Ray? Sphere? Intersection?

It was at about this point that a strange Pommy walked up to me holding a cup of tea, and with his high strung accent he said to me - "Why hullo there dear chap, I must say, can I have your ray-sphere intersection when you're done with it?"
Strange delusions... many... =]

Anyway, over this weekend I have accomplished one of those little dreams I've had since I started 3D all those years ago. A raytracer that is one hundred percent mine all the way from the ground up!

I started off writing a 3D vector library which makes handling a lot of tasks (creating and normalizing the original viewing vectors etc) and make them simpler. A short while after that, I implemented the ray-sphere intersection code and ended up with -

"Oh... Hmm..." says Smerity.
Oh hmm indeed - much to my annoyance it appears one of my spheres decided to go elliptical on me...
I thought it was an error in my ray-sphere intersection code so I redid the entire ray-sphere intersection code to use vectors... Much prettier - but didn't change anything...
Eventually I discovered I was an idiot and that, just like real life, I had the 'camera' too close to the spheres so that you end up with a depth of field effect - when you go that close it ends up with a smudge basically...
USER ERROR BY THE GUY WHO MADE THE PROGRAM :(
Silly Smerity =]

Anyway, after I got my bearings with the depth component I ended up with a thing of beauty... Nearly =]


That concluded late Saturday night. Sunday I went to work on diffuse shading, which came out quicker than I thought. By the end of the day I ended up with -

My attempt at adaptive antialiasing failed... Horrifically... Teaches me to try and come up with my own theory flying in the face of convention - this is just me cheating with low supersampling ^_^


My 'ghostray', the results of the failed adaptive AA =]

Today however I ended up going into the University of Sydney. Yes, the coding sprint that I thought was off was in fact back on =] A disturbingly short email alerted me to its re-occurrence but luckily Cat filled in the gaps... Got there today, extra early as I was expecting trouble on the Bridge (the Tunnel went toll free) and freaked out for most of the morning as no-one was there =] After a while a guy or two appeared and the day got underway. A bunch less people that I expected, but that was as they were either overseas or they were panicking about a commercial project due in some time last week =]
I ended up working on a web interface to help users classify questions. I won't say too much about it as I'm not exactly sure how open/closed it is, but here's the result of day one of two -

Obviously the question has the answer in it, but that's only as we haven't done our own questions yet and are running on previously tagged corpus questions.

Anyway, until next time, NLP FTW!
Smerity