Database design advice needed regarding including nullable fields

Question

As seen in the following DB design:

• each record in Table2 has a type
• most of the records in Table2 belong to say, type one
• type one records in Table2 are linked to Table1 using table1_id
• that means, if I wanted to find out the value of a type one attrib1 record, I would join Table2 and Table1 based on table1_id
• but since all records in Table2, regardless the type, need the same columns (attrib1, attrib2 and table3_id), I feel forced to leave those columns nullable while all other records that belong to other types will get their values inserted

I narrowed my options down to:

• stick to the current design by leaving the fields as nullable
• change the fields to be not nullable and insert into all values required for type one records, which smells like unnecessary redundancy to me since I could determine those values through the simple join mentioned earlier with Table1
• another way around is to split Table2, according to the type, into smaller tables, but the problem here is that I need a unique id for each record regardless its type; therefore, splitting Table2 would mean that some of the records in all these tables would have identical ids since those ids would be incrementally seeded, just like table2_id column currently is.

Both Table2 and Table1 are linked to Table3 for the sake of data integrity and so are attribute1 and 2 to other tables but omitted here for the sake of simplifying

Should I stick to this design since I could easily filter Table2 by type_id?

What is a better more robust approach to this dilemma?