Shannon Larratt is Zentastic

Really, I'm just fiddling with it a lot now to fine tune it and see what needs improvement and what malfunctions with odd input (so far worst case it just outputs nothing; I haven't been able to induce a full-on crash yet)… Below are some more sample maps. I can't imagine anyone except me will find them particularly interesting.

This is the same map as in the entry below, just rendered using the night-time settings.	These are a few of the places I've lived in the past few years.	I've also lived in these places at one point in time.	These are some of the places I've travelled to in the past few years (big map test).	These are all places that Rachel has lived in.

First of all, I've made the following addition to the TOS… I've done my best to write a policy that (a) serves the purpose of IAM, (b) meets the wishes of the IAM community, but (c) doesn't restrict the personal expression rights of the members… I hope this works alright, and feedback on it is still welcome of course:

Avatars

• If you use an adult avatar (that is, a gential or breast shot, or similar photo), you must rate is as “not safe for work” in order to allow safe browsing for IAM members who use IAM in environments where adult avatars would be inappropriate. (Click the blue RATE button under your main picture).
• Your avatar should represent you or your personality. Misleading avatars used to “get people to click” are strongly discouraged as they are both contrary to the community purpose of the site, and artifically inflate the site's bandwidth (thereby increasting the cost of the site). Most commonly this means men using naked photos of women as their avatars.
• Pages that ignore the above will simply be taken out of main-page rotation. That is, you can still use the avatar you choose (assuming you have legal rights), but it won't be placed on the first page when you update.

Now let me tell you something that really sucks. NASA's budget is now under $14 billion annually. They don't even have enough money to upgrade the sapce shuttle in order to make it safer, although the cost of doing so is less than the giant pile of cash big business handed to Bush's election campaign.

Since Bush seized power, the defense budget has had its anuual spending increased by about $70 billion yearly (from $329 billion to about $400 billion), not to mention the $20 billion annual increase on Homeland Security in that same time period and other associated costs.

So if there's $100 billion dollars just lying around (since he's claiming to reduce taxes), is a wise decision really being made by funnelling that to war efforts, rather that valuable space research? Here are a few of the pro's, as I see them for sending that money into the space program:

• The research it provides saves lives and improves lives for millions of people.
• It bolsters American companies.
• It's good for international relations.
• It increases national pride.
• It increases the international community's respect for America.
• It encourages the international community to work together in peace.
• It is the future, and any nation that doesn't embrace it will fall behind.

And here's the kicker: it's very expensive! Space research is by no means cheap. Which means that all these corporations that are currently getting very rich off the war efforts could be getting rich off the space program as well… So does that mean they prefer war?

My brain hurts! I've gotten so mentally flabby on doing this type of thinking… My internet connection's been down for a lot of the day because of snowfall, so I've been tinkering with the dynamic mapping system. The problem is the size of the dataset. My tests that you've seen here have been rendered using the giant map — 21,600 pixels across. That's something like 200 meg of image to crop and rescale, so it chews through at least a second of CPU time. No big deal for my testing here, but unacceptable once it's on the server.

So how to render fast? My first experiment involves subdividing the map into 16 sections — four vertical divisions and four horizontal ones. From this 4×4 grid I create nine 2×2 submaps. Each one of those submaps is broken down again, and then their submaps are also broken down, resulting in about a thousand 2700 pixel wide images, with a final resolution of about 1km per pixel at full zoom, although it will do an infinite amount of digital zoom past that (not that it's meaningful past 200%).

Anyway, below shows a quick drawing of what I'm talking about:

Then, on the rendering level, once I have the edge points of the desired map (in radians), I calculate the area of the desired map. Then I seed the process using the world map's data. I then check to see if the desired map fits inside any of the nine submaps. Any that do are recursed again down to the maximum depth. At the end of it all, the software uses the map with the least differential between actual area and desired area. In the case of ties, it picks the most centred version.

2700 pixels across is still a fairly large map size, but I think, given the power of this server, that it should strike a happy medium. After all, if I instead went with 1350 pixels in width (my original plan), it would result in almost ten thousand images… And with every iteration the level of redundancy increases (and from a quality point of view, nothing much is saved).

In any case, if my thinking is straight on the matter, it means that a 50% resolution map of 2km per pixel (worst case scenario; most maps will render at 100% resolution) can be rendered for any rectangle on the globe). If I get ambitious I'll make the map loop (so for example, Ankorage AK and Tokyo Japan would be rendered as a tight crop, not as a world map), but that's most definitely a project for another day.

Anyway, hope that wasn't toooooo boring.