A new and quite exciting hobby of mine is photography - not just taking photographs, but experimenting with various techniques, post processing software and algorithms, new hardware and so on. To this end I've taken several computer science courses (CSC320 - Introduction to Visual Computing, CSC420 - Introduction to Image Understanding and CSC2503 - Foundations of Computer Vision) and attended numerous talks.

• You can see my pictures on my flickr photostream.
• Below, I include some simple examples of my experiments with photo processing.
• I'd like to note three wonderful talks on computational photography that were given at UofT - the first, by Marc Levoy on Light Field Photography and Microscopy, the second by Ramesh Raskar on Computational Photography: From Epsilon to Coded Photography, and most recently, as part of the Distinguished Lecture Series, Shree Nayar gave an outstanding talk on Computational Cameras: Redefining the Image

From wikipedia: Bokeh is a photographic term referring to the appearance of out-of-focus areas in an image produced by a camera lens using a shallow depth of field.

The flickr community has exploded over bokeh, even creating a slang greeting HBW (Happy Bokeh Wednesday), as well as several other less used ones. This is my most celebrated bokeh effect picture, where instead of choosing a subject and creating a bokeh in the background, I chose to make bokeh itself the subject. This is a close-up of a lava rock.

A slightly less popular technique is that of panoramas. Here, one takes several pictures of a scene, generally from a single location, and stitches the result into panoramas. I first learnt about these and some algorithms used in my CSC320 - Introduction to Visual Computing course. Aside from being cool, they can give you significantly better resolution of certain scenes, since you don't have to limit yourself to one photo at your camera's resolution. Here I include an HDR Panorama (a panorama where each stitched frame is an HDR picture) of City Hall, Toronto. Credit for the idea of the picture goes to Marc's (his Sony alpha just ran out of juice at the time...). I wrote a tutorial for HDR panoramas here.

Another wonderful development is the so-called 'little planet', which refers to the result of creating a (certain type of) stereographic projection from a scene, usually a panorama. Under certain simple conditions this transformation produces a so caller 'small world', where your scene takes the shape of a globe, with the center at the position of the photographer.

On the left I include such a planet from downtown toronto, near the UofT campus.

An interesting result can come from pointing your camera at the night sky - the appropriately called startrails. As the earth rotates we can trace out the 'trail' of the stars, creating a rather neat effect. There are two main ways to do this: either set your camera on very long (or indefinite) exposure, or take many pictures and combine them. I find the first more accurate and realistic, but you need a completely dark sky - if there is a city nearby, your picture will get saturated with light relatively soon. The picture to the left is taken with about a 20 minute exposure, near Algonquin park - without a city anywhere near - at 2AM, in pitch black. Still, one can see that in the 20 minutes the horizon caught colour. The second method works wonderful in city situations, where you catch the stars but don't want to saturate the picture - however, you may also get other artifacts, and if you live in the middle of a large city, even this method won't allow you to capture many stars (just too much light polution).