Is it necessary to use the aligned BAM file to calculate k-mer in GENOTYPE stage?

[GouZiHui]

Hi ！
Thanks for developing this powerful tool!
Can I use Kmers calculated by KMC3 directly from paired-end reads in fastq format generated by paired-end sequencing without aligning them to a reference genome? Or must the paired-end reads be aligned to the reference genome first, and then Kmer statistics calculated using the generated BAM file? What are the differences in genotype calling?
Best wishes!