I’ve been a big fan of grammar-based procedural images generation techniques for a long time. This tool has its strengths and weaknesses but allows to create pretty cool looking images.

Vertu Constellation Launch Event from InteractiveLab on Vimeo.

Flash is still worth something.

P.S. follow us on Vimeo, Facebook or Twitter.

Since then FWA has grown into a well-known and respected award in the industry of  Flash sites, my resource has well been forgotten.

But recently (a couple of years ago) favouritewebsiteawards transformed into favouriteeverythingawards featuring really everything on the net and some offline stuff. How can you compare a heavy interactive informative site with great UI and lot’s of videos with a WebGL experiment, or a music video, or an offline installation? Let’s examine this year’s Sites of the Month:

• January, Lego Star Wars — Half game (which is supposed to be multi-player, but I was alone ther), half site with great graphics and animation.
• February, Die Hipster — Looks like a game where I don’t gave a clue what I am supposed to do, so I don’t like it q:
• March, Greenpeace – A New Warrior — This is a typical multi-media content site which is well-done, contains a lot of data and serves some real purpose.
• April, Wall of Fame —  Great multi-user experience. A site with user-generated content which actually has it (the content). I voted for it in Peoples Choice Award.
• May, Pleasure Hunt — Another half game, half site. Entertaining way to present a brand.
• June, 3 Dreams of Black — WebGL experiment/interactive music video. Great demonstration of the technology.
• July, The Museum of Me — This is actually just a music video, not even interactive. But the idea behind it is brilliant.
• August, Sexy Fingers — This site doesn’t load on my machine, but I checked on a colleague’s PC and it looks like complete crap. Seems to be a bunch of sex-related games which are totally not work-safe.
• September, The Planet Zero — This one is awesome! The third half game, half site but uses pre-molehill software 3d rendering engine. ROXIK is the greatest!
• October, Being Henry — Really well-done interactive video.
• November, The Honda Experiment — Some kind of HTML “The Incredible Machines” game made with pop-up windows.
• December, Androp Bell — That’s just an awesome game. Great idea and really well-done.

So, I’ve tried to find out how big teams work on large Unity projects if even we are struggling. But no luck.

Ones were mumbling about source control, separating control over scenes editing etc. That’s kind of obvious and shows that those people don’t have any experience at all.

But recently I watched all the videos from Unite 11 conference. I really wish I could visit it and talk to Unity gurus. But anyway, here are the videos I found useful. Videos where speakers share their experience on working in Unity3D as a team. The information which is practical and useful:

Unite 11: Scaleable Game Development at Schell Games (this video is definitely must see)

1. Unity tends to crash building a large project. That’s why they split the game into several projects: entry point project, assets projects, code project.
2. Entry point project is simple, contains minimum stuff to show loading screen or error screen. Loads asset bundles from asset projects and DLLs from code projects.
3. Content projects got their copy of DLLs, they are used to export assets to bundles and there are external tools to rebuild assets.
4. There’s no single source file in Unity projects.
5. That allows to build custom framework, use namespaces, custom compiler preprocessor and obfuscator (if needed).
6. But there are problems with this approach like some deeply derived from MonoBehavior classes not showing in Unity3D IDE, harder to debug, etc.
7. #if preprocessor instructions are used to exclude platform specific code.
8. Custom GUI with relative sizing and positioning of controls. Saved and loaded using ScriptableObjects.
9. A lot of external data handled by ScriptableObjects which are serializable, editable like Behaviors and not tied to a GameObject.

Unite 11: Postmortem – Smuggle Truck

1. ONE project to build for all platforms. I think the project is not large, so they don’t see Unity crash much.
2. There’s a Level Editor which was used by players to create levels which actually shipped with the release.
3. Asset Settings tool to configure assets importing parameters for every platform. For example, background textures could be half-size in iPhone builds to increase performance.
4. Platform Specific properties can be configured with special Behaviors attached to objects. For example, you could specify object’s size depending on whether the project is compiled for iOS or PC.
5. Custom build process allows them to use the data from 3 and 4 to set needed properties and delete unused assets before the actual build. This ensures that assets from PC build will not go to iOS build increasing its size.
6. At the end of the presentation they advised to have 2 copies of the project for different build targets if you have a lot of assets. Because converting from PC formats to iOS formats takes a lot of time.

Update: recently they released Multiplatform Toolkit to the Asset Store.

Unite 11: Jagged Alliance Online Engineering for Large Web Player Game

1. Large games have different problems: automatic builds, light map baking which takes 4 days to complete.
2. Big monolithic project fails to build. (heh, been there done that)
3. The project is split in SEVERAL small projects: code libraries, asset project(s). This approach allows to share code between client and server. Unity3D .NET project uses 3.5 and can be referenced from 4.0 server code. Asset projects are compiled into assets bundles to be use with code.
4. Programmers don’t spend an hour a day waiting for assets to be imported.
5. Code projects are separate Visual Studio solutions kept outside of Unity3D Assets folder. It allows using namespaces (not for MonoBehaviors though). Also they use Unit Tests with modified SharpUnit framework. There are problems with built-in Unity3D classes.
6. Shared code between code projects and assets projects are separated in yet another project which is referenced by them as DLL libraries.
7. Separate VS projects are configured to build DLLs into Temp folder not to be loaded by Unity3D and causing names conflict with existing source files.
8. There’s an advanced automatic build system which executes light map baking and asset bundles exporting in parallel reducing average build time.

Greg Wilson – What We Actually Know About Software Development, and Why We Believe It’s True from CUSEC on Vimeo.

The talk is rather long but is definitely worth watching.

The author talks about:

• Studies and references of “well-known” information concerning software development. Which are lacking.
• Time estimates and that all techniques and applications to calculate time needed for specific tasks are worthless.
• When refactoring is useless.
• How software complexity increases with adding more features.
• Code metrics don’t work.
• And how we need to ask for the sources of information we get.

The same iPad app made with Flex and AIR 2.7. The first list contains 480 items, the second one 280 items. Scrolling looks smooth. Not as smooth as native one, but looks good.

As you see it runs MUCH better. Scrolling is smooth, transitions are OK. It’s not as fast as native iOS applications but looks like finally you can make more complex iPad apps using Flex than a simple scrolling list.

Adobe did great job with AIR 2.7!

Multi-touch Fluid Pong from InteractiveLab on Vimeo.

What’s more, it is open source and is availabale at GitHub.