Fruit color in a particular plant is controlled by a set of three QTLs 

By using a punnet square to do a test cross of a heterozygote ‘AaBbBb’ (the trihybrid) with a homozygous recessive ‘aabbcc’ (tester parent), you can obtain the eight possible offspring genotypes I posted. Your question asks "what proportion of the offspring will have a fruit colour phenotype that is different from both the trihybrid and the tester parents?" Your post also says that each dominant allele is a ‘dose’ of yellow pigment. A heterozygote ‘AaBbCc’ has three dominant alleles, so you can think of it having a pigment level of three. A homozygous recessive ‘aabbcc’ has no dominant alleles, so its pigment level would be zero. So basically, the question is asking “how many of the offspring have either a pigment that is different than three or zero”. Of the eight different possibilities for genotypes, you’ll see that one offspring had the same genotype as the heterozygote ‘AaBbCc’ and another had homozygous ‘aabbcc’. So out of eight total different possibilities, 6 will be different than either the trihybrid plant or the tester parent. 6/8=3/4. I hope that helps!

How is the parental/tester phenotype 2? You lost me with the phenotypic ratios.