The f1 generation will take a gene from each of two parents, so imagine the original parents were an orange and a blue and orange is dominant, you would get the gene combination Ob (Orange from one parent blue from the other)
Two plants from the f1 generation crossed, with each plant giving one of its two Ob genes gives
OO Ob OO bO bb
That means if Orange is dominant three out of four plants will grow orange and one (bb) blue.
Also two of the four will have the same genetic background as the F1 generation.
It is true that OO and bb wil not be like the f1, but they will be like the original parents, which presumably were selected for their good qualities, so it is not really true that it is not worth saving seed from f1 hybrids.
Of course these are only averages, but Gregor Mendle worked out the concept of a 'Sex factor', a discreet piece of information passed by each parent, based on counting numbers after carefully breeding pure types of dwarf and giant beans and wrinkled and smooth peas then crossing them. This was long before chromosomes were discovered, and explained the fact that a new variant didn't simply get 'absorbed', which puzzled Darwin. Some people think he cheated because his numbers work so well, but this was in the 1860's in rural Austria, industrial pollution is quite likely to mean the experiments would not repeat exactly. He lost his eyesight trying to work out what happened with the special genes in hawkweeds, the tiny parts in the flowers meant they were almost impossible to cross pollinate in a controlled way at that time. A remarkable man.