gvcf or final vcf used to construct accession specific genome

I have 100 rice accessions and its DNA sequencing bam files. My goal is to construct the genome in a specific region for each rice accession. Could I directly use the gvcf file from the variant call? Or it is better to use the vcf file after final genotype call? If I use gvcf file, is there any filtering step I should use to guarantee the quality? Thanks.