Terraria Wiki talk:Projects/DPS

Average crit chance

How would you determine average crit chance? Maximum and minimum are easy, but there needs to be a defined "average" that would be used. Would it be halfway between the max and min? Or would it be what the average player would have? If that, would it be what the average player has when that weapon is typically found, or would it be what the average player has at end-game? Even then it brings up the question of what you can consider to be used by the average player. Every world generates differently, I had one generate recently that had absolutely no movement accessories in surface chests, when usually I find one. Beyond that, every player plays differently. Some players will grind for weapons and accessories and reforge for optimal stats, while others may just play with what they find.

This is by no means a criticism, I think this would be a great addition. This is just some food for thought when planning out what all you want to cover in doing this. Acbooster (talk) 21:09, 15 January 2019 (UTC)

I think what is meant by "average crits" is to assume that, in the long run, crits show up with the frequency they're supposed to. So, just use the stated crit chance on the weapon (4% for most weapons I believe). If you swing a weapon that gets a crit 4% of hits five times, it's entirely likely you won't get any crits. You may even get 5 in a row, although that's obviously unlikely. But if you swing that weapon many thousands of times, as you might with some play styles, you should expect to see 4% of those swings produce crits, to a reasonable precision. What we can do then, to find the average damage with average amounts of crits, is to take a weighted average: .04*(critical hit damage) + .96*(normal hit damage) = average damage.

Since we're on the topic, i really don't think a "all crit" column is really necessary. The highest crit rate in the game is something like 30%, so you're never going to have a long run of many crits in a row. I don't feel it's helful info, even if it's interesting.

I also have a concern regarding how the base damage is to be calculated. For simple weapons like pre-hardmode swords, it seems obvious that we assume the weapon hits. But plenty of wepons have multiple damage modes, or even just several shots. Do we count all parts as having hit in the calculation? For instance, with Beam Sword, is the projectile included or not? And with Daedalus Stormbow, do we make the usually unreasonable assumption that all the arrows hit? If not, how many are assumed to hit? I think a more fleshed out set of rules are needed for calculating these values, but I don't feel I'm experienced enough to make them haha. Also, 1.4 comes out in less than 2 weeks as of my writing this, and it sounds like a lot of weapons are getting reworked, so dps calculations done now might be wrong soon. Jaekymba (talk) 01:29, 6 May 2020 (UTC)

These are my suggestions for the ground rules, but unless others agree I'm in no way suggesting these should be used:

- Use the base stats for the weapon (default modifier, no armor or equipment, no potion effects)

- Calculate the average dps over 10 seconds of constant usage of the weapon (so shooting the weapon as often as possible for 10 seconds for example) - taking a video helps with this and provides a more reliable dps. Over 10 seconds could also be used but is not recommended (to keep results consistent and therefore more reliable with each other)

- Use one target dummy

- Beam swords should have it done twice - once at melee range and once for the beam only.

- Weapons with a ranged attack (the beam only part of the beam swords too) should be tested from 15 blocks away (average ranged weapon usage range for most weapons).

- Exceptions to the above room are spread-based weapons such as vampire knives and shotguns, which should be tested at 5 blocks range.

These are just my suggestions, feel free to add/change any! DanZeMan (talk) 19:50, 2 June 2020 (UTC)

Inconsistencies between weapons

I don't think having a single DPS value is enough to measure how well a certain weapon goes, and that's assuming all weapons can be measured using the same fair rules.

You could get a theoretic DPS value calculated based on damage, max speed, critical chance and projectile count, but it wouldn't be representative of the weapon during gameplay. That's because weapons have spread inaccuracies (i.e Uzi), spread projectiles (i.e Vampire Knives), homing (i.e Magic Missile), delays based on distance (i.e boomerangs), dependence on accessories (i.e yoyos), damage over time (i.e Daybreak), large areas of effect and inaccuracies (i.e falling stars/arrows), large or short piercing range (i.e Piranha Gun or flails vs Solar Eruption), semi-piercing (i.e Terrablade, only 3 targets), etc. DPS will also wildly vary depending on the target (i.e slow moving projectile vs fast moving enemy, large spread vs large enemy, piercing vs multi-segment enemy) so a dummy isn't fair to use as target for all weapons. Certain high DPS weapons against dummies might actually do worse than other low DPS weapons against dummies, so then now you would need a second layer to help inexperienced players see how well weapon DPS does against all types of contexts because inexperienced players would now pick the highest DPS instead of picking the highest base damage.

It just seems to me it would be moving the problem somewhere else and not helping inexperienced players at all, while putting a lot of effort doing so. — Dig494^talk 14:11, 5 June 2020 (UTC)

I agree; this is not a negligible issue. The main motivation behind this project, quoting from the page, is to "help inexperienced players see how well high attack compares to high attack speed or crit chance, because most inexperienced players just see a high damage stat and pick that, but it is not always the best". While it is true that a weapon with a high damage stat is not always the best pick, attack speed and critical strike chance are not the only other relevant factors at all. It is virtually impossible to compute a single number that is supposed to rate the usefulness of a weapon under every possible circumstance.

I don't want to discredit the damage per second stat as a whole, I just don't feel like it is a worthwhile addition – if carried out on a large scale, that is. I can see its usefulness in clearly confined spaces, e.g. comparing the Megashark to the Minishark: Both weapons are functionally basically identical, are not affected too much by the many variable factors Dig494 outlined above, and their DPS can be determined quite reliably. Perhaps the focus of this project could shift more in this direction. --Rye Greenwood (talk) 05:42, 17 June 2020 (UTC)

Seperate article

As said already by Rye Greenwood, giving a weapon only one DPS value for one specific situation isn't optimal. But intead of listing all the details on the weapon's page, wouldn't it be better to make a seperate article for a sortable table first? This way, new players can see which weapon performs best in chosen situation without switching from one weapon's page to another. The list would then have a table for every category of weapons, each having multiple columns for different situations (e.g. Boomerangs having values for short and long range).

Schwetziktheepic (talk) 17:31, 21 July 2020 (UTC)

Using Mobile

On the mobile version every weapon has auto swing so we can more accurately test the dps for weapons that don't normally have auto swing. Colinzer (talk) 16:11, 23 December 2020 (UTC)

I agree, but if you measure it on mobile, then what is it going to be like for PC players? Are we only measuring this on PC? Or mobile too? What about console? Here is another thing I want to say: on mobile every weapon has autoswing. What about the people that are on PC or Console? Every weapon on PC doesn't have autoswing. Neither does Console. Thanks12 (talk) 13:21, 25 December 2020 (PST)