Just a quick update.
im looking at XML feeds for the widget thing and although there are some good feeds, they are either U.S Based or only provide the temp.
I havnt given up yet, so ill post again when i find somthing.
Saturday, September 09, 2006
Thursday, September 07, 2006
Sun, Moon, Clouds and SWFs
Thought I'd post a progress update.
Done some work in flash recently. I've uploaded the .swfs for viewing.
Firstly : Motion Capture.
This is just our basic motion capture system. Using Colour tracking. Hold a colour in position over the x in the bottom right corner, press the 'space' key, and the program will attempt to track that colour. Still needs work. ( you need a webcam to view this ).
URL : Colourgrab_05_testthresh.swf
secondly : The Sky.
Because we're doing two things (interactive sky, and simulated sky), We needed a sun that would accurately plot time. So I made this :
timeofday_timelapse.swf
It's a timelapse version of an accurate program. The code places the sun / moon in the approprate spot depending on the time of day. This timelapse program is exactly the same code, however it's me manually setting the time at a faster speed. Just so you can see the effect.
finally : Clouds.
I've spent a long time trying to get our clouds to look right.
We had so many options. Using actual images of clouds. 3d modelled clouds. Perlin Noise generated clouds.
I wanted to go with the perlin noise ( for those of you who dont know what perlin noise is, it's fractal - based noise. Great for organic textures / shapes like wood, or clouds. More information here ).
However I had *great* difficulty in creating clouds that looked right.
The closest I got was this : Noise_Xp_2.swf
I was not happy with that.
So I sat down and started again.
I decided to create clouds out of particles. Each cloud would consist of a number of these particles depending on the 'density' setting. Then I could create a randomly generated dynamic cloud with a dynamic number of particles.
Chuck a couple of filters on it, and voila - we have clouds.
I also discovered something neat along the way. Dynamically LIGHTED clouds.
Here are the clouds, minus all the filters, have a look at the shading on them as they pass the sun.
particle_clouds_03.swf
The dynamic lighting looks GREAT.
But is VERY VERY VERY cpu intensive. Unless I can find another way to do it. Im scrapping it for now.
Anyway
This is what Im happy with for now :
particle_clouds_04.swf
Keep in mind, this is just for the prototype, and I can add detail and the likes later. But I think they look pretty good! ( and most of them are draggable ).
So that's just a quick update of what I've got working so far. Now it's just a matter of writing some more code over the weekend. And I think by monday I should have 3 working prototypes. Maybe longer for the motion capture. I m still waiting on a response from a flash expert for some help with that.
Ben is currently working on getting me some RSS / XML data
Bonnie is collating sounds
And we're still waiting on a spot / shorthrow projector so we can get our specs and (Hopefully) start building soon!
Done some work in flash recently. I've uploaded the .swfs for viewing.
Firstly : Motion Capture.
This is just our basic motion capture system. Using Colour tracking. Hold a colour in position over the x in the bottom right corner, press the 'space' key, and the program will attempt to track that colour. Still needs work. ( you need a webcam to view this ).
URL : Colourgrab_05_testthresh.swf
secondly : The Sky.
Because we're doing two things (interactive sky, and simulated sky), We needed a sun that would accurately plot time. So I made this :
timeofday_timelapse.swf
It's a timelapse version of an accurate program. The code places the sun / moon in the approprate spot depending on the time of day. This timelapse program is exactly the same code, however it's me manually setting the time at a faster speed. Just so you can see the effect.
finally : Clouds.
I've spent a long time trying to get our clouds to look right.
We had so many options. Using actual images of clouds. 3d modelled clouds. Perlin Noise generated clouds.
I wanted to go with the perlin noise ( for those of you who dont know what perlin noise is, it's fractal - based noise. Great for organic textures / shapes like wood, or clouds. More information here ).
However I had *great* difficulty in creating clouds that looked right.
The closest I got was this : Noise_Xp_2.swf
I was not happy with that.
So I sat down and started again.
I decided to create clouds out of particles. Each cloud would consist of a number of these particles depending on the 'density' setting. Then I could create a randomly generated dynamic cloud with a dynamic number of particles.
Chuck a couple of filters on it, and voila - we have clouds.
I also discovered something neat along the way. Dynamically LIGHTED clouds.
Here are the clouds, minus all the filters, have a look at the shading on them as they pass the sun.
particle_clouds_03.swf
The dynamic lighting looks GREAT.
But is VERY VERY VERY cpu intensive. Unless I can find another way to do it. Im scrapping it for now.
Anyway
This is what Im happy with for now :
particle_clouds_04.swf
Keep in mind, this is just for the prototype, and I can add detail and the likes later. But I think they look pretty good! ( and most of them are draggable ).
So that's just a quick update of what I've got working so far. Now it's just a matter of writing some more code over the weekend. And I think by monday I should have 3 working prototypes. Maybe longer for the motion capture. I m still waiting on a response from a flash expert for some help with that.
Ben is currently working on getting me some RSS / XML data
Bonnie is collating sounds
And we're still waiting on a spot / shorthrow projector so we can get our specs and (Hopefully) start building soon!
Wednesday, August 30, 2006
Rss Feeding
Hey.. just a note for the RSS feeds for our weather 'widget' here is the link:
http://www.users.on.net/~johnson/weather/
As far as i can tell it is ok, and it gives us the options for more places other than brisbane if we want to 'experement' with that...
http://www.users.on.net/~johnson/weather/
As far as i can tell it is ok, and it gives us the options for more places other than brisbane if we want to 'experement' with that...
Monday, August 28, 2006
Colour Tracking
I've done some experiments over the last few days.
I've got flash to isolate a colour, and track it on a live stream.
It is very crude at the moment, but I think its possible to get this working. Will show other group members on wedensday, as well as lecturers and will get some feedback. Im very excited!
Will post the .swf's here later.
I've got flash to isolate a colour, and track it on a live stream.
It is very crude at the moment, but I think its possible to get this working. Will show other group members on wedensday, as well as lecturers and will get some feedback. Im very excited!
Will post the .swf's here later.
Sunday, August 27, 2006
Motion Detection and Input Devices
Evening.
Since I finished work today, I've been doing research on possible motion capture and input devices for the weather room.
An *extremely* viable option seems to be using flash 8.
However, whatever program we choose to write this in we're stuck with for the whole implementation, so we need to pick something that is fast, and stable. ( Im not saying flash isnt, Im just saying we should be careful ).
So I'll have to speak to the lecturers on wednesday about this. So far I'm leaning towards one of two programs.
Either Flash 8, or a java based program called 'Processing'. ( alternatively I could just code the whole thing in C++ or the likes ).
The reason I'm leaning towards flash8, is I know it can be done!
I've seen something on the internet that would be absolutely perfect. If you've seen the movie "Minority Report", then this should be pretty familiar. You've got a 'glove'. This glove has two LED's attached to two fingers. The two 'colours' are recorded by the program and stored. Then, by bringing the two led's together, the user effectively 'clicks'. To 'unclick' he simply drags the two fingers away again.
That would be INSANELY good for our assignment.
the other option would be to track the colour of the skin on peoples hands. and use two hands the same way the two fingers would work.
So far, I know how to get webcam data into flash. And I know how to isolate a colour from this stream. What Im having troubles with is tracking this colour across the screen.
When you get the RGB value from the Bitmap image saved from the webcam, it's very accurate. It's a hex colour. There's no +/- leway. I can then say 'any colour that is different to this colour, remove it'. But since the colour picking is SO fine, it deletes most of the image bar a few specks here and there.
What I've got to figure out, is a way to increase the area of colour, even by a few hex digits either way... just to give us some leway. There might be a way to do this in flash, I just have to look a bit further.
More as it happens.
Since I finished work today, I've been doing research on possible motion capture and input devices for the weather room.
An *extremely* viable option seems to be using flash 8.
However, whatever program we choose to write this in we're stuck with for the whole implementation, so we need to pick something that is fast, and stable. ( Im not saying flash isnt, Im just saying we should be careful ).
So I'll have to speak to the lecturers on wednesday about this. So far I'm leaning towards one of two programs.
Either Flash 8, or a java based program called 'Processing'. ( alternatively I could just code the whole thing in C++ or the likes ).
The reason I'm leaning towards flash8, is I know it can be done!
I've seen something on the internet that would be absolutely perfect. If you've seen the movie "Minority Report", then this should be pretty familiar. You've got a 'glove'. This glove has two LED's attached to two fingers. The two 'colours' are recorded by the program and stored. Then, by bringing the two led's together, the user effectively 'clicks'. To 'unclick' he simply drags the two fingers away again.
That would be INSANELY good for our assignment.
the other option would be to track the colour of the skin on peoples hands. and use two hands the same way the two fingers would work.
So far, I know how to get webcam data into flash. And I know how to isolate a colour from this stream. What Im having troubles with is tracking this colour across the screen.
When you get the RGB value from the Bitmap image saved from the webcam, it's very accurate. It's a hex colour. There's no +/- leway. I can then say 'any colour that is different to this colour, remove it'. But since the colour picking is SO fine, it deletes most of the image bar a few specks here and there.
What I've got to figure out, is a way to increase the area of colour, even by a few hex digits either way... just to give us some leway. There might be a way to do this in flash, I just have to look a bit further.
More as it happens.
Friday, August 25, 2006
Fog Screen / Helios Display
I've been doing a bit more research about the fog screens.
www.fogscreen.com is going to be wayyyyy too expensive for us to get our hands on. For starters it's quite large. Too large to fit into the space we have to work with. And secondly, apparnetly it retails for around $150,000.
There's a section option out at the moment, called the "Heliodisplay". ( http://www.io2technology.com/ ) - This one is smaller, ( probably too small for our needs ), but has a pricetag of around $15,000. ( US I think ).
As it stands at the moment, I think we might have to drop the fogscreen idea. Will begin research into the other options availble to us, including that 3d projection .pdf that was posted on the newsgroup.
More on this later.
www.fogscreen.com is going to be wayyyyy too expensive for us to get our hands on. For starters it's quite large. Too large to fit into the space we have to work with. And secondly, apparnetly it retails for around $150,000.
There's a section option out at the moment, called the "Heliodisplay". ( http://www.io2technology.com/ ) - This one is smaller, ( probably too small for our needs ), but has a pricetag of around $15,000. ( US I think ).
As it stands at the moment, I think we might have to drop the fogscreen idea. Will begin research into the other options availble to us, including that 3d projection .pdf that was posted on the newsgroup.
Subscribe to:
Posts (Atom)