There is still a kind of uncertainty about short URLs that prevent some people (like me) from clicking on them when the source is not entirely trustworthy. For e.g., how do you know if the target site would be SFW? Is it just a link to a Wikipedia article, a tweet or a Facebook profile? Could we pass this information in the short URL itself?

My idea of doing that is to dedicate the first letter of the URL to the general category of the domain being redirected to. Highly used domains such as Wikipedia, Google, Facebook, Twitter etc. can have their own letters assigned to them. In base 62 URLs, there are more than enough symbols to categorize most of the major domains, with letters such as X used to denote unsafe websites.

Dedicating a letter to a domain may seem wasteful but once this categorization is well-known, these URLs are more likely to be clicked than completely strange ones. Also, this may in turn help reduce the total length of the key by reducing the bucket size for some domains.

Some examples:

Wd21k = http://en.wikipedia.org/wiki/URL_shortening
Gb2H1 = http://www.google.co.in/search?sourceid=chrome&ie=UTF-8&q=url+shortening
RP1 = http://www.reddit.com/r/programming/

Backwards Raytracing

The commonly used form of raytracing is backwards raytracing where we trace rays from the eye to the objects and light sources as opposed to the real world where the opposite takes place.

Recursive Raytracing

This is where raytracing beats most other algorithms in photo-realistic reflections and refractions. Once a reflective/refractive surface is encountered, we place a camera on that point and another image is raytraced from that camera and sent back as the colour of that surface, which accurately simulates how reflection and refraction take place. Clever, huh ?

Global Illumination: Caustics and Radiosity

This is the monster which truly makes images photo-realistic. You see, the usual raytracing algorithm doesn’t take into account the light that bounces off surfaces. For example, a white wall beside a red one looks reddish near the common edge. This is called Global Illumination, an effect which isn’t ordinarily reproduced by backwards raytracing.
Radiosity divides the whole seen into ‘patches’ and keeps bouncing light between them until a steady state is achieved, similar to the real world. This, of course, requires much more computation and hence most efforts are aimed at optimizing the algorithm.
Caustics are effects arising due to bending light after reflection and refraction, which illuminates areas which wouldn’t normally be lit. It also involves the phenomenon of diffraction.
Henrik Wann Jensen is the dude who came up with the brilliant Photon Mapping. It basically involves shooting photons at objects in order to determine the effect of Global Illumination and Caustics. And it’s fast too. Check out his page for more.

I’ve been interested in all this since a long time and I somehow managed to come up with some images over the years, which I intend to show off now. Mind you, what I show you is the output of a computer program (c++ based on djgpp, and later visual c++) which I wrote myself, with lots of help from various sources.

This is using Monte Carlo ray tracing with Quasi-random samples:

This during path tracing experiments:

This one is my attempt at Image Based lighting:

This is my very early attempt at Radiosity, which I later found out was how Henrik Wann Jensen implemented it:

This is implemented by considering all the ‘things’ as charged particles tied to each other with springs. For more, check Force based algorithms