In v11 we have:
- Hands for ball-only animations. Use C and T to modify a toss's catch and throw orientation with respect to the hand. For example, 3 ball penguin is represented as "3C{P}" where C indicates a catch modifier and P is the type of the modifier. 3 ball claw is represented as "3C{C}T{C}" where C{C} indicates a claw catch and T{C} indicates a claw toss. "3C{C}" would just be a claw catch followed by a standard, palm-up, toss. Right now we just have claw as an option for toss modification and penguin/claw available for catch. In the future I think fork catches and tosses would be a cool option as well.
- Dwell ratio modification. Use D{dwellRatio} to modify an individual toss's dwell ratio, overriding the global dwell ratio specified for the whole siteswap. For example, 3D{.9}3D{.4} is a half-shower, you toss one ball over the other by extending the dwell ratio on one side of the pattern.
- Started using jQueryUI, liking it so far.
As an aside, I bailed on using the minified JS as it was making debugging a real pain, but I think I'll revisit it sometime in the future.
Right now I'm inspired to go back to the SiteswapGenerator that I wrote a while back. I'd like to bring back the "siteswap explorer" feature in gunswap and perhaps even add a "guess that siteswap" game.
Also added some example GIFs to imgur - http://imgur.com/a/3QcTQ.
Monday, April 20, 2015
Sunday, April 5, 2015
v10
Shoulder injury means I'm juggling less, but also means I'm spending more time on the animator. Lots of new stuff!
I've modified the UI (yet again) to consolidate user inputs in a single text area. Definitely less user friendly, but it will allow for more control. Also added a "Basic" inputs menu where you can just specify the siteswap, prop type, hand movement, and pattern height (ie. beat duration). I got rid of all the bootstrap components since I didn't think they were adding much value and that kind of styling really didn't feel appropriate for this kind of application. Things look a bit uglier for the time being, but I think I'd rather take the opportunity to experiment with the styling a bit more on my own.
The main driver behind changing the UI was to incorporate bounce juggling on arbitrary surfaces. I developed a prototype for this (explained in a previous post) and I was able to quickly incorporate it into the animator.
I had to change the bounce syntax a bit. Like before, you use "B" to denote a bounce. Now you can modify the bounce parameters by using curly braces and then specify the number of bounces, bounce surface order and type of bounce. If no modifications are provided, we assume 1 bounce on the first surface using the "lift" bounce. For example "5B{10L}5B{11F}" means throw a 5 for 1 bounce against surface 0 using a lift bounce, then throw a 5 for 1 bounce against surface 1 using a force bounce.
Surfaces are now user defined. As mentioned above, the "Advanced" menu has a text area where you can specify all the inputs (probably need to write a post outlining the full input syntax). The final line(s) are reserved for surfaces which are defined like this "p.x p.y p.z n.x n.y n.z s" where p is the position of the center of the surface, n is the normal, and s is the half-width. The surfaces are assumed to have an edge parallel to the x-z axis. For example, "0 0 0 0 1 0 5" is a simple floor where the center of the surface is at (0,0,0) and the normal (0,1,0) is pointing straight up. The surface is a square with a half-width of 5m, so its actual dimensions are 10m x 10m.
I also added a GIF writer. If you go to the "GIF" menu you'll see a "Generate GIF" button that...you guessed it...generates a GIF. I've been wanting to build this feature for a while, and I basically just reused the code from here: https://dl.dropboxusercontent.com/u/7508542/sandbox/omggif/index.html. The only catch was I had to set the "preserveDrawingBuffer" property to true in the WebGL renderer. Apparently this may cause performance issues on some machines. I haven't seen much degradation on my machine now.
Also beefed up the build process a bit. Added a concat and uglify task in Grunt and modified index.html to reference the minified output. The whole point of this project is to stay sharp on non work related programming and I'm really intrigued by all the Javascript tools out there. I'm on a refactoring/organizing kick right now, and I think I might give Requirejs and Bower a shot.
I've modified the UI (yet again) to consolidate user inputs in a single text area. Definitely less user friendly, but it will allow for more control. Also added a "Basic" inputs menu where you can just specify the siteswap, prop type, hand movement, and pattern height (ie. beat duration). I got rid of all the bootstrap components since I didn't think they were adding much value and that kind of styling really didn't feel appropriate for this kind of application. Things look a bit uglier for the time being, but I think I'd rather take the opportunity to experiment with the styling a bit more on my own.
The main driver behind changing the UI was to incorporate bounce juggling on arbitrary surfaces. I developed a prototype for this (explained in a previous post) and I was able to quickly incorporate it into the animator.
I had to change the bounce syntax a bit. Like before, you use "B" to denote a bounce. Now you can modify the bounce parameters by using curly braces and then specify the number of bounces, bounce surface order and type of bounce. If no modifications are provided, we assume 1 bounce on the first surface using the "lift" bounce. For example "5B{10L}5B{11F}" means throw a 5 for 1 bounce against surface 0 using a lift bounce, then throw a 5 for 1 bounce against surface 1 using a force bounce.
Surfaces are now user defined. As mentioned above, the "Advanced" menu has a text area where you can specify all the inputs (probably need to write a post outlining the full input syntax). The final line(s) are reserved for surfaces which are defined like this "p.x p.y p.z n.x n.y n.z s" where p is the position of the center of the surface, n is the normal, and s is the half-width. The surfaces are assumed to have an edge parallel to the x-z axis. For example, "0 0 0 0 1 0 5" is a simple floor where the center of the surface is at (0,0,0) and the normal (0,1,0) is pointing straight up. The surface is a square with a half-width of 5m, so its actual dimensions are 10m x 10m.
I also added a GIF writer. If you go to the "GIF" menu you'll see a "Generate GIF" button that...you guessed it...generates a GIF. I've been wanting to build this feature for a while, and I basically just reused the code from here: https://dl.dropboxusercontent.com/u/7508542/sandbox/omggif/index.html. The only catch was I had to set the "preserveDrawingBuffer" property to true in the WebGL renderer. Apparently this may cause performance issues on some machines. I haven't seen much degradation on my machine now.
Also beefed up the build process a bit. Added a concat and uglify task in Grunt and modified index.html to reference the minified output. The whole point of this project is to stay sharp on non work related programming and I'm really intrigued by all the Javascript tools out there. I'm on a refactoring/organizing kick right now, and I think I might give Requirejs and Bower a shot.
Wednesday, March 25, 2015
Genetic algorithm for finding bounce paths
Prototype: Bounce GA
https://github.com/yDgunz/BounceGA
Overview
Bounce juggling against a set of arbitrary surfaces seemed like a cool feature to add to the gunswap animator. However, given the scope of this addition, it seemed logical to first create a prototype that can calculate a single bounce path given a set of surfaces, start and end position, and a number of other inputs discussed below.
Simulating the Bounce Path
In order to simulate the bounce path we first need to define some parameters, primarily the surfaces and the start and end positions of the ball. Assuming the surfaces are flat, we can describe them by a position vector, a normal vector, and a scale. For simplicity we will also assume that the surfaces have an edge parallel to the x-z axis. The figure below (TODO) shows the start position (p0), final position (pT), and a surface described by position vector (Sp), normal vector (Sn) and scale (Ss).
The bounce path begins with the ball leaving p0 with a velocity of v0. Moving forward in time in increments of dt we can calculate a new position p(t+dt) = p(t)+v(t)*dt and a new velocity v(t+dt) = v(t) + g*dt. This applies until we reach the first bounce.
The first bounce occurs when the ball is colliding with a surface .We can determine if a ball and surface are colliding if the distance from position of the ball to any point on the surface is less than or equal to the radius of the ball. We can describe this condition as (p(t)-Sp) dot Sn <= R. The figure below shows this (TODO).
Assuming an elastic collision, the new velocity will be v(t) - Sn*C*2*v(t) dot Sn, where C is the coefficient of restitution (link). From here we can continue to calculate the new position and velocity as described above until any other collisions occur.
Given the correct velocity v0 we will ultimately end up at pT at time t = T. So this whole problem boils down to finding the correct velocity v0.
We could find v0 by generating random velocities and running the simulation described above until we've found a velocity where the final position in the simulation falls within some minimum success threshold. However, considering a three dimensional bounce path with a small success threshold, this method may take too long to find a solution. It seemed perfect for application of a genetic algorithm.
Genetic Algorithm
A genetic algorithm is an optimization technique that mimics the process of natural selection. The approach described below may be more aptly described as an evolutionary algorithm or differential evolution. I'm not really familiar with the terminology and ultimately whatever you call this heuristic, it got the job done.
Let's first define a population of candidate solutions called G0. Each member of G0 is an initial velocity vector for which we run the simulation described above. The members are assigned a fitness which is the distance from pT the ball ends up if it starts with the member's initial velocity.
If none of the members of G0 have a fitness that falls within the success threshold, we need to generate a new population G1. We do this by selecting members from G0 with a bias towards members with a higher fitness. We then randomly mutate these members and calculate their new fitness values. If none of the members of this new generation, G1, meet the success criteria then we repeat this process until a solution is found, or some limit of attempts is met.
Input Syntax
There are two categories of inputs necessary to run this genetic algorithm. We need the parameters for the bounce path simulation (the fitness function) and the parameters for the genetic algorithm such as population size, max number of generations, etc.
The prototype uses the following input structure:
Fitness Parameters
p0.x, p0.y, p0.z, pT.x, pT.y, pT.z, minT, maxT
R, C
numBounces[, bounceOrder]
tossSign, catchSign
normal.x, normal.y, normal.z, position.x, position.y, position.z, scale - (surface 0)
...
normal.x, normal.y, normal.z, position.x, position.y, position.z, scale - (surface N)
maxGenerations, populationSize, mutationChance, mutationScale, initialScale, fitnessThreshold, noGA
https://github.com/yDgunz/BounceGA
Overview
Bounce juggling against a set of arbitrary surfaces seemed like a cool feature to add to the gunswap animator. However, given the scope of this addition, it seemed logical to first create a prototype that can calculate a single bounce path given a set of surfaces, start and end position, and a number of other inputs discussed below.
Simulating the Bounce Path
In order to simulate the bounce path we first need to define some parameters, primarily the surfaces and the start and end positions of the ball. Assuming the surfaces are flat, we can describe them by a position vector, a normal vector, and a scale. For simplicity we will also assume that the surfaces have an edge parallel to the x-z axis. The figure below (TODO) shows the start position (p0), final position (pT), and a surface described by position vector (Sp), normal vector (Sn) and scale (Ss).
The bounce path begins with the ball leaving p0 with a velocity of v0. Moving forward in time in increments of dt we can calculate a new position p(t+dt) = p(t)+v(t)*dt and a new velocity v(t+dt) = v(t) + g*dt. This applies until we reach the first bounce.
The first bounce occurs when the ball is colliding with a surface .We can determine if a ball and surface are colliding if the distance from position of the ball to any point on the surface is less than or equal to the radius of the ball. We can describe this condition as (p(t)-Sp) dot Sn <= R. The figure below shows this (TODO).
Assuming an elastic collision, the new velocity will be v(t) - Sn*C*2*v(t) dot Sn, where C is the coefficient of restitution (link). From here we can continue to calculate the new position and velocity as described above until any other collisions occur.
Given the correct velocity v0 we will ultimately end up at pT at time t = T. So this whole problem boils down to finding the correct velocity v0.
We could find v0 by generating random velocities and running the simulation described above until we've found a velocity where the final position in the simulation falls within some minimum success threshold. However, considering a three dimensional bounce path with a small success threshold, this method may take too long to find a solution. It seemed perfect for application of a genetic algorithm.
Genetic Algorithm
A genetic algorithm is an optimization technique that mimics the process of natural selection. The approach described below may be more aptly described as an evolutionary algorithm or differential evolution. I'm not really familiar with the terminology and ultimately whatever you call this heuristic, it got the job done.
Let's first define a population of candidate solutions called G0. Each member of G0 is an initial velocity vector for which we run the simulation described above. The members are assigned a fitness which is the distance from pT the ball ends up if it starts with the member's initial velocity.
If none of the members of G0 have a fitness that falls within the success threshold, we need to generate a new population G1. We do this by selecting members from G0 with a bias towards members with a higher fitness. We then randomly mutate these members and calculate their new fitness values. If none of the members of this new generation, G1, meet the success criteria then we repeat this process until a solution is found, or some limit of attempts is met.
Input Syntax
There are two categories of inputs necessary to run this genetic algorithm. We need the parameters for the bounce path simulation (the fitness function) and the parameters for the genetic algorithm such as population size, max number of generations, etc.
The prototype uses the following input structure:
Fitness Parameters
p0.x, p0.y, p0.z, pT.x, pT.y, pT.z, minT, maxT
R, C
numBounces[, bounceOrder]
tossSign, catchSign
normal.x, normal.y, normal.z, position.x, position.y, position.z, scale - (surface 0)
...
normal.x, normal.y, normal.z, position.x, position.y, position.z, scale - (surface N)
- p0 - starting position
- pT - ending position
- minT - minimum amount of time for bounce path
- maxT - maximum amount of time for bounce path
- R - ball radius
- C - ball coefficient of restitution
- numBounces - expected number of bounces
- bounceOrder (optional) - comma separated list of surface indices representing the bounce order. So 1,0 means bounce off surface 1 first, then surface 0
- tossSign - 1 means the toss is up, -1 means the toss is down, u means it doesn't matter
- catchSign - 1 means the catch is made while the prop is traveling up, -1 means the catch is made while the prop is traveling down, u means it doesn't matter
- surfaceNormal - vector for the surface normal
- surfacePosition - vector for the position of the center of the surface
- surfaceScale - all surfaces are squares with an edge parallel to the x-z plane, the scale provides the half-width of the square
maxGenerations, populationSize, mutationChance, mutationScale, initialScale, fitnessThreshold, noGA
- maxGenerations - max number of generations to run GA for
- populationSize - total size of each generation
- mutationChance - percentage chance that a member will mutate
- mutationScale - maximum magnitude of mutation
- initialScale - maximum magnitude of newly generated members
- fitnessThreshold - maximum distance from pT to be considered a success
- noGA (optional) - set to 1 to turn off selection by fitness and mutation (ie. iterating through random solutions)
Conclusion
The genetic algorithm has proven to be an adequate solution to the arbitrary bounce path problem. Next steps are to integrate this into the gunswap animator and to figure out the best way of inputting all the necessary parameters. Thus far most of the gunswap inputs have been handled through form elements, but there's a lot here and I may consider reworking some of the existing inputs in favor of a more text based approach like in the bounce path prototype.
v8 and v9
Released v8 a while back, but never made the release notes. Just pushed a v9 tonight for some stuff that was sitting around locally. For these two releases the big updates were:
- Changed prop toss/spin orientation syntax. The siteswap now just encodes the number of spins, so you can say "5S3" for 5 club triples, but you can't say "5S3XY" to define the toss/spin orientation. The hand movement string is now used for the prop orientation. You can now add {x,y,z,th} to the end of each hand movement point to define the axis of rotation (x,y,z) and the rotation (th). There's a lot going on with this, and it's kind of half finished, so I'll follow-up in a future blog post.
- Changed the default prop toss/spin orientation. It now looks much more realistic for clubs. Even if I don't continue on with the ability to change the orientation, I think the default now looks pretty nice.
- Added a bunch of advanced controls for the dwell path parameters. You can now modify the impact the toss/catch velocities have on the dwell path, including whether or not to match velocities on toss and catch. These parameters are really meaningless to the average user who would just be interested in a good looking animation, so I'm going to see if there are any magic numbers that always look nice.
- Added more examples
- Changed prop toss/spin orientation syntax. The siteswap now just encodes the number of spins, so you can say "5S3" for 5 club triples, but you can't say "5S3XY" to define the toss/spin orientation. The hand movement string is now used for the prop orientation. You can now add {x,y,z,th} to the end of each hand movement point to define the axis of rotation (x,y,z) and the rotation (th). There's a lot going on with this, and it's kind of half finished, so I'll follow-up in a future blog post.
- Changed the default prop toss/spin orientation. It now looks much more realistic for clubs. Even if I don't continue on with the ability to change the orientation, I think the default now looks pretty nice.
- Added a bunch of advanced controls for the dwell path parameters. You can now modify the impact the toss/catch velocities have on the dwell path, including whether or not to match velocities on toss and catch. These parameters are really meaningless to the average user who would just be interested in a good looking animation, so I'm going to see if there are any magic numbers that always look nice.
- Added more examples
Subscribe to:
Posts (Atom)