syncing stim times and subTR grid for gPPI

Hello AFNI experts,

I’m hoping to do a PPI analysis, given a multiband EPI time series data with a TR of 766ms and stimulus duration of 500ms (jittered fixation crosses are also multiples of 500: e.g, 3000, 5000, 10 000, etc.). For precisely synchronizing the stimulus onset times with a subTR grid, I would need to downsample my time series to a subTR grid of 2ms, which is not possible using 1dUpsample (upsample factors possible seem to range between 2-32). Given that, the best option I can think of is to downsample to a subTR grid of 383ms (using 1dUpsample with an upsample factor of 2). My questions are: is there a way to synchronize precisely my subTR grid and my stimulus times? If not, would an approximation (e.g., downsampling to a TR of 383ms) produce usable results? One of my candidate PPI regressors is blocked (negative emotion picture block; neutral emotion picture block), and the other is event-related (attention shift required vs not).

Thank you!!