Procedural City Generation - Developer's Log
shamusyoung.com
shamusyoung.com
Also check out his "terrain" series, it's terrific. http://www.shamusyoung.com/twentysidedtale/?p=141
http://media-cdn.tripadvisor.com/media/photo-s/01/12/00/de/t... http://abodenyc.com/images/New%20York%20City%20Images/New-Yo... http://edp.org/Travel/Chicago/ChicagoNight.jpg http://www.flickr.com/photos/arjunpurky/2507602603/ http://www.picturecorrect.com/wallpaper/japan/tokyo_1024x768... http://www.flickr.com/photos/yayatan/35736937/ http://www.postcardsfromhonshu.com/wp-content/uploads/2009/0... http://www.markmyhsieh.com/Blog/files/page0_blog_entry78_15.... http://leejones-san.tripod.com/sitebuildercontent/sitebuilde...
Projects like these are great examples of hitting that 5% that gets you 95% of the way home. When he talks about avoiding simulations, that's precisely right. I wish he would talk more about how he's choosing the right things to focus on. Perhaps it's just luck? For example, I think that a few of his steps, like bloom, were detrimental. Irregular streets might have solved the streets problem by hiding them (cf. 1st pic). Brightening up buildings in the distance could also have made this look even better.
http://www.youtube.com/watch?v=-d2-PtK4F6Y
Also, if you are interested in procedural cities generation, there is a very cool Swiss startup:
It's a spin-off from ETHZ graphics research. You can find their SIGGRAPH papers here:
Nice article.
Why is the city's nighttime generated based on lights?
this is the opposite experience to blogs posts that run out of steam halfway and then end with some dumb conclusion. well done shamusyoung. wish there were more of these kind of project reports on hn.