This is a review for Super Juicy, a fun Iphone app that involves popping bubbles in color combinations much like tetris. You can rotate the screen around to assist in getting similar colored bubbles together. Comes with a bonus mode that involves a lot of fast paced tapping!
Quite fun as Iphone games go, great to play while waiting for the bus.
Written by a friend of mine, Greg Hjelstrom.
Tuesday, October 06, 2009
Thursday, October 01, 2009
Google Wave for Text Adventures
Some friends (who are either game designers or programmers) and I have started a text based adventure game using Google Wave. You know, those old text computer games, such as Zork, where it says something like:
Computer: You are in the middle of no where and a mail box is in front of you
You type: Open Mailbox
Computer: Inside you find some letters and an ancient map
etc...
However, in Google wave, you can replace the computer with a live person, usually called in gaming parlance a Games Master.
You can see this as a form of modern, interactive story telling that is open ended. Like Choose Your Own Adventure books, but with a bunch of friends.
In order to help with this, I wrote a MUD helper Wave Robot, which participates in the conversation. You can ask it to roll dice, so to speak, to add some randomness to the text adventure. Try it out and let me know what kinds of functions you'd like to see and I'll add it as time permits. On googlewave, just add the contact mud-robot@appspot.com and then type in /help
Computer: You are in the middle of no where and a mail box is in front of you
You type: Open Mailbox
Computer: Inside you find some letters and an ancient map
etc...
However, in Google wave, you can replace the computer with a live person, usually called in gaming parlance a Games Master.
You can see this as a form of modern, interactive story telling that is open ended. Like Choose Your Own Adventure books, but with a bunch of friends.
In order to help with this, I wrote a MUD helper Wave Robot, which participates in the conversation. You can ask it to roll dice, so to speak, to add some randomness to the text adventure. Try it out and let me know what kinds of functions you'd like to see and I'll add it as time permits. On googlewave, just add the contact mud-robot@appspot.com and then type in /help
Wednesday, July 29, 2009
Google Similar Images: Image Specific Related Search
Just wanted to let friends and family know about something I worked on that's finally visible to the outside world:
Google Similar Images with Image specific related searches
Its the thing at the bottom that says "Related". It tells you what queries the image is possibly related to.
Sunday, June 21, 2009
Using the Image Search to Identify Birds
I was listening to a bird singing the other day and wanted to find out what species it was. All I knew was that:
1. It lives in California
2. It has a black central tail with white stripes on either side
3. It has a white stripe in its wings
4. It has a huge repertoire of calls
So I went to Google Web search and typed in [black and white bird], [bird identification], [california birds], [california song birds] etc... to no avail.
Then I changed tack and tried Google Image Search (full disclosure: I work on this). Typed in [black and white striped tail bird] went down a few pages and found the image of the bird I saw. Then I realized it was a Northern American Mockingbird. Armed with this information, I did a web search for [mockingbird song] and got confirmation of what it was by listening to the calls and the description. Indeed, the bird I was listening to had three repeated random calls from other birds then went on in a long sequence. Success! However this seems to be a really difficult search problem that requires using different modes of search in order to find the information I wanted.
Ideally with "Star Trek" like technology I would just ask my cell phone - what bird is this?
1. It lives in California
2. It has a black central tail with white stripes on either side
3. It has a white stripe in its wings
4. It has a huge repertoire of calls
So I went to Google Web search and typed in [black and white bird], [bird identification], [california birds], [california song birds] etc... to no avail.
Then I changed tack and tried Google Image Search (full disclosure: I work on this). Typed in [black and white striped tail bird] went down a few pages and found the image of the bird I saw. Then I realized it was a Northern American Mockingbird. Armed with this information, I did a web search for [mockingbird song] and got confirmation of what it was by listening to the calls and the description. Indeed, the bird I was listening to had three repeated random calls from other birds then went on in a long sequence. Success! However this seems to be a really difficult search problem that requires using different modes of search in order to find the information I wanted.
Ideally with "Star Trek" like technology I would just ask my cell phone - what bird is this?
Friday, March 06, 2009
Youtube video remixing
[Web 2.0] ThruYOU kutiman came up with this really cool remix of youtube user uploaded videos:
Kutiman's site itself is down due to high traffic as on the morning of March 6th but more can be found at this site.
Kutiman's site itself is down due to high traffic as on the morning of March 6th but more can be found at this site.
Thursday, March 05, 2009
District 9
One of those internet movie clips I have really enjoyed was Neill Blomkamp's Alive in Joburg (embedded below) about refugee extra-terrestrial aliens living in South Africa and having to deal with humans who are treating them as a lesser race. I love the way he interviews real people in South Africa about another topic (such as apartheid and the AIDS epidemic) then overlays great CG over it of giant saucers hovering over the city of Johannesburg.
Well, it turns out, Sony Pictures is going to make a movie out of it called District 9 and they have started viral marketing with several websites:
Official movie website
D-9
A blog by one of the alien characters
MNU Spreads lies
A website by the multinational united, some kind of organization (the bad guys?)
Multinational United
Well, it turns out, Sony Pictures is going to make a movie out of it called District 9 and they have started viral marketing with several websites:
Official movie website
D-9
A blog by one of the alien characters
MNU Spreads lies
A website by the multinational united, some kind of organization (the bad guys?)
Multinational United
Sunday, February 22, 2009
Higher dimensional classification
I just watched a great video entitled Nearest Hyperdisk Methods for High-Dimensional Classification by one of the paper's co-author's Bill Triggs.
In plain English, computer vision folks usually represent an image as a some vector in high dimensions e.g. the image of a parrot bird is at location [4, 5, 6, 7] and the image of a owl is at [5, 5, 6, 7] but the image of say a flower is at [100, 0, 20, 50] which is very far away from images of the parrot or owls. One popular technique right now is to find a maximum margin separating hyper plane that divides the two points. Just think of it as a moving a wall between pictures of birds and flowers and what you have is a method to distinguish between the two. One method of finding this plane is called a Support Vector Machine (SVM). The plane/wall probably the simplest decision boundary that separates things. You can make the decision boundary more complicated by making it curved, or say enclosing all pictures of flowers in a ball (hypersphere) etc. This video explains several alternatives that could be more specific of the decision boundary such as using points, balls, disks, convex hulls and affine hulls.
In plain English, computer vision folks usually represent an image as a some vector in high dimensions e.g. the image of a parrot bird is at location [4, 5, 6, 7] and the image of a owl is at [5, 5, 6, 7] but the image of say a flower is at [100, 0, 20, 50] which is very far away from images of the parrot or owls. One popular technique right now is to find a maximum margin separating hyper plane that divides the two points. Just think of it as a moving a wall between pictures of birds and flowers and what you have is a method to distinguish between the two. One method of finding this plane is called a Support Vector Machine (SVM). The plane/wall probably the simplest decision boundary that separates things. You can make the decision boundary more complicated by making it curved, or say enclosing all pictures of flowers in a ball (hypersphere) etc. This video explains several alternatives that could be more specific of the decision boundary such as using points, balls, disks, convex hulls and affine hulls.
Subscribe to:
Posts (Atom)
