I think the OP is talking about selective breeding, a bit like what has been done in carpet pythons, dogs, cat, horses . . . Really anything where just breeding for simple mutations isn't the point, but playing with the polygenetic side of inheritance, as well.

Unless I mis-read the initial post, it kinda looked like you answered your own question. If not, I apologize. It's quite late here. Really, it's just a long, patient, generational process. Breed what looks best, what's 'typiest' in each generation to each other, looking for improvements and dropping animals that don't improve the look from the breeding program. Pastels are actually a great example, as they suffered from a deficit in selective breeding when they became super popular; pastels were thrown to any old female and the resulting babies were often dark and browned out quickly, whereas pastels from brighter normals tend to hold color much better.

Also in pastels, if we're running with that, they have the tendency to brown out with age. About how bad they will brown out is fairly predictable looking at babies, as babies with more orange turn more brown . . . meaning less orange and more yellow in a hatchy is something to select for. Also waiting longer to breed pastels, for a better idea of how individuals look as adults, can be a way to select a more moderate brown-out. But it's still polygenetic, with an unknown number of modifiers, and takes a lot of time, and a lot of work. Folks do it, though. It can definitely be worth it.