aiming direction. Pressing the right bumper enters strafe mode where the player can continue to move in any direction but can no longer change the aim direction. Strafing allows you to do things like aim upward then move left and right.

You can see the video here.

The first distraction: the absolutely amazing Ico and Shadow of the Colossus collection for PS3. Having never had a PS2 (both these games were originally released on that system) I missed out on these when they came out. And man did I miss out. The scope, design, music, and pacing are superb and engrossing. In fact these games were so engrossing Run + Gun has pretty much been on hold.

The second distraction is not really a distraction but more just general fatigue with a bonus helping of family obligations. I truly enjoy game development, but family and friends, well, them I love, so they get precedence. It seems like every weekend for the past month or two has been occupied, putting Run + Gun on hold. This is okay though, as it made me realize I was getting a little — not tired of, not burned out on but, well, — tried of working on Run + Gun. What I really needed was to take some time off it and come back fresh.

Which leads us to distraction number three. I’ve been working on a Javascript prototype game called Attack Vector. It’s a canvas-based game that features ballistic attacks and terrain deformation ala Scorched Earth and Gunbound.

The Attack Vector prototype has reached a point where I’ve accomplished just about everything I set out to do. Terrain deformation is in, the physics work nicely and the level editor is effective and simple. It’s been a very successful diversion and the first game project where I really just started things up and got rolling right away. It’s really made me feel like I’m starting to find myself as a game designer and developer.

Run + Gun hasn’t seen a lot of love on the past four weeks but it’s not forgotten. I’m getting back to work on it with some new perspective, a few new tricks and a renewed enthusiasm.

The PSK does a lot of stuff out of the box for me, like load up simple levels, draw enemies, define objects in the world and handle simple physics and collisions. Rather than worry about building everything from the ground I was able to focus on pulling out the existing collision system and inserting my own.

This went quite smoothly and now I’m in the midst of replacing the simple tile based levels from the PSK with more dynamics ones created in GLEED2D.

The number one reason, and I knew this going in, is that Farseer really isn’t built for use in a platformer game. Just getting my Hero class up and running required a few clever short circuits of Farseer’s default behavior. Still, the Hero and Baddie classes were working out quite nicely.

Then I got to guns and bullets and grenades. This is when things started getting wobbly and lots of exceptions started getting thrown. I’m not really sure what caused all this, but it happened on cue once a ton of projectiles were flying around the world. Not really being sure about what was going on was part of the problem. Farseer has a fairly active community and I’m sure I could devote the time to sussing all these problems out but then I’m back at my first reason for getting away from Farseer: it’s just not meant for what I’m doing.

Dropping Farseer sets me back a bit, it being the foundation of my game and all so I started looking into alternatives. One option was X-Flixel, a port of Flixel 2.43 for XNA. This had some appeal to me as I’d created both Defuse and Beat Boost with Flixel so I’m familiar with the API and flow of things in Flixel. Getting X-Flixel set up was a breeze and it just works, just like Flixel, but in the end I’m not sure it really suits my needs. Plus I want to dig into XNA a bit here.

Further searching did not yield another off-the-shelf XNA game engine that caught my eye so the next option was to roll my own code on top of XNA. I was trying to avoid this since I don’t really enjoy writing spatial partitioning and collision systems, yet it seems the best option to realize Run + Gun. The question then becomes: how to divide up space. I was leaning toward quad trees until I came across this wiki on spatial partitioning and its discussion of quad trees:

Don’t bother. They’re not worth it are and generally slower than grids and hash grids.

So that settled it (Side note: I’m not really convinced that is actually true, however, grids certainly are simpler than quad trees. When in doubt go with simple unless that doesn’t cut it.) I’m currently using a hash grid to partition my game world into equal square cells, then using AABBs to resolve collisions. Bitmasking it also coming it handy for collision filtering.

SpyParty really stole the show for me. I’ve been reading little bits about it around the web but playing, oh man, it really cannot quite be described. There is an intensity — my heart was pounding when playing as the spy — that I just haven’t felt in awhile, if ever, in a game. The other things about SpyParty that really hooked my as a game designer was the manual, specifically how it called out (I’m paraphrasing here) the idea, “that the core game mechanics need to be nailed down, super solid and really enjoyable before adding a ton of content and art and polish.” I’m doing this with Run + Gun so it’s good to see that I’m not being some starry-eyed idealists with my approach.

Chris Hecker is the main man behind SpyParty who also happens to write about game design. One article of his that I had previously read called Achievements Considered Harmful? is an excellent piece about extrinsic vs. intrinsic rewards. My quick summary: there is nothing wrong with extrinsic rewards (e.g., Xbox Achievements) but your players will be more invested in the game if you skew toward intrinsic rewards (e.g., in game power ups). This got me thinking about Bastion, and Xbox Live Arcade game I’ve been playing lately.

Bastion is an isometric dungeon crawl ala Diablo but with an excellent narrator who lays out the story and chronicles your deeds as you do them. Since Bastion is an XBLA game it has the requisite twelve achievements but there is so much more. Each weapon in the game has a corresponding Proving Ground. On a weapon’s Proving Ground you may only use that weapon and must overcome three levels of difficulty to obtain three levels of prize. So far I have gotten no achievements related to Proving Grounds but I have gotten a bunch of sweet loot and spent valuable time honing my strategy and skill with each weapon. This type of intrinsic reward has added so much to Bastion on many different levels. It’s giving me more to do, it’s extending my play time, it’s making me a better player, it’s giving me better gear and so on.

As promised in the opening paragraph, some pix.

Simple Baddie from Sean Monahan on Vimeo.

Guns and Bullets from Sean Monahan on Vimeo.

I’ve expanded on my Hero class adding a gun that can be aimed in eight directions and a muzzle (the blue circle) that indicates the direction you are aiming. There are also two aiming modes. The first is run and gun: press the left stick (on the 360 game pad) and the player aims and moves in that direction. The second is stationary aim: hold the left bumper and move the left stick to aim but not move.

This post will be a quick summary of where I’m at along with some details on how I’m currently implementing my hero class.

In my previous post about Run + Gun I detailed all the tools and libraries I’d be using to make the game. Since then I’ve gotten GLEED2D setup and have a basic level loading. The videos on the GLEED2D site have everything you need to get started (there’s even a class for loading levels into XNA!) so head over there if you want to see how that works. Once the level is loaded all the points are added to a Farseer Vertices object and then a Fixture for the ground is created from these Vertices using FixtureFactory. In case you’re wondering this is borrowed from one of the Farseer demos.

The Hero class represents the player in the world. Farseer is really great for rigid body physics but it’s not really designed for a 2D platformer game. To get things to work I needed to hack things in a bit. Fortunately some other folks were not only smart enough to figure most of the details for me but they also posted about it here and here.

The gist of things is that a rectangle is used to represent the Hero’s body (from the waist up) and a circle is used to represent the Hero’s legs and feet. The circle and rectangle are pinned together with a RevoluteJoint. To move I set the motor speed on the RevoluteJoint which rotates the circle causing the Hero to move. The body will never rotate or fall over because I’ve attached a FixedAngleJoint to it.

Two caches, wallCache and groundCache are used with collision and separation handlers for both the circle and rectangle. Using the ground as an example. When the circle collides with an object the Fixture is added to groundCache and the flag onGround is set to true. When the circle separates from an object the Fixture is removed from the cache. If the cache is empty the flag onGround is set to false. Doing this allows me to know when the Hero is colliding with a wall or on the ground so I can do things like enable (wall) jumping.

Code of the Hero class is here. And here’s a short video demo.