30 Days 30 Photos - 2011

For the last 30 days me and some friends have been taking one photo every single day.

Why? No reason other than we can really. We did it a couple of years back and it was good fun, so thought it would be worth repeating :)

You can see a rather nice slideshow of all the images which I have uploaded to my Web Album.

I took each one using Instagram for the iPhone so all of them can also be found on my Instagrid.

I really enjoyed it although I must admit it was hard to find something worth photographing each day. I also wish I was better at taking pictures, but perhaps now I have a good camera Ill learn.

PostToTumblr 0.8 - 8000 Users and Counting

I cant believe how well my humble little extension for the chrome browser is doing. 8000 users when just a few months ago I was celebrating 1.5k.

There were so many requests for new features and things that I thought I would push an update out this evening to add a little more functionality.

From the change log:

- v.0.8 -- 19/06/11 --
+ Text, Page, and Links are now supported in addition to Images
+ Added an app icon to the bar for easy access to the options

To highlight these new changes I decided to update the promo video too:



I have also added a donation button into the options. I have no idea if anyone will click it, an interesting experiment tho.

If you have it installed it should auto-update, if not go grab it over on the chrome app store: https://chrome.google.com/webstore/detail/dbpicbbcpanckagpdjflgojlknomoiah

Bye Bye IntenseDebate, Hello Disqus

Just a quick post. I have swapped over my commenting system from IntenseDebate over to Disqus. I really didnt like some things IntenseDebate did. It also seemed to get alot more spam which I had to manually approve / disapprove all the time.

Hopefully all the comments should have been preserved in the move over.

Chrome Crawler, HaXe, Three.js, WebGL and 2D Sprites

Had a little free time this weekend so thought I would scratch an itch that has been bugging me for a while.

I started the second version of my Chrome Crawler extension a little while back. I have been using the language HaXe to develop it in. It's a great language and I wanted to explore its JS target a little more so I thought, why not make a chrome extension using it. I have had several emails from various people requesting features for Chrome Crawler so I thought I would extend the extension and rewrite it in HaXe at the same time.

I managed to get the basics of the crawler working a few months back but through lack of time got no further. The second thing I wanted to work on after the basic crawling code was how to represent the crawled data. The current method is simply as a list:



 

A while back however I recieved a mail from "MrJimmyCzech" who sent me a link to a video he had made using Chrome Crawler and Gephi:

 



 

As you can see its pretty cool, visually graphed out as a big node tree.

So it got me thinking, can I replicate this directly in Chrome Crawler? To do this I would need to be able to render thousands of nodes and hopefully have them all moving about in a spring like manner determined by the relationships of the crawled pages.

The first thing I tried was using the HaXe version of the Raphael library. The library is designed for graphing and uses the Canvas with SVG for rendering, so I thought it would be perfect for replicating Gephi. I tested it however and only managed about 300 circles moving and updating at 25FPS:



 

300 nodes just wasnt going to cut it, I needed more.

Enter the recent HaXe externs for Three.js and its WebGL renderer. Three.js is rapidly becoming the defacto 3D engine for Javascript and takes alot of the headaches away from setting up and rendering to WebGL.

After a little jiggery pokery with the still very new externs I managed to get something running:



Thats 2000 sprites running at 25fps which is less that I would have hoped for WebGL but still probably enough for ChromeCrawler. Im not too sure why the sprites are upside-down, nothing I can do seems to right them, perhaps someone can suggest the solution?

If you have a compatible browser you can check it out here. I have also uploaded the source to its own project if you are interested.

The next step is to take the data from the crawl and then render it as a node graph, exiting stuff!