Thank you, Krakob.

I'm very certain I made an attempt at measuring the guardian shield's defences and hit points at some point, but I can't remember where I wrote the result down.

The method itself involved hitting the shield with something that did X amounts of damage, and then something that did Y amounts of damage, then observing the ratios of of the length of shield bar that got reduced, comparing that to ratios the damage values would have if they were undefended, and kinda of zequaling it from there.

I seem to recall the health amount being more than a knight without trinkets but much less than a knight with trinkets, so 775 doesn't sound wrong.

Edit: It was right here! The number I arrived to was 575.

@Lady-of-War:
You know how your weapons do different amounts of damage depending on depth? Weapons do the damage they would do in Depth 25 for T3 Lockdown. I think T2 is depth 15.

I think I get it now. So, would that make T1 be set at Depth 5?

After all:
T2 = D15
T3 = D25

Retequizzle.

I know the buttons can be a bit elusive.

Let me do that maths again I guess.

My post doesn't state this, and this is obviously kinda critical to the end-result, but tests I performed would almost certainly have been with max damage bonus. I must have went with as large damage numbers as I could to avoid treshold shenanigans. To be clear on this, it was long ago and I don't actually remember. Should probably perform new tests altogether.

Let's bring up the old chart of Lockdown sword damage.

Flourish: 412 (swipe) and 618 (charge).
Hammer: 477 (swipe), 358 (dash) and 715 (charge).

Let's go with just the highest and the lowest number for now. The ratio in question i 715/358 ~= 199.72%

The damage in pixels for those attacks were 352 and 133. This ratio is 352/133 ~= 264.66%

To find our defence, we're looking for a number where (715 - x) / (358 - x) ~= 2.6466165413533833

I'm lazy so I just throw that into wolphram alpha and get a number around 141. Okay then.

This puts the damage for the hammer charge at 715 -141 = 574.

574 * 360 / 354 = 583.

I probably chose some other attacks to go off of, but you can see how I ended in the neighbourhood of 575, certainly.

My post doesn't state this, and this is obviously kinda critical to the end-result, but tests I performed would almost certainly have been with max damage bonus.

That would make a world of a difference. The part that confused me though was the 123.5% which appeared during my +0 damage bonus. When I find time again, I'll take another look with the max damage numbers.

@Krakob

The range of deathmarked (armorless) damage is after rounding (where damage is rounded up, even 19.000001 would result in 20 damage).
If the only damage value recorded (more often than not comparing green/black in the data) is only the +6, you add an additional 4% per boost.

The same thing applies when comparing damage against armor.

An example using my proto sword D2 numbers in an attempt to predict the neutral armor... [for simplicity, lets assume my 24.167 raw damage estimate is 100% accurate].
If I only took one unarmored damage value... lets say +4 (the 29 recorded, which is actually about 0.999 off from the actual value). This brings the raw damage estimate to be exactly 25 (off by 0.833).
Now lets get to armor. The theoretical armor is pretty close to 13 (between 13.00 due to +0 20% pickup and 13.03 due to +4 armor). However, if I were to again take the reduced damage value at +4, the very limited data would suggest 14 points of damage reduction.

This is the most extreme simple example I can find, which results in an error of 1 during the armor estimation. Hypothetically, the error may be as high as 2 in a worst case scenario. Extremely unlikely, but possible.

Did I miss the Vitasuit Deluxe?