Eyeballing damage isn't going to give us concrete data unless you're willing to do some very tedious parsing, but there is a method that can. /ra damage is basically static at capped pDIF. Find a zone that has both a high-level monster (to damage you) and a nearby low-level monster (to test with). Use a crossbow and compare damage with/without Scarlet Delirium. Tests on magic won't require any special equipment.
I used West Sarutabaruta mandies and T2 VNMs to conduct some brief tests earlier. After taking ~half of my HP in damage, my ranged attacks on the mandies did 13.3% more damage (170 vs 150, possible 35/256 enhancement when accounting for flooring?). Absorb-INT went from 22 absorbed without Scarlet Delirium to 30 with it up and most of my HP lost. I also tried Blizzard III but couldn't seem to get higher than a 3% bonus, didn't get a chance to test magical WS but I do intend to go back and Herculean Slash a few NMs eventually, as well as reviewing my other tests if no further information has arisen from properly structured tests. However, it will be a few days before I can do so.