Jump to content

tyjoti89

Members
  • Posts

    1
  • Joined

  • Last visited

tyjoti89's Achievements

5

Reputation

  1. Not sure if anyone will be notified of posts over a year later, but I've put quite a bit of time into thinking about this. VR is becoming much more accessible with standalone versions of the Vive and Oculus not requiring a high end computer. Also, more "lower end" computers are becoming VR capable as standard hardware improves. And VR tech in general is coming down in price as newer systems are being released. Ideally I would love to see this game in AR with the ability to battle other people, but so far I have not seen anything that is accessible. I have several friends who have either a Vive or Oculus Rift, and all of them happen to be big Sanderson Fans. And then there was that little VR game about Stormlight. So things are happening. Anyway, my purpose in posting in this seemingly dead thread is to see if anyone has done any more research on developing this kind of game. Even a simple mobile game where you draw your stuff on your screen and either play against AI or you shoot off lines of vigor at your living opponents on their own devices. I do software development for a living, but this was my first foray into animation. Here are my findings so far: Object recognition is a significant challenge, but not insurmountable. Lines of Forbiddance are the easiest. You must be able to distinguish imperfect straight lines from narrow lines of Vigor. I believe this could be accomplished with some sort of threshold of variance from the vector between the start and end point. Lines of Vigor I would love to hear more from Sanderson (perhaps in book 2) about what happens when a person intends to draw one thing and does a poor job of it. It is clear with lines of warding (bad circles = bad strength) but for vigor lines, if the peaks and valleys of the "sine wave" aren't very consistent does it ruin it altogether or does it change the strength/speed of the attack? Distinguishing between these shapes and lines of Revocation might be a bit more challenging however I think it can be done with what I will describe below for lines of Making (chalklings). Lines of Making (Chalklings) This is the real challenge that has made me start and give up several times on this project. But recently I have been reading more about machine learning and object recognition. I have a pretty rudimentary understanding of how it works, but (and one of the triggers for me posting this) I found someone who is already managing to classify doodles, and their system works brilliantly! Google Quickdraw and checkout their data. If you haven't tried the game, do it and you will be impressed. It won't allow for unlimited creativity like you see in the examples of chalklings from the book but it would certainly make for an interesting game. I think recognizing the commands (go, attack, wait, turn) would be similar, but there would need to be some sort of game mechanic that says (I'm done with my chalkling, now I'm writing commands, so that the object recognition tool can distinguish them. One thing that would be very advanced would be to first, recognize the type of creature, but then also recognize features of that creature, like a shell, or wings or claws to augment the stats of the chalkling in combat. This will also be important in animation (see below). Animation This game would still be impressive if all that happened was a simple translation (sliding) of the drawings on the 2D plane. However what I really would like to do would be to animate as much as possible. I have very limited experience with animation but the basics are this. First You have to recognize the object. Then you have to recognize a few points to build a sort of "skeleton" for the object. Google Human pose detection for an example of what I'm talking about. For Lines of Vigor you would want to recognize the peaks and valleys of the "sine wave" and extrapolate it so that it can "slither" toward the enemy rather than just slide. For Chalklings you would want to be able to bind to points of interest like heads, arms, legs (however many), tails, and wings. Then you have to have an animation profile that knows how to manipulate that skeleton. And while I was thinking it would all be 2D animation, it would actually probably be closer to 3D if you think about seeing a dragon flying towards the enemy from a top down perspective. It would be traveling along the 2D plane, but the wings would likely need to transform to appear as if they are flapping in the 3D plane. But even if the Chalklings could just wiggle around in their simple 2D form that would be impressive. If anyone is interesting in discussing this further, I am all ears. I can't commit a ton of time but it would be an incredible hobby project. Quite honestly, if done right, it could be an incredible PhD project in some particular fields of study. And the tech for automated doodle classification and animation could be used in several different contexts. Reply here if you have any interest
×
×
  • Create New...