Add an option to encode two bits in a square of 2 * 2 = 4 pixels.

[Nemgathos]

Because we do not know whether YouTube will one day change its algorithm to compress full HD videos that consist of 1,920 * 1,080 = 2,073,600 pixels, a more robust way to encode files in a video would be to encode one bit of data in a 2 * 2 = 4 pixels square. This would result in (1,920 / 2) * (1,080 / 2) = 518,400 bits per key frame.

In order to prevent files from becoming about four times at large, I would suggest to use colour coding that does not fall victim to chroma subsampling. Four possible states could be distinguished by using the complementary colours black (0, 0, 0) and white (255, 255, 255) as well as red (255, 0, 0) and green (0, 255, 0).

Maybe, the video files can even be compressed a little bit because the probability that information is lost when a full HD video that consists of 2 * 2 pixel squares is uploaded can be eliminated.

Would this be a robust way to encode files even if YouTube decides to change its compression algorithm for full HD videos? What do you think?

[mevimo]

Using 4 pixels squares instead of single pixels could extend YouBit's ability for robustness. However, that is not what it was designed for and I would personally rather see improvements in performance in the other direction (increase information density rather than decreased).
I would still approve a PR that adds this functionality however as it does not hinder anything else.

I am not sure I understand your point on colors. Why would these proposed colors not fall victim to chroma subsampling?