Cumulative DISTINCTCOUNT with time intelligence

v-juanli-msft  really helpful example but we are not quite there yet...

Whilst measure 2 gives a DISTINCTCOUNT, it does so cumulatively over the whole reporting period. i.e. August is a DISTINCOUNT from the earliest date (March) to the August dated rows. The resulting lin chart shows an increasing trend over the full period; when each Monthly period actually has fluctuating DISTINTCOUNTS e.g. August has 3 DISTINCTOUNT values, but Measure 2 reports it as 4. I can get the monthly values using Measure 1 (DISTINTCOUNT of ID).

I need the values to use in a formula = ID count by Month / ID count by Quarter

Maybe creating a separate table with the Distintcounts over Day/Week/Month/Quarter/Year periods would be the answer?

Once we have the above working I need it to show the DISTINTCOUNT for Weekday to Date, Week to Date, Month to Date, Quarter to Date in the context of the full reporting period. 

Also, how can I calculate the DISTINCTCOUNT for the last 90 days to get a rolling 90 day value each day?