I love that we both arrived at post-apocalyptic fiction. This played a bit like a madlib but way better, because the game and I were taking turns building off each other's prompts. It was like having a simulated improv partner.
It did occur to me that if the entire image was in grayscale, rather than a colourful background and a grey object, then players would be forced to identify the object purely from the shape described, rather than taking its grey colour into account (consciously or otherwise). It would also fit with the protagonist being a robot; I'm imagining him assessing his surroundings Terminator-style.
I'm looking forward to seeing all these games streamed live because none of us know what to expect.
P.S. Your text box system is way better than mine, may I ask how they were done?