Share this project


Share this project

Neal Stephenson wants to revolutionize sword fighting video games.
Neal Stephenson wants to revolutionize sword fighting video games.
9,023 backers pledged $526,125 to help bring this project to life.

Technical Update #1

Our initial set of videos did a great job of putting us on the map and bringing in enough donations to get us 2/3 of the way to our goal. As our donor community has grown and taken on definition, we've begun seeing more requests for updates that talk about the technical challenges. The ability to make comedy skits doesn't necessarily go hand-in-hand with game design acumen, some have pointed out.

We're pleased that the campaign has progressed to the point where this has actually become a matter for conversation!

-- We hope it doesn't put anyone off if we continue posting the occasional video with high entertainment value. We trust that even the most technically serious donor will understand that this sort of outreach is necessary in a campaign that pivots on reaching a large number of potential donors.

-- Putting a gifted comic actor in a giant hamster wheel and pointing a camera at him is relatively straight-forward. Answering technical questions in polished-looking video clips is slow work and consumes a lot of resources.

-- While we’re working on a few video clips, we’ll default to writing out our answers. The forthcoming updates might be on the TMI side of things.

Motion Capture and the Combat Tree

In our lingo, MASE stands for Martial Arts System Embodiment and it means the aggregation of code, assets (e.g. animations), and in some cases hardware needed to make a particular martial arts style work in CLANG. The first MASE we are making is based on Il Fior di Battaglia (Flower of Battle) by Fiore dei Liberi (1410). This system is sometimes referred to as Armizare.

(In keeping with our long-term goal of trying to allow a knight and a samurai to get into a fight, MASEs should be interoperable. That's a challenge for later, though.)

Every martial art--at least, every one that we would consider trying to turn into a MASE--includes a finite set of stances. Armizare is no exception. If there is more than one stance, then the system needs to include transitions between stances. In the particular case of Armizare, these transitions are frequently synonymous with sword movements: typically attacks (both cuts and thrusts), sometimes parries, and sometimes just changes in position with no offensive or defensive content. So, as a first move toward creating a MASE, it makes sense to get someone who knows the art in question, put them in a motion capture (mocap) environment, and have them go through the list of stances and the transitions between them. In our case, that person has been Guy Windsor, who runs the School for European Swordsmanship in Helsinki, Finland.

The mocap technology is a sensor-equipped suit from Xsens that we borrowed from Valve (and a second one that we borrowed from Suckerpunch) We shot this video a couple of years ago while we were capturing a number of basic Armizare moves.

Of course, stances and transitions do not make for much of a martial art. There will be a range of attacks, parries, counterattacks, and so on. So as long as we had Guy in an expensive electronic catsuit, we captured a number of those too. That's why we needed two suits: so that Neal could act as his tackling dummy.

We clean up and process these short segments of mocap data into discrete animations embodying stances such as Posta di Donna (where the sword is slung around behind the back of the neck) and moves between stances, such as mandritto fendente (an attack out of Posta di Donna from high right diagonally downwards) as well as various parries, counterattacks, etc. With a little help from the game engine and some clever programmers, these can be strung together according to rules in the combat tree--which is another data structure that makes up a part of the MASE. How you move the controller determines the path that you'll take through the combat tree.

All stances and moves in a system are therefore embodied, within a MASE, as sets of coordinates and paths. Some of those moves are so obvious and basic that even the lowest-level character will have immediate access to them. More advanced moves will have to be acquired over time by learning them from trainers in the game world. Learning a move adds it, in effect, to the inventory of things your character is capable of doing. You gain access to more of the combat tree, giving you the ability to execute a wider range of moves.

If you play Warcraft, you can make an analogy here to the inventory of spells, attacks, etc. that a character acquires over time. For a high-level character, this becomes a vast array of options, many of which are context-dependent in some way. The interface becomes the bottleneck. Actions need to be mapped to keys on the keyboard. When the keyboard fills up, players need to create multiple key maps and switch between them. Much of the screen real estate is taken up with icons, some of which may be grayed out or timed out in a context-dependent manner. In CLANG we seek to replace that style of interface with a swordlike object that you can move through space to select options from your "inventory" in a more obvious and intuitive way.

In-system vs. Out-of-System Movement

No hardware controller can measure every nuance of position and movement. By doing some basic pattern recognition on how the controller is oriented, however, it's possible to make a pretty good guess as to what stance the player wants his character to be in. If the data from the controller matches up with a known stance from the MASE, then we are In-System. The character in the game world drops into an animation that is biomechanically stable and martially sound. Various options in the combat tree become available. For example, if CLANG recognizes that you are in Posta di Donna, it will light up a number of relevant options on the attack star (a user interface widget that we'll talk about in a future technical note), cueing the player that a wide range of angles of attack can be launched from this stance.

An early prototype shows getting into Posta Di Donna and a mandritto fendente:

Sometimes CLANG might not be able to map the player's position onto a known stance from the MASE. In that case we are Out-of-System. CLANG then defaults to using the built-in capabilities of the game engine (inverse kinematics) to make the character stand and move in a plausible way. Many of the offensive and defensive options in the user interface become grayed out, since they can't be launched from arbitrary stances. In other words, being Out-of-System is not a desirable situation when you are playing the game, and most players will quickly learn how to avoid it. A simple interface (a reset button on the controller, for example) can be used to get back In-System when things have gotten into this state.

Sword Sync

Our single most frequently asked question concerns what happens when the position of the on-screen sword gets out of sync with that of the hardware controller, which is something that might happen, for example, when your attack gets blocked and you keep swinging through even though your character's sword has been suddenly stopped. Sword Sync is what we call that. We'll cover it in more detail in a future technical note and video. Other than a good UI that helps you know when you are blocked, one answer is that real swordfighters move in such a way as to minimize the effect of being blocked; and part of becoming good at playing CLANG is learning to be more like a real swordfighter.


Only backers can post comments. Log In
    1. Missing avatar

      Jonathan on

      "More advanced moves will have to be acquired over time by learning them from trainers in the game world. Learning a move adds it, in effect, to the inventory of things your character is capable of doing."

      Could you please explain this in more detail? This is motion game... logically, if I move a sword into a certain stance, and this stance IS capable of being recognized by the video game... my character should go into that stance, right? He shouldn't just randomly not go into that stance because I haven't "learned" the skill yet...

    2. Missing avatar

      James Roy Peters on

      Wow, this looks fantastic! I'm a student of Fiore's works in Santa Clara, California, and it is SO awesome to see the things I'm learning show up in a video game.

      The comment about being prepared for your opponent to block is key. You must always have a plan B, because Plan A rarely works right out of the box.

      I am very excited for this project.

    3. Samuel R. Plymale on

      The combat tree really interested me. Very nice post. :)

    4. Missing avatar

      Zappo Darks on

      Another commenter was discussing the "kinect disconnect" which I agree the Extra Credits video (cited here covered pretty well. I get the impression you're trying to cross the uncanny valley with the swordsmanship by moving to a relatively direct physical translation, while using conventional controls (the -- potentially -- couple of thumb joysticks and trigger buttons available) for moving around, looking around, buying, looting, etc. Is that fair to say?

    5. Missing avatar

      John Haverkamp on

      what you describe really sound fascinating, better than i hoped. can't wait to give it a try. as well as maybe program and mocap other styles into the system with the promised SDK.

    6. Subutai Corporation Creator on

      @John The only advantage "In System" fighters will have is one of technique. If your moves are better, you can still fight out of system, but the UI won't ba able to help would and the animation will be relying completely on IK, which maybe be less splendid than our lovely hand-polished mocap.

    7. Missing avatar

      John Haverkamp on

      I recognize the direction Clang is heading here as necessary, but lament that it will turn into more of a 'sport' than a true physics based simulation where any style can have a go and be tested. I see the mentioned MASE functioning much like 'right of way' does in western fencing - an artificial constraint on combat that evolved to facilitate safety and judging. Don't get me wrong though - I'm in to the hilt and eagerly awaiting the release - (fingers crossed!)...

    8. Missing avatar

      Tadanori on

      This is even more interesting than I had originally believed. You are designing something that will be accessable to a wider range of individuals by having these transitionary stances acting as waypoints in the style. Elegant.

    9. royblumenthal on

      I'm shocked and horrified to find out that the CEO-in-a-wheel video is some kind of a skit! That's just not on! I frankly don't believe you! I think you're SAYING it's comedy to lull people into a state of doziness, and that you're going to attack them with a Clang controller.

    10. Dee Abson on

      "[P]art of becoming good at playing CLANG is learning to be more like a real swordfighter."

      Hmm. So for those that have pledged at a reward level of $25, which eventual delivers unto them a copy of the game, I'll take this statement to mean that if they were to become more familiar with the mechanics of the game, vis-a-vis the various MASE which encapsulate swordfighting styles, they could become much better at the game.

      One might conclude then, that something like, say, a PDF of the illustrated Clang fighting manual at the $75 reward level might assist with this endeavour.

      Done! (and *hint*hint*nudge*nudge* to my fellow Kickstarters).

    11. Subutai Corporation Creator on

      @Ryan. Out of system moves will still function with regards to hits and damage. But our swordmasters feel that trained responses will be more effective. But CLANG will allow you to prove them wrong, if you can!

    12. Missing avatar

      charlieb on

      Your comments about sword sync and in-vs-out of system have interesting game implications where being out-of-sync or out-of-system can mean the difference between victory and loss. I can't think of any other game or system where this sync between the human and the control system is part of the responsibility of the human and thus part of the game. Do you think this is an immersion destroyer or conversely perhaps an aid to getting into the flow state?

    13. Ryan Murphy on

      Sweet! Love tech updates, all kinds of interesting stuff going on here. So out of state attacks and movements will be just directly those movements, the difference will be that they won't get the benefits to hit and damage and stuff that an in-state attack will?

    14. James McKendrew on

      Hey! What's all this technical crap? Where's the comedy?