Difference between revisions of "Testing Movie Caption Scroll"

From Viewpoints Intelligent Archive
Jump to: navigation, search
Line 2: Line 2:
 
service=youtubeIA
 
service=youtubeIA
 
|id=R81egpgDzbY
 
|id=R81egpgDzbY
|alignment=center
+
|alignment=left
 
}}
 
}}
  

Revision as of 17:17, 4 November 2017

Our first example of Shadama is a simple physics simulation of particles and gravity.

As in the LOGO programming language, we refer to mobile objects as turtles. Here, we

declare a breed of turtle such that each turtle will have properties

"x", "y", "r", "g", "b" and "a".

A method is defined with the keyword "def". Methods are executed on

each turtle in a breed. Here, we've defined the "setColor" method to use the turtle's x and y

coordinates to assign them a red and green value.

In the static function "setup", we set the number of

turtles in the breed to be 3,000 and then set all turtle's x and y coordinates to be a

random number between 0 and 512. We run "setup" by hitting Cmd-S.

Those dots may be hard to see, but we can increase the number of turtles to 30,000, 300,000 or 1 million.

What we want to do next is to move those turtles around. So we define

a method called "move". Move increments the x and y of the turtles. We call

the move method from a static function "step".

When we click the "step" clock, "step" is repeatedly executed and the turtles move.

Interestingly, while they are moving, we can edit the program on the

fly and the changes take effect as soon as we hit Cmd-S. We can add an argument to "move", use it in the code, change the call site and the

turtles that went off screen come back.

We could even allow each individual turtle to have its own

velocity. So let's add properties "dx" and "dy" to each turtle. Then, we use them in "move". Here, instead of "n", we increment the "x" and "y" of each turtle

according to their "dx" and "dy".

We call the primitive "fillRandom" to set the "dx" and "dy" to be the x and y components

of a randomly-chosen direction vector. Now, all turtles move

in their own individual directions.

Let's run it again by executing "setup" manually.

Now we'll change this to be a simple gravity simulation. To do so, we add constant

acceleration to the "y" component of the velocity. Now all turtles are

accelerating toward the bottom.

We don't want those particles to go out of bounds. So we modify

the "move" method. First we store "dx" and "dy", as well as the possible

new location of "x" and "y" into local variables.

When "newX" is out of bounds, we make the turtle bounce back.

When "newY" is out of bounds below, we wrap the value around to the top.

After adjusting those variables,

we write the values back to the turtle's properties. Now one million turtles are falling down, bouncing off the side walls, and wrapping around

to the top.

We are now going to create a simulation with large objects. To do so, we use what's called a "patch". A

patch is a 2D grid of cells where turtles can store values. Here,

we create a patch called "Field", and add "r", "g", "b", and "a" properties for each cell.

And, we define a breed called "Filler" that operates on the Field.

Now we define a method called "fill". The argument for this method is the patch, and turtles executing this method update the patch cell value at their x and y coordinates.

We modify "setup" to use a primitive called

"fillSpace", which creates turtles of the Filler breed and places them at all integral grid points. When this "setup" is called, and the "fill" method is called, each "r" and "g" of the patch get values derived from its position, resulting

in this color gradient.

Let's make some shapes now. We define the "fillCircle" method to replace "fill". Let's also add the "clear" method, which clears the field's values.

"fillCircle" works by having each Filler turtle

check whether it is within the radius of the circle. If it is, it writes

the value into their corresponding cell.

In the "setup" function, we call "fillCircle" and get a circle at (100, 100).

Shadama programs can also be interactive. Let's split the "setup" function and create another one called "loop".

Notice to the right that there is a variable

called "mousemove", which changes its value as we move the mouse pointer in the display. That is an indication that we can use the value in our

program.

Now let's tie it all together. This code has methods we used

previously, such as "setColor". "fillCircle" is slightly modified so that

it sets normal vectors for the circle in the patch.

In the "move"

method, each turtle checks the normal in the patch cell where the turtle is located and computes a new direction from the dot product with the normal. When we run

this, particles bounce off the circle and make a beautiful pattern.

However, it is a bit hard to see because the colors of the turtles are too

similar. Here, we change the "setColor" method so that instead of using "x" and "y" position, we use the velocity "dx" and "dy" to determine the turtles'

color.

Then we call "setColor" from "loop", so that the colors are updated continuously.

Here, horizontal velocity corresponds with reddish color, and vertical

velocity corresponds with greenish color. The visualization can be quickly adjusted

to match how you prefer to view the simulation.

For fun, you can make two circles that move symmetrically.

Here's an interesting variation of our program. Instead of modeling gravity as pulling downward, gravity follows

the inverse square law in relation to the mouse pointer location. It looks like there are large structures, but it's actually just a

quarter million particles from regular initial positions, all following the inverse square law.

Shadama allows the user to import images. The built-in "fillImage"

takes an image object, creates enough turtles to cover it, and sets the turtle colors based on the image's pixel values. This looks like an image but it is

actually a bunch of turtles.

We reuse our previous "move" method, set the direction vectors randomly, create the "loop" method that calls "move", and run "loop".

For fun,

We can use the turtles' red and green values to determine turtle velocity, causing an

an interesting effect.

Let's use those pixels in our gravity simulation.

The code checks the timer, and every 10 seconds, it gathers all turtles to their original position.

We can also try some mathematical functions. This is a program to

visualize the Mandelbrot set. Notice how we can change the color scheme on the

fly.

This is Conway's Game of Life.

Again, we can change the program while

the simulation is running, and the code change is reflected in the

simulation right away, without needing to restart.