Use Dense_Rank() to Calculate 2 Different Distinct Counts

I need to calculate two different distinct counts in the same query, based on different criteria. I'd like to accomplish this with dense_rank() if possible.

I need columns for:

1. By resource, how many unique customers are in category X?
2. By resource, how many unique customers are in category Y?

Sample data:

Expected result:

DROP TABLE IF EXISTS #test_data 

CREATE TABLE #test_data
(
line int 
,res varchar(1)
,cat varchar(1)
,customer_id int
)

INSERT INTO #test_data (line,res,cat,customer_id) VALUES (1,'A','X',1)
INSERT INTO #test_data (line,res,cat,customer_id) VALUES (2,'A','X',1)
INSERT INTO #test_data (line,res,cat,customer_id) VALUES (3,'A','Y',1)
INSERT INTO #test_data (line,res,cat,customer_id) VALUES (4,'A','X',2)
INSERT INTO #test_data (line,res,cat,customer_id) VALUES (5,'B','X',1)
INSERT INTO #test_data (line,res,cat,customer_id) VALUES (6,'B','Y',1)

SELECT  DISTINCT 
    RESOURCE = t.res
    ,[UNIQUE CUSTOMERS CATEGORY X] = CASE WHEN t.cat <> 'X' THEN NULL ELSE DENSE_RANK() OVER (PARTITION BY CASE WHEN t.cat <> 'X' THEN 0 ELSE 1 END, t.res ORDER BY t.customer_id ASC) END  +
                                        CASE WHEN t.cat <> 'X' THEN NULL ELSE DENSE_RANK() OVER (PARTITION BY CASE WHEN t.cat <> 'X' THEN 0 ELSE 1 END, t.res ORDER BY t.customer_id DESC) - 1 END
    ,[UNIQUE CUSTOMERS CATEGORY Y] = CASE WHEN t.cat <> 'Y' THEN NULL ELSE DENSE_RANK() OVER (PARTITION BY CASE WHEN t.cat <> 'Y' THEN 0 ELSE 1 END, t.res ORDER BY t.customer_id ASC) END  +
                                        CASE WHEN t.cat <> 'Y' THEN NULL ELSE DENSE_RANK() OVER (PARTITION BY CASE WHEN t.cat <> 'Y' THEN 0 ELSE 1 END, t.res ORDER BY t.customer_id DESC) - 1 END
FROM
    #test_data t

Incorrect results (I don't want the NULL values and I want two lines instead of four):

You seem to be looking for conditional counts:

select resource,
    count(distinct case when cat = 'X' then customer_id end) cnt_cust_in_x,
    count(distinct case when cat = 'Y' then customer_id end) cnt_cust_in_y
from #test_data
group by resource

I don’t see how dense_rank would be helpful here.

In databases that support the standard filter clause of aggregate functions (such as Postgres):

select resource,
    count(distinct customer_id) filter(where cat = 'X') cnt_cust_in_x,
    count(distinct customer_id) filter(where cat = 'Y') cnt_cust_in_y
from #test_data
group by resource