Bedlam Update 20140311a

More landscape exploration. It’s amazing how quickly I can put to use all the preamble work I did over the summer with perlin generation. I played around with box noise (diamond maps) but honestly the results just weren’t that spectacularly different from what I’m getting now with layered noise.

To which end here, I’m layering perlin noise at a couple of different octave/frequency combinations. The compositing rules are pretty simple because I’m working entirely in grayscale (although Processing does have a habit of throwing in some weird-ass color artifacts from time to time which keeps things interesting.)

After each layer pass I apply a simple convolution matrix to the whole image. These are applied in descending iteration frequency so that the closer to the “top layer” things get, the sharper they are. Pretty simple stuff really but I love the “volcano” accidents. These maps are generated at 1024 so I’d eventually have to serialize these and regenerate higher frequency noise on top of them.

Before I do that though, I have to learn a few new tricks regarding decay and erosion models. There should be cracks, fault lines and rivers in these maps that are missing right now and those are going to be essential for gameplay.

The attention to detail, the addition of cracks, fault lines, and rivers as mentioned above is going to really pull this all together nicely. 

Bedlam Update 2014030902

I’ve reduced the workflow in Enki to taking raw images as input for the various height-maps and other informative structures required for initializing a world build. This made a lot more sense than trying to constantly generate procedural content at the server. Noise support in NodeJS is kind of not great and frankly I didn’t feel like spending the time to get libnoise set up on the server, etc. It just adds a ton of additional pre-requirements to installing the local development build.

This way, we can use whatever tools we’re comfortable with, Processing is my personal favorite for quickly roughing up procedural content and eventually I could see a real need to use photoshop for building the base dataset for the world for Bedlam.

Above are a few samples from the work I did this morning on building a new dataset. The final world dimensions for Bedlam are (quite frankly) astronomically large. So large that I can’t even generate a single world in processing with my current memory setup. Instead I’ll be producing several different land-masses and then stitching them together for the final import set.

You can now paint textures and have them persisted in the database!

In a surprising turn of events we decided to start streaming data from the Enki server into our chunk renderer. The result is above which, I realize looks an awful lot like the earlier mesh data, but I can assure you it’s not.

Pretty soon we’re going to be working on data persistence. Last week I stood up a new NodeJS service to warehouse all our data and keep all the members of the team working in the same space. I’m not entirely sure if this is how Unity saw their tools being used, but I have to say I’m loving it.

There’s been some discussion about whether or not to keep the entire world persistent, as a single player game, with remote save game storage. The more that we work with these network services; the more value I can see in doing things this way. Our world is “procedurally generated” but in a way that’s not really at all like the way Minecraft or Banished are.

We generate a single “bible” map out of a few different permutations and math that I barely understand. Once that’s done we can all work in the same space making changes that get merged together in the single datasource on the server and redistributed.

I keep thinking about the idea that this could be incredibly fun for users. What if we could do global events that change the entire plot of the game mid-stream. “Didn’t finish the main quest yesterday? Tough it’s gone forever. Here have a demon lord!” Crazy, crazy stuff.

There’s also the possibilities for multiplayer that I think are most obvious. We could very easily build match-making into Enki and have massive tactical battles in turn-based arenas without a whole lot of extra coding. I’m sure the discussion will keep going, we’re kind of evenly split on the whole idea of requiring at least periodic connections to the home server and we don’t have any way to manage the cost of the hosting itself.

In any case, the above is an example of the newest version of the editor (now running as a desktop application instead of a web player as previous versions were because of cross-domain scripting policies.) Ultimately I think we’d like to get back to the web player possibility because that would open the door for the Facebook market.

Geoff made a really good point. Requiring the user to keep clicking to cycle through tile art is dumb.

I knew I could do better so I sat down this morning and crafted the new “Chooser” palette. Once you’ve assigned a tile set to a layer, whenever you have the paint tool active you get a list of tiles you can paint in to the map. This works out really well when you’re trying to plan and I know that we can go one step further (and even show you the tile that will be dropped before you paint it down.)

Geoff made a really good point. Requiring the user to keep clicking to cycle through tile art is dumb.

I knew I could do better so I sat down this morning and crafted the new “Chooser” palette. Once you’ve assigned a tile set to a layer, whenever you have the paint tool active you get a list of tiles you can paint in to the map. This works out really well when you’re trying to plan and I know that we can go one step further (and even show you the tile that will be dropped before you paint it down.)

As a kind of testament to the amount of code we just wrote; this most recent screenshot demonstrates our editors new ability to modify the tile maps we’re creating. Right now we’re brute forcing the placement of tiles but in the future we plan on having some moderate heuristics in place to speed the process up. The above also demonstrates layer isolation; which is one of those things you don’t really think about unless you do graphics for a living. :)

The way tile editing currently works is each “Texture” group contains a stylesheet that defines the available tile sets. The texture is then constructed from “Layers” and each layer can have a single prefix-type (tile set) that can be painted.

When you paint into the surface we place the midpoint tile (solid color.) You can then use the turn tool to swap out the specific tile inside of the prefix set to get things looking the way you want to create a shape.

Blog Log #1

Launching your own startup company can be challenging. Everyone comes with their own work experiences; their own ideas of what processes can and should work, usually predicated on what has worked for them in the past.

So when it comes to TUG’s ambient environment sounds there are again a few basic layers I start from.

Distant environment sounds

Close-up environment sounds

Basic wind layer

Musical wind layer