BEGINNING THOUGHT TRANSMISSION FROM AUGUST 2015
I've modeled the city, and it's time to render the first scene. I want a zoom in from space, to place the hidden images.
*a bit of fiddling around in Blender later*
Turns out it's a real pain to model something the size of a planet. Especially if it needs to blend well with things that aren't
the size of a planet. I know that Kerbal Space Program dealt with this be having a separate copy of the solar system, scaled down to 1
scale, that gets swapped in when you go far from a planet. I think I will use a similar idea and make a very small Planet Earth in another scene, and zoom into that instead.
What does Earth look like from space in the year 3000? Well, let's assume that if we've managed to completely screw up the climate in 150 years, and then don't do anything for another 50, the remaining 950 years will be enough to mostly fix it. So the greens and browns will be basically where they are on the current Earth. However, I want cities to have expanded to huge amounts - think Los Angeles sized cities on all the coastlines. So I need to replace the coast with bigger cities.
I also need to cheat the time a little bit. I want to zoom into the earth at daylight, so the whole thing is visible, but the actual closeups of the city need to be just before sunrise.
Finally - I need a sign to let cars know the bridge is closed. But - what should it say? What language even should it be in? In the last thousand years, English has changed so much as to be completely unrecognizable. Languages of the area - Spanish, French, Italian - arose out of Latin. And the languages spoken in particular areas have changed significantly as well.
Well, in the year 3000, globalization has taken hold. Currently the lingua franca
is English; so let's suppose the language of the future is largely English, with some Spanish and perhaps a bit of Mandarin. Given the progress with digitizing everything, there will be a huge amount of data that is interacted with frequently that uses language from centuries past; so perhaps language change has slowed. The main problem with English is how inconsistent it is, especially with spelling. Hmmm...
[skip a week or two]
*lightbulb* I should use the writing system I developed all those years ago! We will assume there was a standardization effort at some point which created a new alphabet to record data consistently between languages. The neat thing is that it sort of looks like Beanish, too, so perhaps the Beanies are writing with the script as it has evolved over the next 9,000 years.
...But how to write it? Handwriting will not look good on signs and such.
The solution: create a font! There is an awesome website called Glyphr, which lets you build a font, complete with all the little things I need such as ligatures. I can draw all the letters once, then when I need to make a sign or something, simply type it out in GIMP.
Let's see, what should the first sign say? Obviously a last-minute sign like this can't tell people that they need to drive all the way around the Mediterranean. Perhaps there's a under-sea tunnel, as well. Let's put it at Ceuta, at the other end of the strait.
"Gibraltar Bridge closed. Use Ceuta tunnel."
Because I am deciding what happens, I can choose how consonant/vowel shifts happen so that they're easier to write with the new alphabet. Let's make all G/J's soft, and remove any non-essential vowels before R's and L's. (The latter is sort of happening already.)
"Zhibraltr Brizh klowsd. Yws Siwta tanl."
And now we can let this render.
...Hmm, it takes around 15 minutes per frame, and it's still noisy. The noise I can live with, because we need to dither it anyway. But that lets me know I will have to do images in large batches.
[skip forward to the beginning of September]
Final step: draw the black-and-white images that people will actually see. My plan is to have it pretty slow and not doing much for a while. Then at some point someone may figure out the secret, or I'll have to reveal it, and it will turn out there was a second story hidden underneath all along! (Like we thought Randall might have done at one point, but he didn't.) The process is pretty simple: have a giant XCF with the whole scene in it. Have one layer with just the border in it. When I'm finished with a frame, I crop the image to the border layer, export it to a new image, and undo. (I messed this up a bunch of times - exporting over the last frame and having to re-make it, or forgetting to undo and realizing a few frames later that I've accidentally deleted everything outside the border.)
...How do I ONG the images without anyone knowing who it is? I asked mscha, and the solution was to put the images in a folder where newpixbot could find them. So, I put them on my website, in a folder with no link to it. They've been in http://taixzo.com/t-1
ever since I started uploading them. Newpixbot would check the folder every 4 hours and ONG the next image.
I need a way to secretly encode the colormap information. The frames use an 85x3 palette. bftf did not need a key, because with only three or four colors you could put ones in by hand. I suppose the best way is to hide it in the hash of the image.
*proceeds to try and upload the images to the server, gets an error*
After a bit of research, it turns out most filesystems, including the one my server was using, are limited to 255 characters. Unfortunately, because I need them to be valid filenames, I can't just name the frames with 255 bytes - I need to name them using text bytes, so I encode it as base64. 255 bytes encode to 340 bytes of base64 - longer than they are allowed to be. The solution: store the hashes in a text file. mscha modified Newpixbot to check the text file and grab the hash for the current image. (I call it a "hash", but it's not really a hash, is it.)
And now we wait...
END OF THOUGHT TRANSMISSION FROM SEPTEMBER 2015RSIR