So someone at Reddit posted this interesting comment in regards to my inquiry.
I think it might be the way the preview works.
The preview probably compares the position of your character to the position of your cursor. The actual attack likely uses the position of your character and the anchor point of your target.
This likely isn't the full story, but it might be part of it, judging from how I was getting advantage/normal roll previews based on nothing but moving my cursor around the enemy model. It made me think back to that one specific bug that Larian kept emphasizing in the patch notes. Three times to be exact. That whole thing about the one NPC peeing higher than should be anatomically possible from a visual perspective.
What if a different but similar version of that is at play here? As in, perhaps the previews are accurate, the real issue might be that the game could be reading our positions incorrectly upon launching an attack (as in, calculating our characters to be slightly higher in elevation than we actually are)? It would explain myself being able to get advantage rolls when the preview says I shouldn't, and it would also explain disadvantage previews resulting in a normal attack roll at the same time. A widening of the break points should mean that it would be harder to get a low ground disadvantage penalty AND a high ground advantage bonus at the same time. But I have not observed advantage previews turning into normal rolls throughout any of my tests, only the exact opposite.
Of course, that also brings up the question, why does this possible bug favor only us? To which I would answer that it's not really just us. The calculations probably only adjust the attacker upwards, not the defender, and we don't get that much visual data about how the enemy AI calculates high ground/low ground.
I'd say the UI is definitely bugged. I've got some high ground tests to supplement that. I'm still leaning towards the logic to granting advantage/disadvantage has been changed. It may technically be bugged right now too, but combat has been a lot more fun not having advantage and disadvantage be everywhere xD
I think the logic for the UI was also changed from patch 3. I'm not yet ready to go back to patch 3 to full test it out as combat was a slog, lol.
The crypt seemed like a smart choice to test out high ground, since the fight started the player with high ground advantage in Patch 3.
Here we can clearly see high ground advantage is not being given to Gale.
Gale in the CryptNow we can see the result matches the UI, something in the logic/code has changed.
The following combat logI chose to confirm this with Astarion, since he has dark vision.
Astarion's prediction in the UIThe result in the combat logWe see an expected outcome from the change.
Now check out what happens when Tav targets an enemy a greater distance away.
UI says it will be a normal attackThe game rolled the dice with advantageIt seems like there is a minimum distance check now for high ground advantage, and the UI is struggling to keep up with the change.
Here is Tav attacking a closer enemy from the same spot/elevation.
The UIThe Combat LogAgain, the attack rolls are consistent with positioning and the success rate has been what we would expect from Advantage/Normal/Disadvantage. The UI is a bit of a mess explaining what roll you will get.
If you don't have percentage to-hits that are even multiples of 5, you have Advantage or Disadvantage. Those images clearly show that you are at disadvantage from low ground (56% isn't an even multiple of 5) and the combat log is what is messed up (it isn't showing the disadvantage).
What is really interesting is that with repeated tests, the outcome is what a player would expect if it was a normal attack roll. The combat log seems to be showing what was actually rolled by the game. It's definitely different from Patch 3 and combat has been more fun because of it.
Overall it seems like there are more checks in the game over whether advantage or disadvantage should be applied.
Disadvantage was not applied.
