So the turquoise and the lavender samples work very well. Blue I got to work a couple times, but it is very difficult. I had to keep trying and moving the blue sample around until it would change the LEDs. The green just was not working at all. I think maybe I'll go over the blue with another layer of paint, and make the green larger, and try these colours one more time. Or perhaps there is something about the way these particular paint colours reflect light that is not so friendly with the colour sensor.
If my memory of my childhood is correct, I think I had fabric paints that dried a lot faster and were a lot thicker than these ones before. These paints seem to soak through the fabric a lot more and take 4 hours to dry. I may visit Michaels soon and see if they have any other brands to try.
Update: After going over the green and the blue one more time, the blue is not any better. I was able to get the LEDs to register the green once. This is not good enough results for a performance situation. These two shades are darker than the other paints, so maybe it has to do with that. The orange also works, but I find it looks rather similar to red with the LEDs. I'll look to see if there are lighter blue and green paints that I could try.
Another note, I think I may have written about the wired connection issue already (can't remember). Basically, the ESP8266 refuses to spit out data through serial unless if the computer is connected to the wireless network first. Then after it starts printing the data through serial you can disconnect the wireless network and it is fine. This didn't start happening until we added the update to send brightness data from Max to the GLOBE. It hasn't been a problem so far since I'm not doing public performances where that backup method is needed (yet). While I was testing for the colours, I noticed that the GLOBE refuses to turn on its LEDs or change colours unless if it first connects to the computer wirelessly - which I think is related. It's like the ESP will not do anything until after a handshake is made, which is hopefully something I wish to fix (though it's lower on the priority list).
If you are looking for a summary for my Masters thesis, it is here.