It’s been well over a year since my last blog post. What’s changed since then? I’m living in a new city and working at a new interactive shop with a talented crew. What hasn’t changed? I’m still getting to develop cool interactive work and still enjoy bringing my work home with me.
Here’s an interim demo of Paper-Robo V2, recoded ground up. It’s come a long way from the single document class I wrote in school, but will be always be a work in progress and learning tool.
It came to my attention, after watching the Conan premiere, my IMWITHCOCOROBO is no longer screen accurate. Here’s a quick demo on how to add a beard to your favourite Paper-Robo:
There’s more than just a paint brush tool at your disposal, so feel free to be creative:
With Chris Bosh’s contract with the Raptors expiring soon, there’s been a lot of talk of him being shopped around next season. Sources say that the Cleveland Cavaliers have targeted Bosh to play along side King James.
While I don’t want to see Chris leave Toronto, it wouldn’t surprise me to see him follow the path of other marquee players before him : Stoudemire (he was big back in the day), Carter, McGrady and Camby.
Where do you think Chris will go?
With this latest flar demo you can make your predictions by redesigning his jersey in AR, download the parts to your desktop, and print/assemble your own Paper-Robo! You can also upload your own custom skin to preview what the robo will look like in AR.
For anyone who designs a killer Bosh skin through this app or even better photoshop, please send them my way and I’ll post them here! I bet some of CB4′s twitter army could make some sick custom bots.
Adding mouse events to collada models and manipulating materials
This functionality is simply an extension of the helloMouse3D example that ships with the Papervision3D libraries. The only difference here is that instead of manipulating the material of a Papervision3D primitive, you are manipulating the material of a 3D collada model.
One thing I discovered is that material lists for colladas are wonky, and in order to add interactivity to 3D models you have to add listeners directly to the to the mesh associated with that material and only when the model has completely loaded. As you can see be the live demo, the focus of the mouse on the model sometimes gets lost. I suspect that this is due to the collada model I’m using which has been to hell and back to get it rendering properly in Papervision3D. I built the model 2 years ago in Sketchup when either Papervision 1.0 or 1.5 around! From Sketchup I exported it as a collada, and brought it into 3DS max to make a uv map. I exported from 3Ds max as a collada, only to import the result into Blender, to obtain a collada with settings the Papervision3d can render.
When I have time I will try the demo with a simpler collada to see if this loss of focus is still a problem. If you have any ideas why I’m losing focus on my model please let me know!
public function addMarker(marker : FLARMarker) : void
{
// store marker
var markerList : Vector. = this.markersByPatternId[marker.patternId];
markerList.push( marker );
var container : DisplayObject3D = new DisplayObject3D();
robo = new Collada( FLARManager._GATEWAY + "/resources/models/myrobo4flippedblender.xml");
A colleague at work pointed me towards a JPEG Async Encoder class that allows you to to merge large BitmapData objects to form a large JPEG, getting around Flash player 10 BitmapData size restrictions. I didn’t need to encode something beyond 4,095 x 4,095 for this particular demo, but it’s useful to know that there is a work around.
Flash Player 10 reading and writing files locally
This demo uses a some new features of Flash 10, reading and writing files locally between the browser and OS without the need to write any server side code.
Using the FileReference.load() method you can allow a user to select a file locally and load it directly into the Flash plugin without having to upload it to a server.
This is my first demo to include the newest FLARManager update. Eric Socolofsky has updated his framework to include some new optimizations, one of which is adaptive thresholding, which allows markers to be detectable in low light conditions.
AR Musings…
Is the wow factor is still there AR apps? Most offerings simply involve displaying a model on a glyph, which captures the user’s attention for about 2 minutes. There are a few hoops to jump through, in order to get a user to interact with an AR app: you have to print out a marker, need a webcam or have to download a specific plugin to view AR content (not as big a deal these days with broadband connections). Also imagine all the sheets of paper used around the world for a website that promotes renewable energy resources! I’m not sure what the solution is for saving our trees, other than offering digital versions of as AR marker that could displayed on a mobile device…Alternately, AR websites could offer a keyboard shortcut that mimics a marker without having to print it out.
Here are some interesting examples of interactive AR apps that have popped up recently that have started to push AR beyond displaying simple models in 3d space:
There are keyboard controls for both models. ASDW to move Starscream and directional arrows to move Boba Fett. Each character also has jump, taunt and attack controls. The UI controls are still pretty rough, and not quite as polished as online games available over broadband connections from o2 or similar providers. There are about 14 other animations I haven’t had time to program.
I re-factored the earlier multimarker flar prototype to use the FLARManager 0.3 from Eric Socolofsky at transmote.com. I was able to trace out the string names of the individual 3D channels of the MD2 models I’ve been using. I can now play the individual animations I want to play instead of looping through them all.
This snippet might help those that are trying to figure out the string names of the 3d channels of their MD2 Quake 2 models:
I also implemented Saqoosha’s hole effect using Papervision3d, in the form of trap doors, with some slight modification to what cube faces are visible. I think this has to do flipped coordinate space, when comparing a standard PV3d scene and to the flarnode. Here’s the mod:
Instead of :
var hole:Cube = this.scene.addChild(new Cube(new MaterialsList({all: cmats, bottom: cmatb}), 300, 300, 300, 1, 1, 1, Cube.ALL, Cube.TOP)) as Cube;
this._green = this.scene.addChild(new Cube(new MaterialsList({all: new ColorMaterial(0x00ff00)}), 300, 300, 300, 1, 1, 1, 0, Cube.TOP)) as Cube;
Changed to:
var __hole : Cube = new Cube(new MaterialsList({all: __cubeSideTexture, bottom: __cubeBottomTexture}), 80, 80, 80, 1, 1, 1, Cube.ALL, Cube.FRONT);
var __green : Cube = new Cube(new MaterialsList({all: new ColorMaterial(0x00ff00)}), 80, 80, 80, 1, 1, 1, 0, Cube.FRONT);
I went through at least 10 different asymmetrical patterns for the Starscream flar marker. Most of the time the camera would keep losing the pattern or confusing it with the Boba Fett pattern. Bad lighting and problematic marker patterns were the biggest issues.
I don’t have to use keycodes very often, but when I do it’s always a pain trying to remember what uint corresponds with what key. Here’s a snippet I picked up for using as3 to generate keycodes:
I just got back from volunteering at FITC Toronto 2009. I HIGHLY recommend volunteering. It’s a great way to attend the sessions for free and meet amazing new/old talent in the interactive field. I’ve come away from the conference with a lot of new ideas that I can’t wait to try.
Here’s my list of things that were cool to me in no particular order:
1) Papervision X – Ralph Hauwert
- 3d renderer optimized for Flash Player 10
- Ralph used Alchemy and Pixel bender to run Quake in Flash at 60 FPS!!
2) Colin Mook – Your Phone is Your Controller
- http://www.playmegaphone.com/
- http://unionplatform.com/intro.html
3) Sourcebinder
- http://sourcebinder.org/
- drag and drop node based programming
- break out game built on the fly with audience interaction using a webcam
4) Gary Stasiuk – Visual PV3d - Flash Visualization Application that provides a GUI interface to create and manipulate objects in 3D scenes using Papervision3D
7) Jared Ficklin
- music visualization through physical objects
- nirvana songs visualized on paper show similarities between songs; verse – chorus – verse
- showed 4 patterns and asked which one was likely a nirvana song, the one most similar was actually a foo fighters song!
Mario Klingemann
- bitmap manipulation turning ordinary bitmaps into Van Gogh like paintings.
- http://www.aviary.com
- using Keith Peters minimal components
I was able to revisit the spaghetti code on my FLARMultiMarkerDetector prototype and improve performance 100 fold. Thanks to the guys at Squidder.com for sharing their source for the virtual drum kit.
The MD2 models I’ve been using should be familiar by now and are from Polycount.
I was writing this app on the train and didn’t have a printer to print the patterns. Now I have them on my iPhone, so I can test my Flar apps where ever I happen to be.
One thing I’ve noticed, is that if you cutting the patterns out or are displaying them on an iPhone it seems you have to leave a white margin in addition to the black border for the pattern to be recognized.
I'm currently a Flash Developer in Toronto, Ontario. This blog is a resource library and journal, researching emerging new media and whatever else I see fit.
Some of my personal and professional work.
Mario Klingemann