(The lazier way, can still be helped by using the temporaries separately instead of chaining their changes)

Temp version before benchmarking