Not much to show yet, but a brief taster of my latest experimentation in Unity.

Playable version below in the Unity Web Plugin.

Please view the full post to see the Unity content.

Rendering Synthetic Objects into Legacy Photographs from Kevin Karsch on Vimeo.

This mind-boggling technical demonstration from Kevin Karsch et al. from UIUC shows just how far algorithmic interpretation of imagery has come. The possible uses for it are many and varied, but the potential for games has really piqued my interest.

The age of the point-and-click adventure has come and gone, but many will remember fondly the days of Monkey Island, Space Quest, and Gabriel Knight. The defining characteristics of the genre were simple enough: a linear, narrative-led journey through a series of static locations with interactive puzzle elements that had to be completed in order to progress to new locations. With such a limited mechanic it wasn’t the gameplay elements that we loved — in fact they were often obscure and frustrating to the point that they spawned an entire industry of premium-rate help phone services — it was the stories and the sense of immersion that the intricately-crafted settings offered.

Of course, static locations, no matter how elegantly-rendered, were a limitation of hardware capability that vanished with the advent of real-time 3D. From the moment that Wolfenstein and Doom appeared on the scene the days of the point-and-click as the dominant gaming genre were numbered.

The thing that really captures my imagination about Karsch’s technology, then, is how simply it could be used to generate photo-realistic point-and-click adventures from photographic source material. Want to recreate the black-and-white horror of WWI trenches without employing several dozen artists and animators for a couple of years? Or the gothic majesty of 1930’s New York? Grab your source photos, spend ten minutes setting up the geometry of the scene with this software, then drop in your 3D interactive elements and you’ve got the basics of a game engine with a visual fidelity that current top-end 3D engines would be hard-pressed to rival.

Over the past few years one of my major problems with project development has been getting appropriate feedback on which to judge the effectiveness of change. Naturally, this has made me a huge fan of integrating automated metrics into the development process at a very low level, and nowhere more than in my own game development.

Recently I’ve had the opportunity to hit some frustrations with some of the analytics packages commonly used by indie game developers, such as Google Analytics (good event tracking, patchy game library support, massive reporting lag), MixPanel (excellent real-time updates, based around funnel analysis for sales / conversion tracking), and Lumos (nice debug event tracking, immature reporting tools, Unity-specific).

Enter brief.io, my latest project. The goal is to provide platform-agnostic deep analytics with custom reporting functions — in real time — for game and app developers. I’m currently accepting applications for a closed beta period to test library integration for Flash, Unity, and HTML/Javascript, and hope to be rolling out the first deployment in the next couple of weeks.

If you’re interested, head on over and sign up for the beta, or drop me a line in the comments below!

The final milestone is now complete, and Scavenger Wars now has environmental hazards as described in the previous milestone. As this is the final build, I’ve included a web player version embedded below, as well as Mac and Windows builds.

Some caveats: you’ll want to play in a widescreen resolution, either windowed or fullscreen. There’s only the one level, and as it’s just a prototype there are some slight physics glitches. However, there’s enough here to get a feel for how gameplay would be in a fully-developed version. For what amounts to about 14 hours of work, I consider that a decent achievement.

Windows standalone (2.9MB .zip)

Mac universal binary (12MB .zip)

Web player:

Please view the full post to see the Unity content.

Continuing last weekend’s mini game jam, I’m implementing the last few milestones from the Scavenger Wars design document in short sprints. Combat is now complete, as per the milestone from part 5: with explosive missiles and line of sight behaviour for the AI player.

The AI pathfinding behaviour from the second milestone was very slow and clunky, which has been bothering me, so I spent some extra time and now have a re-worked AI that is far more aggressive and capable of much faster navigation.

For the next milestone I’ll be implementing environmental hazards: radiation, electricity and radiation. Initially these will all be visual variations on a simple trigger mechanism which causes constant damage whilst the player is inside the area of effect. I will also add in some exploding barrels which are detonated whenever they’re hit by anything.

1. Area trigger which causes constant damage on contact with player. (40 minutes)
2. Particle cloud rendering with different colours for the different types. (30 minutes)
3. Barrels which explode when touched, causing damage. (40 minutes)

This is the last functional milestone for the prototype, which I will leave in its current state after this change. Over the week I’ll write up some of the game’s features as tutorials in preparation for next weekend’s Reddit Game Jam.

Unity3D has a decent web-player implementation, which requires no more than a button press to deploy. Compared to working with Flash it’s almost insultingly easy. So really it only makes sense to put up a build of milestone 2 to show how it works in practice.

So click below to try it out!

Controls are arrow keys or WSAD. You start top-right under the GUI (an annoying side-effect of squashing the game down to fit in the width of this website). Give it a shot fullscreen as well. Don’t expect great things as we’re only talking about 12 hours of work here!

Please view the full post to see the Unity content.

Milestone 2 is complete, though it took far longer than I expected mostly due to stopping to watch films all afternoon!

The game now has an AI controlled opponent which will seek out collectibles and manoeuvre to collect them, avoiding environmental obstacles. It’s not complex behaviour, and is a little slow (I use the physics engine for movement so was over-cautious with my approaching algorithm), but it gets there. There are some very obvious tweaks that can be made to make it more aggressive and better at navigation, but it’s a good start.

The next milestone — which sadly I may not have time to complete in the weekend — is combat. Specifically, adding a weapons system to allow both players to take each other out in competition for resources. This is fairly straightforward:

1. Add a projectile weapon that fires in a straight line forward from the ship. (30 minutes)
2. Make the projectile explode when it hits anything. (10 minutes)
3. Test: Fly around shooting things. (5 minutes)
4. Make the object the projectile hits take damage. (10 minutes)
5. Make the explosion add a force to anything nearby to push them away. (30 minutes)
6. Test: Fly around blowing things up and yelling one-liners from action movies. (10–20 minutes)
7. Make the AI player fire whenever the player passes in front of it within a certain angle and distance. (45 minutes)

As there’s a couple of hours of work there, and it’s already late in a day I’ve decided to extend my GameDev Weekend in order to complete the prototype described in the design document. That will mean at least one more milestone after this for the environmental hazards, and possibly another for polish and a couple of well-designed levels to play on if I have the time.

A good night’s sleep, another pot of coffee, and it’s time to start the second milestone. Now that I have a working game mechanic, it’s time to add a little competition in the form of an AI player.

AI is a complicated subject, and one that could occupy weeks of development time. I want to skip any complication however and implement something simple enough to provide competition for the player. The basic scheme for the AI player will be:

1. Locate the nearest collectible and set it as destination. (30 minutes)
2. Rotate towards it. (10 minutes)
3. Determine if path is blocked, if so rotate 30 degrees and set destination at distance of blocker. (30 minutes)
4. Repeat step 3 until path is clear. (10 minutes)
5. Calculate distance to destination. (10 minutes)
6. Accelerate for half the distance to destination. (10 minutes)
7. Decelerate for the remaining half distance. (10 minutes)
8. On collision, rotate towards the movement vector and decelerate until stopped, then go to step 1. (30 minutes)
9. On reaching destination, go to step 1.

This should provide a basic, careful AI player which will avoid collision at the cost of being slow to collect. If it does collide with anything it will immediately stop and recalculate its course. (Should the player try to ram it off course, for example.)

With a little luck and some good debugging, there’s about 3 hours of work to this next milestone, so the best thing to do is just jump right into it.

It’s far later into the evening than I’d hoped for at this point, but all of my tasks from the first 6 hour milestone are complete.

Whilst the amount of time I’ve spent working on Scavenger Wars totals slightly more than 7 hours (after spending some time stuck on peculiarities of prefab instantiation in Unity and C#), the total elapsed time is closer to 10 hours. That means my aim of getting another 3 hour slot in tonight is kaput, and I’ll have to cram tomorrow full of juicy game development goodness.

Speaking of which, tomorrow’s goal is to introduce a little competition in the form of AI. Now that the base game mechanic of moving, collecting, crashing and dying works, the next 6 hour milestone will be the complex business of writing a goal-oriented AI with pathfinding.

A reminder that the code I’m writing for Scavenger Wars is available in a GitHub repository. It’s not much use on its own, but once this prototype is complete I’ll be writing up a series of tutorials based on my experiences, at which point it will be a lot more useful.