Help me make sense of the weapon tooltip dps

embermanemberman Posts: 729
edited July 2015 in TL2 General Discussions
I am looking at the Brasstaxx. In case the link doesn't work, it is a
Level 30 Unique Pistol
272 Damage per second
Attack speed (.96 sec)
69-138 Physical damage
69-138 Fire damage
1 socket

+54 physical damage
-7 armor to all armor per hit
+30 knockback

The 272 dps number displayed at the link makes sense if you add up the average physical and fire damage components and divide by the attack speed 0.96. However, in game, it is displayed as 258 damage per second. Why are these numbers different? It can't be due to any stat or gear differences because I checked the same exact weapon on two different characters with vastly different stats and equipment, and both reported the same exact tooltip display of 258 dps. What is accounting for this difference?

On top of that, I socketed in a Venom Ember Chip (+14 poison damage) and the displayed dps went up by 10. It should have gone up by roughly 14 given the attack speed of 0.96. Why did it lose 30% of the gem damage?

Addendum
I have found an unverified post that explains the difference in the tooltip dps. Unfortunately this information seems to clash with other calculated numbers displayed elsewhere in the game.
d00d wrote:
On weapon dps math:

Flat increases in weapon dmg aren't just added to your weapons min and max dmg. Instead (from my testing), the listed damage increase is added to your max damage and a scaled down amount is added to your min damage.
For example, you would expect a Skull of Vellinque to increase the DPS of a Speed 0.4 Claw by
200/0.4=500 DPS.
Instead, it increases it by 200*0.85/0.4 = 425.
0.85 is the "average dmg factor": (min dmg + max dmg)/(max dmg*2) = 0.85 (for claws)
For swords, its only 0.75.
For axes, its 1!

According to this post, you need to convert the +54 physical damage effect into +27-54 damage instead, and also the +14 venom ember chip into +7-14 damage instead. If you use those modified number ranges, the simplified dps tooltip comes out to the displayed 258 (and 268 with the venom ember chip). However this doesn't make any sense as it is not consistent with how dps is calculated and displayed almost everywhere else in the game. Given the above I am inclined to think this tooltip dips number is calculating damage in a unique way that doesn't match the rest of the game and should be ignored.

Comments

  • GeistGeist Posts: 128
    emberman wrote:
    An unacceptable reason is "I just want you to do the full weapon dps calculation every time you want to compare two weapons because it is important to know the exact effect of all your stats and gear even though they are the same for both weapons because you are only comparing the two weapons ceteris paribus and nothing else".

    Not to make "that kind of post" but the game does that calculation for you if you just press J. Arcane Statistics should tell you anything you need to know when comparing gear.
  • embermanemberman Posts: 729
    The arcane statistics only work on weapons you can equip. The vendors in this game don't have a return policy of any sort so I can't try their weapons before buying.
  • ChthonChthon Posts: 1,855
    This is fascinating. I wish I had time to look into it deeply right now. Since I don't, wold someone mind running this experiment?

    Mod up a monster with huge hp and zero armor.
    Mod up a weapon with level 1, min 1, max 100, 1sec attack speed.
    Make a new character.
    Console spawn the weapon.
    {question 1:} What does the tooltip say? The graph output for level 1 is 21, so I'd expect a range of 1-21 and a DPS of 11.
    Console strength -X to reach 0 strength.
    Now console spawn the monster, console aifreeze, and hit the monster repeatedly.
    {question 2:} Does the observed range match the tooltip?
    Mod up a vellinque clone that can be put in a level 1 item. (May need to nuke the graph... I don't recall how the socketable reqs work off-hand...)
    Console spawn the vellique clone. Plop it in the weapon.
    {question 3:} What does the tooltip say now? (The common understanding -- which, based on your observations, appears to be wrong -- would predict 411. d00d's theory would predict 211.)
    Hit the monster again.
    {question 4:} Does the observed range match the tooltip? (The common understanding would predict 400-421. d00d's theory would predict 4-421.)
    Torchlight 2 Rapid Respec - Putting the "hack" in "hack-n-slash"
    StashNinja - INFINITE Stash for Torchlight 2
    NullMod - Play together in the same multiplayer game with different mods!
  • embermanemberman Posts: 729
    Looks like I might finally have to learn how to do some modding.
  • gytfunkegytfunke Posts: 571
    No need to create a new monster. Just type 'punchbag'. Punchbags have 0 armor.
  • embermanemberman Posts: 729
    Ok I have some possible results but first, two questions to clarify some important numbers.

    First, I used a clone of the Outlander's starting "Initiate Matchlock" to test with. The weapon has an observed attack speed of 0.88, but its SPEED value was 115. In order to get an observed attack speed of 1 I had to set SPEED to 125. This seems wrong to me so I just wanted to check if this is indeed how you are supposed to do things.

    Second, Vellinque's Skull seems to provide +200 physical damage for me, while Chthon you seem to be assuming it is +400. Again, just a quick check to make sure I didn't **** something up horribly. The item database here also says +200 so maybe you remembered it wrong?

    Anyways, on to the test procedures and results.

    For the mod, I cloned pistol_n00 (the initiate matchlock) to testpistol, set its SPEED to 125, and its MINDAMAGE to 1, and SOCKETS to 1. I also cloned tl2_skull044 (Vellinque) to tl2_skull044test and fiddled with the level requirement graph so that it requires item level 1.

    For the test procedure, I spawned testpistol in game. This resulted in an observed weapon with Attack Speed 1, Tooltip DPS 11, and Physical Damage 0-21. Note that is not 1-21. I set my strength to zero, spawned a punchbag, and hit it a lot of times. Observed damage range was 0 to 21, inclusive, in the combat log but 0's would be turned into 1 damage purple numbers. Next I put the Vellinque in the weapon, which resulted in a displayed tooltip dps of 112. However, hitting the punchbag with this weapon resulted in an observed damage range of 200 to 220, inclusive. Just for further confirmation, I added 7 points of shadowshot to this test outlander so that shadowshot deals exactly 100% weapon dps on its primary hit. Testing with Shadowshot 7 (and no charge) resulted in the same damage range (200-220 inclusive).

    Based on the results, I am inclined to conclude that the displayed tooltip dps is doing a calculation that is completely wrong for how damage is actually calculated, which makes it untrustworthy and misleading. Specifically, it is incorrectly downscaling any flat +damage sources on the weapon when calculating the displayed tooltip dps.
  • gytfunkegytfunke Posts: 571
    Thanks for the research, Emberman. That makes more sense than d00d's hypothesis.
  • ChthonChthon Posts: 1,855
    Good work Emberman!
    emberman wrote:
    Ok I have some possible results but first, two questions to clarify some important numbers.

    First, I used a clone of the Outlander's starting "Initiate Matchlock" to test with. The weapon has an observed attack speed of 0.88, but its SPEED value was 115. In order to get an observed attack speed of 1 I had to set SPEED to 125. This seems wrong to me so I just wanted to check if this is indeed how you are supposed to do things.
    I'm not sure. It's been a long time since I've mucked with weapon speed. The only thing I recall is that the actual in-game attack period is faster than the tooltip's displayed period. (Which isn't exactly relevant here.) My guess is that you did it right, since the numbers seem right for 1sec. Even if you did it wrong, you used a constant, so your results should still be valid.
    Second, Vellinque's Skull seems to provide +200 physical damage for me, while Chthon you seem to be assuming it is +400. Again, just a quick check to make sure I didn't **** something up horribly. The item database here also says +200 so maybe you remembered it wrong?
    Entirely possible. Apparently, my memory isn't what it used to be...
    Anyways, on to the test procedures and results.

    For the mod, I cloned pistol_n00 (the initiate matchlock) to testpistol, set its SPEED to 125, and its MINDAMAGE to 1, and SOCKETS to 1. I also cloned tl2_skull044 (Vellinque) to tl2_skull044test and fiddled with the level requirement graph so that it requires item level 1.

    For the test procedure, I spawned testpistol in game. This resulted in an observed weapon with Attack Speed 1, Tooltip DPS 11, and Physical Damage 0-21. Note that is not 1-21.

    That's OK. I was kinda guessing that it would round up to 1 since everything else in the game does.
    I set my strength to zero, spawned a punchbag, and hit it a lot of times. Observed damage range was 0 to 21, inclusive, in the combat log but 0's would be turned into 1 damage purple numbers. Next I put the Vellinque in the weapon, which resulted in a displayed tooltip dps of 112. However, hitting the punchbag with this weapon resulted in an observed damage range of 200 to 220, inclusive. Just for further confirmation, I added 7 points of shadowshot to this test outlander so that shadowshot deals exactly 100% weapon dps on its primary hit. Testing with Shadowshot 7 (and no charge) resulted in the same damage range (200-220 inclusive).

    Based on the results, I am inclined to conclude that the displayed tooltip dps is doing a calculation that is completely wrong for how damage is actually calculated, which makes it untrustworthy and misleading. Specifically, it is incorrectly downscaling any flat +damage sources on the weapon when calculating the displayed tooltip dps.

    I am strongly inclined to agree.
    gytfunke wrote:
    That makes more sense than d00d's hypothesis.

    Actually, it confirms d00d's hypothesis, since d00d's hypothesis would predict the tooltip says 111, and 112 is probably only off by a rounding error. The big find here is that the tooltip is completely and utterly wrong.
    Torchlight 2 Rapid Respec - Putting the "hack" in "hack-n-slash"
    StashNinja - INFINITE Stash for Torchlight 2
    NullMod - Play together in the same multiplayer game with different mods!
  • gytfunkegytfunke Posts: 571
    Oh, it seemed to me that d00d was assuming the tooltip was displaying correctly and the DPSes of different weapons were affected differently by flat damage gems.
Sign In or Register to comment.