next »

[lithander]

Been busy preparing the GC demo so there's a lot of polishing, hacky feature-faking and GUI stuff to do. And thus nothing really novel to present here. But I've implemented some new special abilitys that are quite fun. So for the time being here are some animated gifs:





Btw is anyone reading this going to the gamescom?

[diegzumillo]

Amazing progress! and good to hear this project is getting some boost with the book thing.

When I was playing with these crowd algorithms my plan was to make a zombie apocalypse kind of game, but then the world decided that everything should include zombies and zombie games started falling from trees. So I lost interest. But your idea is way more interesting, direct interaction with the crowd looks much more fun and interesting than avoiding it.

[Mr. Levich]

I'm going to gamescom. Will you have a booth there?

[lithander]

There's no public playable build. But the publisher of our adventure game Book of Unwritten Tales 2 got a booth (Hall 10.1
Booth A060) and there will be a presentation of the same demo that we show to the press (02.1 B040) in the restricted area once per day. I think it was 11 am but Im not sure.

I'm going to do a couple of presentations in the restricted area if you got a trade vistors pass.

[jctwood]

This is progressing so well since you initially posted! Good luck at GamesCom!

[TS42]

The game looks awesome and I love your technical dev blog!
I'm a fan of the Zwerge/Orks/Elfen books, too. Looking forward to play with Boïndil and Boëndal!

I didn't know we have such a nice studio here in Bremen (living near Bremen). Is it possible to visit the studio?

[lithander]

The game looks awesome and I love your technical dev blog!
I'm a fan of the Zwerge/Orks/Elfen books, too. Looking forward to play with Boïndil and Boëndal!

I didn't know we have such a nice studio here in Bremen (living near Bremen). Is it possible to visit the studio?

What we did for Battle Worlds a while ago was that we invited interested people first-come-first-serve spend an afternoon to visit us at the studio and play a pre-release version of the game. We'd watch them play, let them describe their experience and what they liked what they had problems with and just tried to figure out what we could improve before launching. Then there was cake pizza. I guess you could call it a Focus Group Test if you like - it's really helpful as a developer has a hard time to see his own project with "fresh" eyes after while. And it's very likely that we'll do the same thing with Dwarves!

Last time we just posted to the facebook group and our dev forum on a short notice. So maybe you'll want to follow the Dwarves facebook page so you wont miss it! (I can also send you a PM here if I remember)

[TS42]

I think a pre-release study is very interesting and important for development. It points out how customers uses the game/mechanics/menu/software and if they get stuck somewhere or have questions, you get a direct hint where you've to add more info and make stuff more clear.

I'll send you a PM with my email address, because I'm not very active on facebook. Maybe you have an internal player tester mailing list and you want to add me there ;-)

Keep up the good work!

[lithander]

I've been neglecting this devlog a little. One reason is that I've been on vacation and the other that I've started this devlog with a rather technical scope and there hasn't happened much on that front lately. With the gamescom (where we needed a playable demo) and with the upcoming kickstarter (which hopefully goes live next week) we were focused mostly on doing eye-candy stuff and implementing gameplay that helps us to communicate what kind of game The Dwarves is going to be. (both internally and to outsiders)

I jokingly said we're now following a totally new software development method I dubbed "Capture Driven Development" and then was a little surprised to realize that there's more then a grain of truth in that. It's like rapid prototyping but we're not focusing on a playable prototype but on many little, disconnected things that you can screen-capture and that communicate the essence of the game effectively if shown for only a few seconds.

For example so far we had a very simplistic Xbox-Gamepad control scheme implemented that was sufficient to provide input to the core systems to trigger some abilites or move the character. But seeing the gamescom presentations a lot of people were worried about the game playing like a console port on PC. This was a wake-up call because PC is our main platform and if we fail to communicate that we risk that our intended target audience discards the game on a false early impression. So for the upcoming kickstarter video we had to put a bit of work into a mouse driven control scheme, something I would have postponed for a much later time when all the core systems were well in place. But what actually happened is that the experience of controlling the characters in an almost RTS like fashion (by clicking a point in the scene and having him walk automatically over there) is very different from taking direct control with an analog stick. This is now changing _our own_ vision of the game, too. There are a couple of interesting new ideas and - early in development as we still are - we're flexible enough to consider them all. Personally it feels like I'm learning something fundamental about development practices right now even though, had you asked me last year, I would have claimed to know exactly what the best approach to making a game like this is.

Well well... a lot of talk. Actually if you like 'talk' I've got even more. I got asked to post this youtube video.

It's one of the presentations of the game I've given at gamescom. (Actually it's the very first press demo I ever did in my life.) I don't know if it's any good cause I really can't stand to see (or listen to) myself on videos. So I didn't watch it myself! =) (Oh, and be warned it's noisy and in german)

P.S: we are totally blown away by the reception and participation we've seen on the games facebook page. 10000 fans allready and the post with the announcement trailer has been shared over 500 times.

[lithander]

Our Kickstarter Campaign has just launched!

Go have a look - even if you don't intend to pledge - as I think the kickstarter video is really a great introduction to the project.

But of course we're happy and grateful for all support we can get! And if you've got any feedback about the campaign (something important missing? something unclear? just generally tips what we should improve) please take the time and tell me!

[Fenrir]

Wow congrats guys, really nice campaign!

[Sunburned Games]

Great Kickstarter! And in 5 languages, you are raising the bar. Vorsprung durch Technik.

Just backed it!

Hamingja!
Joachim @ Sunburned Games

[TammiDev]

Backed the campaign (whoohoo, early bird, free moneys)! I really liked the look of how you move within a crowd, it looks really natural how the mob walks backwards/ forwards.

[lithander]

Thanks for your kind words and all the support! =) 50k allready and the first day isn't half over.

it looks really natural how the mob walks backwards/ forwards.

I'm really happy that you say that because that crowd-combat system is something I'm very attached to, personally. I haven't seen something like what we plan in other games and I'm glad for the opportunity to work on it. And I've got much more planned, too. So hopefully when we release the game it will feel like I got to contribute something that's more then a rehash/remix of what's been done before.

[lithander]

I'm not cross posting every update from the kickstarter but I think the videos showcasing some voice samples are really cool!

https://www.youtube.com/watch?t=8&v=onNtLE2bui8 (English)


(German)

[lithander]

I've neglected this devlog a little. But today I want to make a post about a feature that was a lot of fun to programm and is now at the core of the HORDE framework: A Dynamically Balanced AABB Tree.

I'm going to explain what exactly that is in a second but let's begin with the problem that it solves: The fact that the performance of the simulation was getting exponentially worse the more combatants (e.g. orcs and dwarves) we were trying to simulate. For example to find all object in contact of a particular one in a naive implementation you have to iterate over all other objects and ask "are you in contact?"... imagine you'd have to deliver a letter by asking all 7.000.000 humans on earth if they happen to be the correct receiver!

So the A Dynamically Balanced AABB Tree is a data structure conceived to make that kind of queries cheaper. It works by putting each object into a box. Thats what AABB stands for. (Axis-Aligned Bounding-Box)


That box is a little bigger then the object so even if it moves a little bit the box will still fit.


By storing each box individually nothing would be gained but the boxes are stored in a tree-like structure. Each box has a parent box that is completely containing it and also a second object.


That parent node might also have a parent and if you move up the hierarchy you end up with a rootnode and an associated box that contains all the dynamic objects you're trying to keep track off.


Now when you want to know if there's an object at a specific position you always start your query at the rootnode. If the position is in that node's box you proceed to evaluate both children. The cool part is this: If the position is not contained in a nodes box you can be sure that all the childrens of this node will not contain the position either.

It sounds like a lot of extra complexitiy but the opportunity to skip over large bunches of objects in your search is absolutely worth it.

The tricky part is to balance the tree e.g. to decide which 2 object-nodes or branch-nodes should share a parent. The general idea here is to keep sum of the sizes of the boxes in the tree as small as possible.

And because many of the objects in question are dynamic and eventually leave their own box the tree is constantly in flux.



When an object doesn't fit into it's own box anymore it gets removed from the tree, a new AABB is calculated and fattened (a margin is applied to make it larger) and then it gets inserted in the tree again. And this is where the automatic balancing takes place... the tree is locally rebuild in a way that it stays balanced which means that all leave-nodes (the nodes that have no children and are associated with one of the objects we try to store) are roughly equally far away from the root node. If you wouldn't consider that metric the worst imaginable tree would be a basically a complicated linked list and all potential efficiency benefits of this data structure would be lost.

But working correctly this data-structure makes a lot of queries like "give me all orcs in this area" fast enough to be used extensively. So in a way this is the foundation of pretty much every more high level behavior including collision detection and response.

[lithander]

New gameplay video!

[Sunburned Games]

New gameplay video!

Erstklassig - Funded! Congrats, and with time to spare for stretch goals.

Best wishes,
Joachim

[lithander]

Thanks Joachim!

[TammiDev]

Congrats on the successful kickstarter! (saw the emails )