I've been a big DnD fan since I was first exposed to it a long time ago (2nd Edition baby! ;D ) and about a year or so I've been perusing the optimization threads and loving every tidbit of info on this forums.
So I was browsing around the forums and playing with my character generator and it suddenly dawned on me that of all the character generators I've encounter, be it web based or download and install flavored: they don't optimize. I mean they can generate a decent character, but they can't stand toe to toe with characters that are generated in these threads. Simple example, I doubt any character generator can create Pun Pun (*bows reverently*).
Now, I'm a programmer by trade and I thought I'd try to create a character generator that would follow some set parameters to generate optimized characters.
So why am I posting about this? Well because I wanted to see if there was a demand or interest present for such a program. I've been working on the preliminaries (choosing a language and platform to work on, what sort of database I'm going to need, interface issues, etc etc). This is going to be a daunting task and I was hoping that the community here would be my support structure in motivating me to keep at it.
So, what say you character optimizers? Should I dive into this project or shelf it?
By the way, I am very open to suggestions and ideas at this point. I was to make this generator as efficient as possible (obviously since the sheer amount of permutations and combinations could just kill the project if it can't generate a character it an acceptable amount of time), as accessible as possible (crossplatform? web based?), and modifiable (new book comes out, just add the data to the database).
Thanks for reading my long post!
I'd love to at least talk it over with you, but it's going to be a fairly odd endeavor.
My impulse is python as it is platform agnostic for the most part, and I've been thinking about the permutation problem and I think if we use an advice driven system that won't be too much of a problem. What I mean is that a lot of problems have known-best solutions, and we can indicate these automatically if we lay out the permutation engine correctly.
Character Concept:
he duel weilds bastard swords and shoots fireballs!
Level of Cheese:
lots of stinky chese my dm is a jerk and I need to teech him a lesson
Other Options:
I roled all 18's and a 17. I get 5 LA for free and flaws.
Not only am I interested, I'm a fellow hacker and was thinking about this just last week quite extensively.
I'd love to at least talk it over with you, but it's going to be a fairly odd endeavor.
My impulse is python as it is platform agnostic for the most part, and I've been thinking about the permutation problem and I think if we use an advice driven system that won't be too much of a problem. What I mean is that a lot of problems have known-best solutions, and we can indicate these automatically if we lay out the permutation engine correctly.
Indeed, I figured it has to be driven by a system to allow for inclusion/exclusion of entire books, or just certain elements (like just certain feats aren't allowed, or no swashbucklers, etc). A thought that's just cropping up in my head is that any decent MySQL or similar DB can handle the data by cross listing the descriptors for an entry, so all that would be necessary for the user to manipulate what is "in" and what is "out" would either be a straight selection menu/section of all the options (which would be insane after inputting ALL of spells, feats, skills, etc) published by wizards...or, just have an interface that allows the user to create customs DB's (i.e. No Eberron DB, or No Dwarf Race DB) for their character generation without screwing around with the Master DB, which contains, of course, all of the data.
As for python, that's in the pool of languages I'm considering. I'd have to refresh my knowledge of it, but that's just a matter of reading through the books I have.
Heck, if I there's enough people interested, and willing to put some work into it, I think a team of us could get this going in no time reducing the coding time for it a great deal.
Obviously, I'd say the first prototype build would only involve the PH and DMG as test data. Start small (rhetorically speaking cause just those two manuals have a LOT of info if you really think about it. I'm actually planning on getting my hands dirty in code after June 8th on this little project cause I have some scheduling issues to deal with at the moment.
This might be a starting place for you haxxors?
So, what say you character optimizers? Should I dive into this project or shelf it?
Shelf it.
This once, I demand a reason from you.
Why?
This once, I demand a reason from you.
Quite simply I think it'd take the fun (and soul) out of D&D and character creation. How interestering would it be to simply press a button and have an instantly CO'd character? Moreover I think the time that'd you'd have to invest in a program like that would not yield enough popularity for it to be worth (I could be wrong tho) So I think if you're doing it to please the masses I wouldn't do it, of course if you're doing it for yourself than by all means.
Heck maybe I'm old school, I still use pencils and papers and actual books, all the other players in my campaign use laptops, PDFs, and electronic character sheets. But to each their own.
I cannot really see this feasible. What I could imagine is a searchable indexed database of existing CO builds. E.g., gimme all fighter 20 builds. Or, all charger builds
It could probably be based partially on just function. Like say I want a grappler. Then first I'd get a race chart of what might be useful, then a chart of what prcs/pcs offer basic bonuses to grappler and then what prc's/pcs have abilities/variants that interact with grappling in a special way.
I doubt any character generator can create Pun Pun (*bows reverently*)
The biggest problem you will run into is characters like Pun-Pun use heavily debated topics which not everyone agrees is valid via RAW. Since RAW isn't a solid object (every person has a different opinion what the rule means) you can't make a generator that suits all people.
Making a generator that is pretty solid on RAW (99% agree or something) is still possible, and if done well many people will find it of interest (such as me.)
Currently, I use a hand created excel sheet that knows how to build any character using the races/classes I tend to build. I fill in each level in a column and it fills in the stats/abilities. I fill in spent skill points in another area and it fills in how many I have left, the totals of the skills including synergies, etc.
Just going that far (a generator that YOU build the character inside) is useful.
Now if I misunderstand you, if you mean to make a random character generator. One that can pump out optimized random characters, I believe you will find that project significantly more difficult than you can imagine.
I imagine something like categorizing several basic ideas. Start with the set in stone things (e.g. I want the swift hunter feat and the monk class and an unarmed combat idea) before it correlates compatible entries on what you haven't specified (perhaps you should be a halfling and use the skirmish-granting racial sub!)
If you have the resources (Rocktapus and I can help) it sounds wonderful. But you should realize how deep you would have to go for this to be useful.
Before I respond to the posts, sorry for disappearing, from the thread, been a bit busy, but I'm going to be checking this thread regularly again.
To respond to a few posts in no particular order...
I use PCGen right now and I like it, the problem I have is that a lot of source materials are missing, at least from the Mac version (I mean there's no swashbuckler for pete's sake). Also it's a bit slow because it uses Java. I want to find a way to make it work on Windows, Mac and Linux without sacrificing performance. I've seen a few web based generators that are interesting but not sure if my idea can work in a web based architecture.
I appreciate the input but I don't think it would take the fun out of it cause it will never replace a person optimizing characters. I'm not planning on shelfing it. Don't misunderstand tho, I started playing DnD beck in the 2nd edition days when I was a kid. I appreciate the old schoolers as I'm a bit of an old schooler myself.
I am very aware at the level of complexities involved with a generator of this kind. That's why I want to see if I have the support of the DnD community behind me cause I know I'm going to get frustrated during the process of making this thing and be tempted to trash it. In fact, when it comes to rule checking, I'd prefer to keep it as strict as possible and as math based as possible. 99% of what goes on in creating a character boils down to playing with numbers.
I want to give it a shot. I still have a lot of prep work to do in choosing a language, working out how to organize the info in a useful manner for the code and for the user.
Thanks again for your support and encouragement. I still haven't decided if I'll start on this project or not, but it's looking pretty good so far :D
I used to use Ema's Character sheets, until they got the C&D letter from WotC. A little sad that resource is gone, especially since I found the spell lists to be a huge assistance. I am zealously guarding the remaining hard copies I have left, since I neglected to make a backup of the PDFs. :/
Additionally, a lot of the CO work is not of a very straightforward nature. You can easily pick out everything that boosts a grapple check, but would you also have all the info on how to boost STR? If you want to abuse a particular spell, would the program be able to recognize exactly what you are attempting to do? There is a large amount of non-numerically quantifiable information that would also have to be entered.
Heck, just spell selections on a caster is bitterly argued, with a large amount being due to play style and campaign/DM setting.
That said, good luck. Would like to see a decent resource but don't fall afoul of any legal isues.
On the plus side, at least 3rd edition material is complete now, you are able to get the diffinitively complete official WOTC material into it if your willing to do the work, though for obvious reasons I can see Dragon and Dungeon magazines or module material being left out.
I agree that it could never quite replace an individual optimizing a char but could get close. Cause ultimately, it will probably end up like PCGen in the sense that you can go in a play with the character data.
I do like a non-optimize option in there. So it just generates a good char without trying to take advantage of every single option out there to make it uber.
The most I can see is something that suggests existing builds/conceptual feats/classes/items/spells, etc.
If you include prestige classes (something I haven't seen a char gen do elegantly yet) make sure it boldly lists their prerequisites, but doesn't require you to have them to put the class in your sheet, in the event that a DM has houseruled it or their getting in by some optimization trick the generator can't recognize.
About picking feats and such, PCGen uses that methodology, where you build it level by level, but it can also pick feats by itself. It also has some prestige classes, and doesn't allow you to add the prestige unless you meet the requirements AND on top of that highlights non accessible stuff (like a prestige class you don't qualify for) in red.
As you can tell I've used PCGen a lot and do like many aspects of it.
The first stage whether it will generate optimized chars or just a normal build, would be to have a database, and allow for choosing what books are and are not included for the build. Ideally, if you pick only PHB, and you want the strongest fighter you can get using only the PHB, it should be able to do it no problem.
Yeah, the point I was suggesting was that stage one would be a simple char gen with all the data available (leave out any necessary details to avoid trouble lol) but to have everything at your finger tips where you could say... for example... tell it to let you choose feats from... Shining South, then you would get a list with a brief overview of what they do, and could then choose feats and drag them into the feat slot you want them to fill.
If you include prestige classes (something I haven't seen a char gen do elegantly yet) make sure it boldly lists their prerequisites, but doesn't require you to have them to put the class in your sheet, in the event that a DM has houseruled it or their getting in by some optimization trick the generator can't recognize.
Just give the program an option to not enforce prerequisite requirements via a toggle/checkbox/whatever. I would prefer that the requirement enforcement be left in (for most cases).
just try to avoid or slay to copyright
sPersonally, I'd say that rather then having an automatic character generator based off of some preset parameters, what would be more interesting is something more along the lines of the following:
User opens program.
Program loads, and asks user for character type (grappler, tripper, archer, Battlefield Control, Healbot, Charge build, etc.)
Program starts listing various widely recognized race/feat/spell/weapon/etc. that fit for that category of character, ranking them by "efficiency" (i.e. how often they come up in realistic optimized builds; Power Attack being as one of the first feat entries for a charger, for example).
User proceeds to use said lists to create their optimized character, without the hassle of having to search through every book in their possession to find the best option.
Kind of like an offline indexer, you know? Only, rather then being listed alphabetically, they're being listed by usefulness.
I think that Mister_Crowley has a point about it taking the heart and soul out of D&D...
Personally, I'd say that rather then having an automatic character generator based off of some preset parameters, what would be more interesting is something more along the lines of the following:
User opens program.
Program loads, and asks user for character type (grappler, tripper, archer, Battlefield Control, Healbot, Charge build, etc.)
Program starts listing various widely recognized race/feat/spell/weapon/etc. that fit for that category of character, ranking them by "efficiency" (i.e. how often they come up in realistic optimized builds; Power Attack being as one of the first feat entries for a charger, for example).
User proceeds to use said lists to create their optimized character, without the hassle of having to search through every book in their possession to find the best option.
Kind of like an offline indexer, you know? Only, rather then being listed alphabetically, they're being listed by usefulness.
Definitely something to consider. I still don't think it will replace a player optimizing a character, so I don't think it'll take away the magic so to speak.
Depending on the coding, it could end up following specific paths of generation which in effect is just using the feats and optimization that people mostly use. So ultimately it may not be "creative" like a person. The problem I see mostly right now is making it generate quickly and have rule checking, and also just inputting the metric A$$ tons of data from SRD and rule books and even dragon magazines if I can get a hold of those things.
That's why I would LOVE for Wizards to want to be part of this. Would make getting the data a LOT easier and would be great to just have a team of people compile all the data so it can be inputted into the DataBase easily.
I'd also want the optimization to be an option that a user can turn off if they just want to generate a random NPC or a regular PC without having to use an optimized char.
I've had to put the idea on hold for the moment. Been very very very busy with my last couple of classes before getting my degree so I haven't done any research, data gathering, planning, anything really, when it comes to this little project.
I'm putting it on hold indefinitely. Fear not, I don't plan on abandoning the project. I'm predicting I'll be able to get to it in Sept again.
I'll still be checking the thread for any other great suggestions and words of support.
i do python, and i'm learning lisp.
i think the first step would be actually defining an algorithm that can design optimized characters based on input. thats going to be hard enough.
I am ready to jointm
i do python, and i'm learning lisp.
i think the first step would be actually defining an algorithm that can design optimized characters based on input. thats going to be hard enough.
Yea there are a few first hurdles to jump over, not just designing an algorithm, but also data organization, how to present this in a meaningful manner to the user so when he chooses to optimize a fighter, it optimizes it consistently every time (technically speaking, if someone chooses the same options over and over the results should be the same), and efficiency since there is a LOT of data to go through, and I can't have the program taking weeks going through all possible combinations to get the "best fit".
My main concern at this point is portability, efficiency and organizing the data. I want it to be very easy to add custom classes, armor, anything. Also I want to be able to turn off optimization if someone wants to just generate characters without making it the strongest damned warlock ever lol.
Between now and Sept. I have a lot of hell to go through for school, personal life, work, etc, so as I said, I'll keep checking back here for support and ideas.
I might have to break out my Python book in the interim and update myself if that's the language I ultimately choose. I still haven't decided which language though...
alternatively, we could use an evil scheme of mine that includes reading the data from pdf files that we search the computer for.
except thus files are illegal. damn.
ill have a whack at an algorithm sketch here in my upcoming vacation(ooh, tomorrow's the day. yay!)
"and also just inputting the metric A$$ tons of data from SRD and rule books and even dragon magazines if I can get a hold of those things."
As for this...I know nothing about coding, but it seems to me that the above stuff is grunt work...sitting there with the book and typing the text into an appropriate medium (program, whatever). I'm sure you could find a small army of people willing to do small chunks independantly, who would then fire them off to you.
And a nod towards the naysayers: I agree that, in principle, D&D belongs on paper and every step away from that risks sacrificing the nature of the game, something that is clearly very special. D&D is the rare example, for me at least, where more complicated and less technologically streamlined equals better. That said, I would love a one-stop encyclopedia of everything D&D, so I don't need to use 28 different books when I want to CO. But I would (probably) not use an automatic CO function, except maybe as a way to sift through feats to get a general census of existing combinations, much like I do when perusing the forums. I'm not building a fighter20 tripper build, but I like to know how it works (respect@Snow Savant) and which elements/feat/class combinations are involved.
Just a suggestion, but count this as my supporting vote :D
I don't know how often I would use the optimising function in such a program, however i would absolutely LOVE to have a program with a comprehensive listing of all 3.5 race/class/prc/feat/item/skill information, with a clean interface that could format to a final character sheet.
"and also just inputting the metric A$$ tons of data from SRD and rule books and even dragon magazines if I can get a hold of those things."
As for this...I know nothing about coding, but it seems to me that the above stuff is grunt work...sitting there with the book and typing the text into an appropriate medium (program, whatever). I'm sure you could find a small army of people willing to do small chunks independantly, who would then fire them off to you.
And a nod towards the naysayers: I agree that, in principle, D&D belongs on paper and every step away from that risks sacrificing the nature of the game, something that is clearly very special. D&D is the rare example, for me at least, where more complicated and less technologically streamlined equals better. That said, I would love a one-stop encyclopedia of everything D&D, so I don't need to use 28 different books when I want to CO. But I would (probably) not use an automatic CO function, except maybe as a way to sift through feats to get a general census of existing combinations, much like I do when perusing the forums. I'm not building a fighter20 tripper build, but I like to know how it works (respect@Snow Savant) and which elements/feat/class combinations are involved.
Just a suggestion, but count this as my supporting vote :D
Yes an army of minio...er I mean supporters :D would be great for data input as long as they all format it the same way (of course those details would be worked out).
The legal aspect mentioned above is going to be a nuisance for sure, but as I've mentioned, I'd love to work with Wizards on this to make it legit. In fact I invite them to contact me and we can talk.
I thought about some aspects of what you mentioned sea_cucumber. I too have always wanted a complete database of 3.5 data (maybe eventually 4.0, but I'm not a fan of 4.0 so I don't plan on implementing that...at least not just yet). I would like for the app to not only generate chars ad generate optimized chars, but also be a repository of everything dnd 3.5.
The beauty of this is idea is, at least to me, is that this isn't all that complex to code really. The hard part is making it efficient and fast.
I agree that there's something beautiful about dnd character creation as a process. In fact I'm the first to appreciate that cause I've never been big on generators anyways. However, I know for a fact that any character I make will never be the best it can be because I just don't know all the rules and numbers in my head to ensure I get it right, and I'm bound to weaken her somehow.
we could stea-i mean use the data from some other sheet generator, like PCgen.
alternatively, we could use an evil scheme of mine that includes reading the data from pdf files that we search the computer for.
except thus files are illegal. damn.
ill have a whack at an algorithm sketch here in my upcoming vacation(ooh, tomorrow's the day. yay!)
BTW since our datasets are covered under the OGL (with the exception of some datasets having a special license to use deity names and such). It wouldn't be stealing.
Of course you would have to add the PCGen Sec. 15 line for every dataset that you used though.
BTW since our datasets are covered under the OGL (with the exception of some datasets having a special license to use deity names and such). It wouldn't be stealing.
Of course you would have to add the PCGen Sec. 15 line for every dataset that you used though.
YEa I need to look through those datasets to see if they are indeed complete and are formatted in a useful way to the way I want to use them. We'll see.
In response to arguments about taking the creativty out of Character Optimization, you're not forced to use the program. If there are people who agree with such an opinion they can just not use it.
Saying that a program takes the creativity out of CO is like saying Treantmonk's Guide to Wizards took the creativity out of wizard creation. The answer is it simply didn't. It's just an extremely helpful tool as this program would be.
I too have always wanted a complete database of 3.5 data (maybe eventually 4.0, but I'm not a fan of 4.0 so I don't plan on implementing that...at least not just yet). I would like for the app to not only generate chars ad generate optimized chars, but also be a repository of everything dnd 3.5.
...
I agree that there's something beautiful about dnd character creation as a process.
Precisely. A one-stop application acting as a 3.5 compendium (races, classes, feats, skills, prcs, spells, rules, monsters, challenges, rewards etc.) can only help make all of 3.5 more accessible. If there is one thing 3.5 suffers from, it's bloated and mismatched publication. An encyclopedic application (maybe with other fun features :D) would go a long way towards cleaning it up and bringing it all together.
So...keep on trucking!!!
Saying that a program takes the creativity out of CO is like saying Treantmonk's Guide to Wizards took the creativity out of wizard creation. The answer is it simply didn't. It's just an extremely helpful tool as this program would be.
+1
Much what I was trying to say, but much more concisely.
first, you define your concept, by ordering modules in a layered tree structure
Hey friends and neighbors!
As mentioned in previous posts, I was going to get back on the saddle of working on this character optimizer/generator (Tentatively called COG at this point in time) in September and by golly it's Sept. already isn't it? :D
So I'm going to start on the preliminaries:
1. Language to use: preferably something modular/object oriented, highly portable so it works on windows, macs and linux machines, and perferable alanguage that I already know...I hate to have to learn a new language, but I'm willing to do so if necessary.
2. A comprensive list of 3.5 titles would be useful so I know what titles I'm missing
3. I need to start making some flow charts and design a basic algorithm for character creation and basic optimization techniques employed by users.
I'm throwing this infor out there so if anyone wants to contribute they can. Thanks again everyone for your support!