Probabilistic Sampling

Rather than a static accept/reject for HSV, we accept or reject like... 10%-60% with probability p=the hsv number.

This should bias us towards having useful patches, while not completely tossing lighter ones, or ones with different information.