DPS Add-on Development

Yep.

It's several years old, but maybe it's a start?

http://www.ffxiah.com/forum/topic/19948/ffxi-melee-damage-comparator-v510-20march2011/1

oh wow Masumanai, there's a name I haven't heard in years.

I was going to send Night a PM to go over the details of what it would involve, and avoid giving tangible work to Rooks since I know he's always super busy (but planning on running the structure by him for suggestions).

If other people are interested in teaming up, they can PM me about what areas they would like to contribute to and I'll try to find a way to optimize the work distribution.

I'll probably also run by some UI prototypes here for feedback.

I had create the original version of that, Masa was doing most of the work by the v4-5 time frame, I don't remember.

The 80 hrs. estimate is pretty generous, for something static is it seems reasonable, but considering the game changes and updates for new gear etc.. its pretty hefty. Unless you're able to parse the dat file and auto load a db with all the stats (possible I'm sure but no idea the effort involved).

I remember I had it fairly well tuned for the 75 cap and then abyssea happened and it became very difficult to keep up. Even at the 75 cap the scope ended up far exceeding what I was expecting. (download the v1.1 version from google code, it is tiny by comparison)

I put it in a spreadsheet so others could view the work, tweak stuff and add their own buffs/gear. I considered some sort of stand alone app/website (in a perfect world probably would have done a perl/sql interface myself). The last thing I wanted to do was make some boneheaded error no one could see, I did get lots of feed back when things did not look correct.

Getting accurate mob data was another issue. I think Mote's still work pretty well, but relevant mob data....(and gear, but the gear is easy to add)

the Augmented gear is annoying to deal with as well, although the format is perfect when looking at Oseem for options.

More old threads of history in case anyone cares:
http://ffxi.allakhazam.com/forum.html?forum=10&mid=1237842147283719653
http://ffxi.allakhazam.com/forum.html?forum=10;mid=1233873439193787346
http://ffxi.allakhazam.com/forum.html?forum=10;mid=1232488844300441375

So to be clear the only thing this would offer over the spreadsheets is updating them and giving a UI? :/ I still don't think it's a very good use of time lol.

I swear just that would good enough to make more people choose gear according to spreadsheet and stop parse low. :(

Well, that's a good point. Though I'm more for function over form and I don't think people too lazy to ask others or without the time and motivation to spend an hour learning to use spreadsheets are gonna change much.

I believe a DB of all gear somehow added to spreadsheets, and searchable, and updated with updates with aug fields would go far into keeping spreadsheets up to date. There's still lots of unknowns though in the game, that when I get time for stuff like this I'd rather spend testing and figuring it out. Ranged damage calcs (especially for snapshot/rapidshot etc) and pet equations for damage (like what exactly is flaming crush's damage formula?) are some of the biggest ? still out there and I play both a lot so :/ Melee calculators have been done and redone in different forms so many times... imo the spreadsheets are the most lasting version of this sort of stuff we've ever had and I think they're great work. Excel definitely isn't easy access though, but I think a guide on how to use spreadsheets would be good. I figured it out easy with no excel background but went to school for computer science, excel wasn't taught as it was considered more for people getting business degrees though excel can take and use programming.

I don't want to curb anyone's appetite for making something good for the community or doing things their own way or making things more accessible though, so it's the last I'll say about it. I wish anyone working on it luck, I'll get out of the way :)

spreadsheets calculate an average damage per second by calculating damage dealt over a cycle, a proper application dedicated to this sort of thing could run a simulation of a real scenario and more accurately quantify things like spillover damage, reengage time, and how those effect gear sets

it's something i've wanted to make for a long time, but like many others i just don't currently have the time to sink into it with or without compensation

I used to be a spreadsheet fan who can spend hours playing with spreadsheet to optimize DD sets and theorycraft dps, some of my friend said they imagine Im some sort of no life player that plays with spreadsheet all day. Ever since SE introduced all that augmented gears I stopped optimize DD sets just because it's absolutely horrible experience to import those gears. Not like people give a damn about COR DPS anyways!

So yeah, there you get one example of lazy player that won't try to DD better just because of spreadsheet UI XD.

I'll be happy with a better UI to import augmented gears personally. Right now I just put together gears that seems to work on average level and find excuses to dodge parse challenges whenever another elite COR challenge me a parse XDD.

Whether it is worth it or not is, of course, subjective; but to me, the worth is measured by whether people would use it.

Benefit over spreadsheets, from a user perspective:

• It's centralized. No comparing different versions of spreadsheets or having outdated versions or bad input (which you and I had an issue with), and no more having to keep up with downloading the latest version.

• Viewing a pooled resource of other peoples' DPS profiles means much less effort and time for the average user.

• It would allow for a much more accurate comparison between jobs (assuming the models are accurate, anyway). We will finally be able to visualize how bad MNK really is.

• No need to maintain certain aspects yourself (like equipment).

• Without being limited to the comparison of two sets, we'd be more readily able to see how a difference in buffs/enemy can impact gear choices.

Benefit for those who update/distribute spreadsheets:

• Most probably won't have to anymore, it'd be up to the site admins.

• Any changes/updates to models would only have to be done once.

At the moment, I'd be spending the most time, which would probably otherwise be spent staring at the wall. The time Nightfyre has to commit depends on a number of factors, but the core of his task might very well save him time in the long run, seeing as he is pretty much the only one updating the math in the spreadsheets.

......................................................

To further explain what I have in mind, a user will be able to create a series of 'profiles'

• Character profile: race, merits, job points

• Buff profile: buffs and debuffs

• Job profile: main/sub, and certain variables related to your actions, like WS, WSTP, melee/ranged, etc

• Equip set

• DPS profile: an assigned job profile, along with assigned equip sets for each of the relevant actions

You (as a user) would select the character profile, buff profile, and enemy profile (defaults are provided). Then view all the DPS profiles that have been submitted to the website (which, I intend, can be filtered by main job and combat skill). Here is a brief/incomplete mockup of what the page may look like when viewing someone's DPS profile.

But a downside to the setup I envision is that the results are not instant, like in spreadsheets. So if you just want to compare the difference between one piece, it would ultimately take you longer. So in this respect, it really is more of a community tool. It would depend a lot on use.

Anyway, feedback (to the structure, usability, design, or anything) is welcome.

D: Lots of people like COR dps, myself included. I admire passion wherever it is found, if it's toward a good end. I think COR does very respectable dps especially if you factor in they can support too, I only wanted to show RNG isn't just a version of COR without support abilities and thus basically useless. I also always find it great when anyone does all they can do with a job, such as yourself in min/maxing COR so much and showing what it really can do gives all the rest of us something to compare to and shoot for. I unlocked leaden lately and it was the easiest thing ever, with sam + tact roll and self chaining wildfire on urganites.



I really like the mockup and what you said. My only critique is to try to keep it transparent and real time as you can. Javascript should do everything well (certainly as much as the spreadsheets do at least) would keep it client side and transparent. If doing intense server side simulations, and making it server side and making it "real timey" with ajax calls etc, consider keeping it open source somehow. Sometimes communities will balk at things not kept open source such as the story about that mnk dps calculator in the linked thread, and it keeps contributions flowing and prevents it being a one bus project (as in, if one person is hit by a bus, the project doesn't die idk where these terms come from sometimes >.>). Of course, that's up to whoever works on it, but that's my suggestions, to be taken with a grain of salt I'm quite sure. Good luck on your ambitious project, I have to say it does my heart good to see people willing to still contribute so much to ffxi.

- I dig the layout

- How are you going to do itemsets? You can pull AH's itemsets with:
http://api.ffxiah.com/ffxi/itemset/index.php?method=getItemList&id=<ID>; if that's helpful.

- Replace the lorem ipsum with something from http://slipsum.com/

I would recommend building a simulation framework instead of a DPS calculator. There are many DPS calculator (spreadsheet style) pitfalls that I think would be made worse if users aren't forced to scrutinize the context of their analysis more thoroughly. Like a few weeks ago, I was doing some Sinister Reign with a BLU friend (BLU vs BLU). Our gear is similar, I would rate hers as slightly better, and she also has Tizona. I think she led most of the fights when she was using Sequence/Colada. Tizona main hand was better on the DPS spreadsheet but the one run she used it, she got completely crushed because of how long it took her to build aftermath TP. She literally didn't even get a single weaponskill off before the first wave was dead. I'm not saying you couldn't work those corner cases into a DPS calculator, but I imagine the number of realistic considerations would result in a very kludgy codebase.

Anything involving DPS calculation would really be up to Nightfyre and how he wants to handle it. I haven't had the chance to talk with him yet, but a lot of decisions I need to make about the backend depend on things I may not completely understand about calculating DPS. My goal is allow the user to provide him any and all the data he needs to return an objective value.

That said, bringing up AM is a good point and I may, in that case, add a second TP set (when AM is down) and second WS set (AM proc) to the DPS profile, which could theoretically be used to calculate the entire cycle. This may involve assumptions about play style, though (SAM or /SAM timing meditate, DRG timing jumps, THF...I'd have to think more about).

But I'm not exactly sure what you mean by 'simulation framework.'

As for doing it client side/real time, if I did choose to use Javascript, it would necessitate a completely different structure for that to affect the user experience. At the point that you have all of the components to determine DPS, you are not in a place you would be editing equip sets; not that it couldn't be handled that way, but it also loses some ease in using the same sets for other profiles. I can think of possible solutions to that, like being able to input a stored item set and then modify it, but I have to think about what happens in the database at that point. I will think more about this.

I considering asking about something like this, but it'd probably only be used for importing sets, since 1. item sets on here allow for multiple equip slots (which I like for my wishlists), and 2. main/sub are not specified.

You could I guess add a "avg fight time" entry which would allow you to have an AM3 down set and the Trigger ws set DPS be calculated in based on a ratio of how long the fight is expected to last and how long it would take that TP set to get 3k tp and ws. A 30 second fight(drg and sam aside) means you probably aren't going to have AM3 up but on a 3+ minute fight you have had the time to gain the full advantage of AM3.

Accounting for things like building to 3k TP is already within the scope of the spreadsheets and has been for years, but the burden is on the user to make use of that potential. Moving to simulation would improve model accuracy in certain edge cases and may allow for slightly easier handling of such considerations from a user perspective, but it's not a magic bullet. Many of the decisions we make to eke out those final percents are difficult to model correctly no matter what your approach, so a certain amount of indirect examination will always be part of getting the most out of any XI calculator.

I can't be certain because I wasn't the one using the spreadsheet, but I believe that was the way the spreadsheet was configured (to take into account 3k TP). Regardless, the point I'm making is that the DPS calculation does not consider mob HP or group dynamics very gracefully. The mob was dead before even reaching 3000 TP. There are time gaps between mobs that eat into the aftermath duration, and I think we even finished the fight long before AM3 wore off anyways. I imagine only 45~60 seconds of the aftermath was actually utilized.

I think generally the DPS calculators are reasonably accurate for determining gear selection (other than RMEA) but fairly poor for predicting actual DPS in a particular situation. If that's what you're concerned about (and it often is in these BLU vs DRK or whatever comparisons) than I think 'edge case' becomes 'nearly every situation of interest'

For my uses (gear selection) the spreadsheets work fine. Perhaps I'm wrong, but what I think Blazed is more interested in is 'what would win in this situation.' If that's the case the I don't see spreadsheets as ever being accurate, at least not in a maintainable way.

Simulation framework vs DPS calculator -

A DPS calculator takes known equations, mob stats, player gear, etc. and uses statistics to estimate your average performance in a particular situation. Estimating actual DPS is difficult because of thing like skillchains, mob HP, other people in your party affecting what you do, etc. These are difficult to model in a way that's both accurate and easy to configure for a specific situation.

A simulation framework takes known equations, mob stats, player gear, etc. and simulates a series of actual fights and reports the results. It's kind of like parsing but instead of playing the game we're using a framework to simulate it. There's nothing a simulation framework can do that a DPS calculator can't do. I think that if we want accuracy for specific situations that a simulation framework would be more maintainable than a DPS calculator.

If the spreadsheet was configured for AM3 build time and a realistic depiction of AM3 uptime, Tizona would have been behind from the get-go. It's not that far ahead even in ideal scenarios, so I'm inclined to think that if there actually was an attempt to specifically model an SR-like scenario (something I've also done in the past), it was done poorly.

I agree that the spreadsheets don't handle current party mechanics very well - updating them to better account for current skillchains and the resulting strategies was/is a major goal for my work on the spreadsheets. They can do a much, much better job of handling such things than they do now.

How do you propose coming up with a mean uptime for AM3 for any specific situation?

You generally don't need to, or at least that's not how I would approach the problem. You can solve for minimum uptime required for Tizona to pull even/ahead (accounting for varying situations such as defensive swaps, changes in behavior/active time, etc as needed) and then compare the result to established/expected fight duration.

solving for a specific variable using a spreadsheet goes pretty far past what the average user is going to be willing/able to do

I think the ability to upload lua files to prepopulate gear sets would be nice. System would have to be designed to easily add augments though and ideally compare augments.

Also an ability to have to have it search and recommend gear could be nice, I had wrote an add on for Mote's sheets a while back (its pretty simple, but still useful and could be improved upon):

http://www.ffxiah.com/forum/topic/39693/damage-spreadsheet-tip#2419243

I'd be willing to help depending on language decisions etc.. made. My time is fairly limited though.

Implementing SC dmg is a good idea tbh. I often feel SC dmg is often left out of most DPS discussion and often overlooked by the community.

If it's something that allows user to set individual ws(with appropriate sets)for each step and calculate and compare the final dmg would be nice. Although it seems like too much work.

My plan for skillchains on the sheets is to accommodate TP return from different weaponskills (ie set 2 can use the TP return from WS 1 and vice versa) and allow the user to set the skillchain type (level and step number), as well as some changes to melee-oriented calculations (an option for minimum time to WS instead of minimum TP to WS). There are a few additional considerations (enemy MDT and SDT for instance) that will probably be user configured, and eventually I'll need to add support for Inundation.

I've been working on a way to easily compare gear set stat totals (mainly because I hate manually inputting gear into spreadsheets).
Anyway, as of now I have it working to compare cumulative augment stats across sets, but was wondering if there's a way to retrieve gear's base stats from FFXIAH, or if they need to be put into a spreadsheet, etc.

Easiest way I've found to pull gear stats is to parse the JP descriptions and sort everything into tables. Would still require manual definitions for stuff like Brutal Earring, but that's easy enough.

Sweet, thanks.

Hmm... interesting but you have to establish boundaries and limits so you can expect something and not having a bunch of feature creep.

What we currently known is that the formula for DPS is fairly constant. As long as you keep your assumptions right then you put in your gear information then you will get the DPS. That means the key to this app is how to update information and how easy and intuitive it is to use compare to the spreadsheets.

Ask yourself a few questions and see what sort of shape you will have your app in.

Why do you want to use this app over the spreadsheet?

+ There are 3 reasons I can see why someone would want this over the spreadsheets. #1 Auto-update on new equipment, #2 streamline UI (e.g. you put on your gear in game and the app will tell you your DPS/WS vs an fixed mob right inside the game), #3 ease of use compare to the over complicated spreadsheets.

You can come up with 3 options:

#1 update the current spreadsheets to make it simpler and possibly more universal and add that to a wiki or somewhere people can access and edit/update as they use it. Cheap option.

#2 Develop an external app if you can find a database or pull from the game. You will want this app to be open-source to let people tweak it and possibly fix/update the thing. You will need the option for people to enter their augmented gear and probably the ability to remember people's set. I highly recommend putting the app into a website like ffxiah or bgwiki so people can have an account with gear sets.

#3 Develop a full fledged windower app that will directly pull the gear data from the game and tell people their DPS on the mob people are targeting (probably use an external database to /check the mob). This will update on pieces of gear that people augmented or new SE gear. You must also add the ability to set a hypothetical set of gear so people can aim for (and compare to). You might want a way for people to easily adjust their 'dream' gear.

Personally, I really prefer the combination of #2 and #3 option. Instead of fiddling through the spreadsheets, I can just put on what I wear and compare that to what I want - right inside the game. However, if I have to pick one, I would vastly prefer #2 simply because it is a lot more useful to know what you should aim for.