It is 2025, and everyone knows about large language models. The world is divided: while some companies want to feed all texts in the world to their next-generation models, some other companies want to keep their texts to themselves and to their human clients.
A new company called CloseAI started to develop a captcha designed to protect human-oriented resources from large language models. To do this, they employ the fact that the latter tend to happily digest and interpret any text, while humans are mostly capable of interpreting the primary message and ignoring the non-important words. The captcha is hence a piece of ASCII art that contains a message for a human and some text in its "pixels" aimed to mislead robots.
The CloseAI representatives hired you to produce the initial prototype, which will be dealing with names of rainbow colours, that is: RED, ORANGE, YELLOW, GREEN, BLUE and VIOLET. Your program will receive two different colour names as input, $$$C_1$$$ and $$$C_2$$$. It has to produce a rectangular ASCII art image that would satisfy the following conditions:
Obviously, using a human to check the output of your program would be too expensive. Instead of a human, a simplified model is going to be used, which:
For the purpose of the latter step, this model assumes that:
Obviously, under these conditions all rainbow colour names can be distinguished easily one from another. This model is far from being human-grade though, as it would also read the following image to contain the word "RED" (one hole, zero holes, one hole):
OOO..P....VV
O.O.PPP..V.V
OOO..P...VVV
Now your task is to implement a captcha generator that satisfies all these requirements.
$$$^{\text{∗}}$$$The neighbours of a cell are the four cells that are: directly above, directly below, directly to the left and directly to the right.
Output the ASCII art picture satisfying the requirements.
BLUERED
BB..L...U.U.EEE B.B.L...U.U.E.. BBB.L...U.U.EEE B.B.L...U.U.E.. RED.RED.RED.RED