.. Much like how gfycat encodes their links
eg.:
- https://gfycat.com/JaggedIdealFrillneckedlizard
eg.:
- https://gfycat.com/JaggedIdealFrillneckedlizard
demo - http://hipku.gabrielmartin.net
explanation - http://gabrielmartin.net/projects/hipku/
The hungry white ape
aches in the ancient canyon.
Autumn colors crunch.
For IPv4 addresses, there's loads of space, so I can afford to use some longer words.
Confirmed for localhost
If you're making full sentences anyway, the grammar of the sentence doesn't need to change much. The vast majority of the entropy is already in the words themselves.
Example sentence (generated by me, not a RNG): "the tiny hairy fish quickly paints a big scary monster".
EDIT: With 10 words, each from the 252 most common words... sentences of this type would have an entropy of more than 10^24 or 2^80. I guess "articles" are pretty much "The" vs "A / An" however, so there really are only 8 words of note...
In regards to word commonality, I'm pretty sure you could in fact use something like the 5000 most common words. The people who care about this kind of stuff tend to have large vocabularies!
You could, for example, use a part-of-speech tagged corpus (a large collection of text where each word was tagged with its PoS by a grad student). Just train a Markov model on the parts of speech instead of the words themselves, and you would be able to generate English-like mad-libs.
Throw in some ambiguous adjectives and you should have a large enough namespace that matches up with common image contents.