Calculating Physical Damage

[ringthree]

I could be completely wrong here, but I believe that I had correctly guessed the potential range for damage on my SAM against level 81 Greater Colibri (VIT 67 and defense 322)as 59-100 without any buffs and just in my TP build gear on (422 Attack and 98 STR with Hagun), but when I tested it out. It didn't work. These are the results that I got.

http://i236.photobucket.com/albums/f...test_graph.png

http://i236.photobucket.com/albums/f...Ctest_data.jpg

http://i236.photobucket.com/albums/f...Ctest_freq.jpg

All the damage are regular hits. No food, no other buffs. What am I missing or not understanding?

[aurik]

I could be completely wrong here, but I believe that I had correctly guessed the potential range for damage on my SAM against level 81 Greater Colibri (VIT 67 and defense 322)as 59-100 without any buffs and just in my TP build gear on (422 Attack and 98 STR with Hagun), but when I tested it out. It didn't work. These are the results that I got.

http://i236.photobucket.com/albums/f...Ctest_freq.jpg

All the damage are regular hits. No food, no other buffs. What am I missing or not understanding?

pdif is not an even distribution.

[ringthree]

I could be completely wrong here, but I believe that I had correctly guessed the potential range for damage on my SAM against level 81 Greater Colibri (VIT 67 and defense 322)as 59-100 without any buffs and just in my TP build gear on (422 Attack and 98 STR with Hagun), but when I tested it out. It didn't work. These are the results that I got.

http://i236.photobucket.com/albums/f...Ctest_freq.jpg

All the damage are regular hits. No food, no other buffs. What am I missing or not understanding?

pdif is not an even distribution.

Well, it still has a limit on the range of damage no matter the distribution, right? And knowing your cRatio you can find the limits of the distribution, but those limits don't match my observations.

I figured that pDIF would be some kind of normal distribution, definitely not even. And obviously there is not enough data to really draw any conclusions, but I figured that it would at least look better than this.

[Callisto]

Idk, I tried to look at it myself and came up with pretty much the same thing you did(59-101~, I can't remember which digit fSTR and pDIF are truncated at, so I'm probably a little off).

[ringthree]

Idk, I tried to look at it myself and came up with pretty much the same thing you did(59-101~, I can't remember which digit fSTR and pDIF are truncated at, so I'm probably a little off).

Well, Amele said that fSTR hasn't been tested in a while, and just that when they tested it appeared that adding 4 STR still increase fSTR by one, but I am wondering if something else was changed in the equation with most recent two-handed update.

I am definitely going to do more testing this weekend.

Does anyone have a parser that will collect all of the individual hits? Right now I am just reading through chat log dumps on DirectParse and entering them into a spreadsheet by hand. I definitely do not have the skills to change a parser to do that.

[aurik]

My data comes from ages ago (2006) so it's not unlikely that things may have been changed. Also I dealt with a lot higher attack value (500-600~ vs spartoi) and lower level, so my data may not be applicable to the entire spectrum.

That said.

pdif distribution isn't a "normal" distribution either. It's more like a flat distribution, with a "correction" for values outside of a certain region, which leads to huge spikes for the frequency of certain values.

[ringthree]

My data comes from ages ago (2006) so it's not unlikely that things may have been changed. Also I dealt with a lot higher attack value (500-600~ vs spartoi) and lower level, so my data may not be applicable to the entire spectrum.

That said.

pdif distribution isn't a "normal" distribution either. It's more like a flat distribution, with a "correction" for values outside of a certain region, which leads to huge spikes for the frequency of certain values.

Yeah, I remember when you posted that. It was really weird, I had experiences like... I just (naturally) assumed they would use something like an normal distribution, but I can see how a flat distribution would be easier to deal with in a game, it just wouldn't be as realistic I guess. A normal distribution would probably be much more difficult to deal with especially in cases of extremely high attack.

When you did your tests were you using your Spharai?

[aurik]

I have both pre- and post-spharai data.

Normal distributions take more calculation power, I'm not surprised they are using a simpler way to calculate.

[ringthree]

I have both pre- and post-spharai data.

Normal distributions take more calculation power, I'm not surprised they are using a simpler way to calculate.

Interesting. Do you think you can post your old data? If it really is a flat distribution then it should still be making a curve as more modifications are added, right?

[ringthree]

After doing a little testing with the calculator and working with the assumption that SE has changed the modifier for two-handed weapons in the cRatio part of the equation (I did this based on two assumptions, first that it would be the most obvious place for this to happen, and the second that SE mentioned this when making the changes) the multiplier for level difference has to be at least .030 or less. The problem with this is that though it matches the fMax of my results, the fMin results are much too high, and I got damage results much lower than that fMin would indicate.

So there would have to be some kind of change to the potential damage ranges also, if I am understanding this correctly. The fMax and fMin changes might well explain all of the results without the level difference multiplier being incorrect too. I am just not sure.

What is the best way to test what the damage cap is on Greater Colibri? Just get on my MNK, make sure that fSTR is capped and then pile on Minuets and attack food? Any suggestions would be helpful.

[Callisto]

After doing a little testing with the calculator and working with the assumption that SE has changed the modifier for two-handed weapons in the cRatio part of the equation (I did this based on two assumptions, first that it would be the most obvious place for this to happen, and the second that SE mentioned this when making the changes) the multiplier for level difference has to be at least .030 or less. The problem with this is that though it matches the fMax of my results, the fMin results are much too high, and I got damage results much lower than that fMin would indicate.

So there would have to be some kind of change to the potential damage ranges also, if I am understanding this correctly. The fMax and fMin changes might well explain all of the results without the level difference multiplier being incorrect too. I am just not sure.

What is the best way to test what the damage cap is on Greater Colibri? Just get on my MNK, make sure that fSTR is capped and then pile on Minuets and attack food? Any suggestions would be helpful.

I had actually been under the idea from my own dicking around with 1h pDIF that they may have actually changed the # that cRatio is multiplied by to find pDIF without changing the actual cRatio calculation, but I play on 360 and it's very difficult to test crap like this without a parser when I'm sitting watching the log and writing down damage #'s on a notebook(ghetto parse ftl). I'm also still rather new to playing w/ game mechanics though so I could easily be miles off, but that's my current guess at what was changed.

Edit: If it helps at all for your 2nd question, there's this: http://www.bluegartr.com/forum/viewt...ri+def#p473909

[ringthree]

After doing a little testing with the calculator and working with the assumption that SE has changed the modifier for two-handed weapons in the cRatio part of the equation (I did this based on two assumptions, first that it would be the most obvious place for this to happen, and the second that SE mentioned this when making the changes) the multiplier for level difference has to be at least .030 or less. The problem with this is that though it matches the fMax of my results, the fMin results are much too high, and I got damage results much lower than that fMin would indicate.

So there would have to be some kind of change to the potential damage ranges also, if I am understanding this correctly. The fMax and fMin changes might well explain all of the results without the level difference multiplier being incorrect too. I am just not sure.

What is the best way to test what the damage cap is on Greater Colibri? Just get on my MNK, make sure that fSTR is capped and then pile on Minuets and attack food? Any suggestions would be helpful.

I had actually been under the idea from my own dicking around with 1h pDIF that they may have actually changed the # that cRatio is multiplied by to find pDIF without changing the actual cRatio calculation, but I play on 360 and it's very difficult to test crap like this without a parser when I'm sitting watching the log and writing down damage #'s on a notebook(ghetto parse ftl). I'm also still rather new to playing w/ game mechanics though so I could easily be miles off, but that's my current guess at what was changed.

Edit: If it helps at all for your 2nd question, there's this: http://www.bluegartr.com/forum/viewt...ri+def#p473909

Well, my observations as of now are that at very least the fMax and fMin have to have been changed, so that sounds about right, but if that is the case then it is a pretty substantial change.

[Callisto]

I noticed the same thing when I was trying to do 1h, whatever it used to be, I think it was a 2.0 cRatio cap with a 1.2 multiplier, that came out to 2.4 fMax, then when I had been playing with it post-update the max I found was 2.46, forgot what the min was, but it led me to believe that the multiplier was now 1.23 with the same cRatio cap, maybe something similar but a different # for 2h.

Edit: Ok, crosspost, here's from ffxio what I had thought when I was testing it for 1h:

Back to the math end, I've been trying hard to figure out which part changed for the new pDIF, if it was the cRatio cap or the 1.2 multiplier. The only way I could think of to figure it out is to see parses from right before the patch and right after under the exact same conditions with uncapped cRatio/pDIF, as if it was the 1.2 multi being shifted to 1.23 even uncapped cRatio would get its damage boosted. At first I was thinking they would have raised the cRatio cap, but if the new Min for capped pDIF is indeed 1.58, which both my #'s and Armando's show, that would make the adjustment from Max to Min .88 instead of .8, and if they're going to make one of the pDIF changes in the conversion of cRatio to pDIF, it would make more sense to me that they would make both changes in that manner, and raise the 1.2 multi.