The content-adaptive LOP2 loop-filter was adopted into NNVC-9 software. Since a new version of LOP loop-filter, namely the LOP3, was also adopted into the NNVC-9 software, this contribution reports the content adaptation per random-access segment for LOP3 loop-filter. The content-adaptive parameters are signalled via a specifically-designed Adaption Parameter Set (APS) called Neural Network Filter Update (NNFU). It is reported that applying this approach in the RA configuration yields average coding gains of 1.02% (Y), 2.27% (Cb), and 1.43% (Cr) compared to NNVC-9.1 with NN intra tool ON and LOP3 tool ON.