Final Piece

joshshard — Fri, 08 May 2009 15:01:17 +0000

Here is a video of the final piece. I have looked at some of the small bugs and already started to fix them.

Project Evaluation!

joshshard — Fri, 08 May 2009 10:06:18 +0000

I have enjoyed this project, I have produced a virtual creature that reacts to user movement. Hotaru also can find his way round user placed obstacles to reach his destination. My aim for the project was to produce a more efficient, and smoother version with more behaviors. I think I have achieved this. I decided to use Actionscript 3.0 over 2. this is because it gave me a good opportunity to learn and push myself, this also meant it would run faster just because of the difference is speeds of the 2 languages.

I’m really happy with the usability of the creature, it runs smoothly and the motion tracking works effectively. The fiducial tracking also worked better than I thought as I had a lot of problems with this, but on the last test before the crit it all came together nicely, the user could line up squares and create walls that my organism would find his way around. However the only thing I am slightly disappointed with is the motion tracking the reason for this, it works incredible well when there is a single user but it doesn’t work with two users. Sadly making it a single user experience. From a single user purpose it runs very well, and is a good experience, but with 2 users it doesn’t work very well.

The code is my pride and joy of this project, it what makes it work, although code normally code isn’t linked with creativity, this is where it all happened in this project. I spent many hours venting, and ranting about it often saying “Why didn’t I build a website?” but now I have seen it working, I am so happy. I could say what all 430 lines do, and I have learnt how to program Actionscript 3 which will hopefully get me a job in the near future. There were 4 main elements that went into the code and they went together without clashing which was a relief although each section need a little modifying.

The setup caused the most problems, this is always going to be the case, when I need to calibrate 2 cameras to a projection at 4 meters off the ground facing directly downwards. The calibration was pretty accurate by the end and I am proud of that but the fact I was using 3 different devices with 3 different lens caused a nightmare. I original had planned to use a single camera which would have made calibrating easier but spliting a feed just slowed the machine down and the piece dropped to such a low frame rate and became unaffective. The only other thing I wasn’t happy about the setup was if I could have got it higher it would have been even more effective but sadly I had restraint restriction and no wide angle lens available.

What I would change about the project, I would have preferred to get my setup ideal but couldn’t because of restraints. I would have also liked to do a lot user testing with various audiences but as I believed the setup would have caused me a lot less problems than it did and this took a lot of time. I would have also liked to experiment with other behaviors, like making the creature a home which when he is scared he would run too, or making it so if you touched him he exploded into loads of smaller bugs and gathered back up over on the other side of the piece. Sadly time limitation stopped this but I will look into developing in the future.

I have thoroughly enjoyed working on this piece, at times I have felt despair but seeing it work has made me realize that this is what I want to do for a career. I have learnt a lot of very useful skills, working with flash, processing, reactivision, and various camera applications. I feel confident as an Actionscript 3 programmer, and I take away from this a skill that makes me employable. I had dissappointments with my setup but they don’t matter they can be refined the software for the creature is there and it is running very well, and can easily be built upon.

Screencast 5 – Compiling(Part 2)

joshshard — Fri, 08 May 2009 01:08:00 +0000

In this screencast I review the other changes I have made while compiling the code that I added, I try to cover all the changes to summarize what I cover in the video:

Mention no changes to the server code, it worked fine.
I mention added variables for the fear and attraction which weren’t added until the final compiled version.
I talk about how the referencing works but I will go into more detail on that in this post.
I talk about the fear and attraction code.
Also the drawing of symbols which I will mention in a little more detail here.

I had problems with my hit testing because I was struggling to get the values from my squares array so to solve that problem I went into my bigArray and got the x and y position out of there. For the X and Y positions to be accurate I had to multiple the bigArray value by the stage width or height so that it transfered correctly across, but still this was incorrect because the values I recieved from processing where decimals and took from the center of the fiducial rather than flash which i was calculating from the top left. I solved this by – or + half the squareSize which was a variable that held how big I wanted the square. So I knew the center of the square was:

_circle.y<=bigArray[4*i+2]*stage.stageHeight

_circle.x<=bigArray[4*i+1]*stage.stageWidth

These explains how to get the x and y value of the top left corner and from there I could find out the other corners, I minus half the square size because the first half of the equation finds the middle:

_circle.y<=bigArray[4*i+2]*stage.stageHeight-squareSize/2

_circle.x<=bigArray[4*i+1]*stage.stageWidth-squareSize/2

These measurements caused me the biggest problem but when I got to grips with it I began working well again. So here is the final screencast enjoy.

http://www.screencast.com/t/hcS8dLjIN

Screencast 5 – Compiling(Part1)

joshshard — Fri, 08 May 2009 00:35:36 +0000

In this screencast I go over the setup of my compiled version, I focus on the “init” function which is used to setup all the various things that I won’t need to return to change. I first cover the imports roughly what they do and also mention how I laid my variables out according to what the work with so the motion tracking variable are together, and the bug ones are grouped together. Finally I go over:

Setting up the server for processing.
Setting up the camera and circle for motion tracking
Setting up the bug
Setting up the square sprites in an array for the fiducial tracking
Setting up my timer and it event listener.

A lot of stuff that is similar to what has happened in the other screencast happens in this one but it is showing them all compiled together. So here it is Screencast 5.1:

http://www.screencast.com/t/IiruLKDvXYO

Screentest 4 – Movement(Part 3)

joshshard — Fri, 08 May 2009 00:17:47 +0000

This is the final part in the movement trilogy. In this screencast I show the code and how it works for multiple object hit tests at once and how I got around the few little bugs. So to sum up what goes on in the screen cast:

I show the piece with multiple symbols.
Explain how I will be drawing the shapes using an array and for loop
Also how I check the symbols and finally how I introduce a counter.

Just a brief explanation of the counter this counts up every frame and if it is hitting on box then the counter won’t reach the number of squares of the stage and that means it won’t run the script to go towards the movement, I fail to mention the reason this is in there is because otherwise the creature would see that it wasn’t hitting square 1 and would jump inside the box it was hitting. This counter solved that problem so here is screencast 4.3.

http://www.screencast.com/t/mOsRmsc5GVM

Screencast 4 – Movement(Part 2)

joshshard — Thu, 07 May 2009 23:58:30 +0000

In this screencast you can see I have made some drastic changes to make the hit test a bit more accurate and I’m also generating the values for the movement every frame which seems to have fixed the major problem of the bug jumping into the square. So to summarize what I have done in this screencast.

Added Actionscript to find out the side/corner of the square it hits.
Changed the movement from a tween class to mathematically calculated.
Made the circle react differently but predictably depending where it hits the square because of the above two points.

So just the final part of this screencast to come. I will be covering in the final part how I dealt with multiple squares. So here is Screencast 4.2.

http://www.screencast.com/t/lx3qGSIvkyp

Screencast 4 – Movement (Part 1)

joshshard — Thu, 07 May 2009 23:33:56 +0000

This covers my first attempt at movement, I cover some of my mistakes in how I was attempting to do it and the basic logic behind how I thought my code should work is:

Check if circle is hitting block – if not then move towards circle
If circle hits the block then move away from the block at an angle dependant on position
Then loop back to step 1.

The screencast going over the code is here: http://www.screencast.com/t/SpderuNIy

Screencast 3 – The Bug

joshshard — Thu, 07 May 2009 23:01:33 +0000

This screencast covers the bug, I go through the Actionscript in it explaining how it works. The physics used are called inverse kinematics. So an overview of how it all works is:

Firstly draws all the segments and stores them in an array
Then it calls a function to find out the angle from the front segment to it’s destination
Then it transforms the position of the front segment(x, y and rotation)
Then with this data it cycles through the whole of the bug applying the same transforms

http://www.screencast.com/t/atGmmPgD2Df

Screencast 2 – Motion Tracking

joshshard — Thu, 07 May 2009 21:50:42 +0000

So this is the motion tracking screencast. I just go over the the soulwire motion tracking code, I don’t delve into the classes just over the main actionscript file that makes it work. They are taking images every frame and comparing it to the previous frame and applying various filters to pick up the difference, they are using BitmapData function to store the images of the current and previous frames. That is detail on how the actual class works, this covers the code I will be using with a few changes to optimise it. So here it is:

http://www.screencast.com/t/Fre9A90Wl

Screencast 1 – Fiducial Tracking

joshshard — Thu, 07 May 2009 21:05:51 +0000

In this screencast I talk about my fiducial tracking code in the first part I cover processing in the second part I focus on the Actionscript. I try to cover it in as much depth as possible without dragging on. This is my first screencast they are quite difficult to do but it went okay.

Part 1: http://www.screencast.com/t/iV0K5PVd

Part 2: http://www.screencast.com/t/UabYSSageE