enter code hereI want to use samtools mpileup to achive batch extraction of variant quality information from CRAM files

I start with a CRAM on which i do samtools mpileup -B -f "$reference_genome" -r "$region" -q 0 -Q 0 "$cram_file" And i get :

chr11 | 56094117 | G | 25 | FALSE | ..A,..,A...AAa,,.,..,,,., | ?!??]!]?!???!]!]?]?!]????

But when i use samtools tview --reference -p "$cram_file" Then i get :

can clearly see the samtools tview first line :there are 4A 1a (no color and no underline ),it means these 5 reads are Q > 30. But when i caculate the samtools mpileup ASCII code ,there are just 3A 1a (Q > 30). Which one has the truest reads quality results? Has anyone else had this happen to them?