|
| Thread: Heroes ranking |  This thread is pages long: 1 2 3 4 · «PREV |
|
Paladin_H3
Hired Hero
|
posted September 02, 2009 08:33 AM |
|
|
Theoritacly yes there is an argument. But as in practical terms how often will you have battles with that spell and also are you willing to give up that opportunity for an alternative spell cast. Plus how many treasure having fire magic as a skill, let alone exp or adv at it.
Counter artifacts and counter battle tactics may sometimes be a huge difference depending on the level of your opponent, thus practically I will have to dismiss that idea.
Specially its not worth considering that when choosing a barb just because the probability of opponent weilding it is practically low.
|
|
pei
Famous Hero
Fresh Air.
|
|
posted September 02, 2009 05:47 PM |
|
|
In standard H3 it is but in wog there are too many anti-mind spell artifacts. Still if u dont get any u will get scre...d.
Just gonna take this excludes WOG or the Hero specialization boost scrpits.
As i usually play in random, i have no inconvinience with the heroes choice BUT when town was Castle or Rampart or Dungeon or Fortress or Stronghold or Conflux i always managed to stay alive.
1) C.H. !!!!!
2) Tazar !!!!
3) Mephala !!!
4) Dessa or Gunnar or Kyrre "Taxi!"
5) Marius "Nobody understands me but im great...really"
6) Neela "Im a human wall"
7) Gem "Goodbye my Coney Island"
8) Vidomina or Isra "So the map is S?"
9) Luna or Brissa "We are pizza in movie night".
10) Rashka "Im bad and can tell u stories".
11)Voy (had to give this fellow its place in my heart)
|
|
angelito
Honorable
Undefeatable Hero
proud father of a princess
|
|
posted September 02, 2009 06:39 PM |
|
|
Gem?????
Rashka????????
Voy?????????????
I'm lost...
____________
Better judged by 12 than carried by 6.
|
|
pei
Famous Hero
Fresh Air.
|
|
posted September 02, 2009 09:48 PM |
|
|
Ok, let me explain...Voy is just getting a star for it has saved me in maps where there was no wood at sight. He really did conquer the sea. The same thing happened to me when a random hero was (PUAJ)Gem. Yes...but she did very well on islands kind of map, not as the main hero but a hero at least.
I have to say that if i didnt take these considerations then my list would be EXACTLY like the rest, i mean change one or two heroes only or move precedences in the list order.
Rashka is not a bad option, i know that Marius could turn to be better as main hero but Rashka works fine for me doing the nasty job.
If i have to reform my list based on your point of view as the main hero then it would turn out to be:
1) C.H. !!!!!
2) Tazar !!!!
3) Mephala !!!
4) Dessa or Gunnar or Kyrre "Taxi!"
5) Marius "Nobody understands me but im great...really"
6) Neela "Im a human wall"
7) Thant "My minions!"
8) Don Mullich "How u doing?" 
9) Vidomina or Isra "So the map is S?"
10) Valeska "Shhh"
@dimis: Your analysis is great. Could u post or upload more of that stuff? Seen percentages but didnt have such a mathematical approach.
|
|
Casihasi
Known Hero
|
|
posted September 03, 2009 05:58 PM |
|
|
I donīt like Neela much...alchemists are not a good class, and who would want scholar on his main.
Give me a barbarian or Overlord with 2 useful skills like off. and tactics any day over her.
Iīd have to say Gunnar is the best hero overall, Hack and armorers are ace too but it doesnīt help much if Gunnar beats you to those utopias.
Also if you play with no rules, facing a hit and running Gunnar is a major pain in the A$$.
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted September 04, 2009 07:48 PM |
|
|
skills
pei,
You can find more information about offered skills in the thread On the internals of offered skills when leveling-up a hero and references found there.
However, throughout the years, that thread has grown to be chaotic. So my actual suggestion would be to go through this one (appeared in The Librarys Heroes 3 Tribute Thread) which tries to gather all the info in a few pages with some examples and has almost all references for up to about a year ago (which is already a very good level). If you have questions about these or questions that are not answered in the summary I propose above, feel free to post in the thread so that everyone can benefit; there are no dumb questions.
This also reminds me that I have to work on some problems that I have left unanswered and I was planning to work on during the summer. hmmmm ... I guess you will hear from me soon.
____________
The empty set
|
|
pei
Famous Hero
Fresh Air.
|
|
posted September 07, 2009 02:28 AM |
|
|
Dimis, i found it extremely interesting.
I must say that i agree with the document, something is missing as sometimes the composition of skills repeats itself exactly from game to game.
I think this could be managed with penalizations when user decides not to choose one skill, then its weight could decrease.
Have you ever seen how the Simplex Algorithm works? Now, i would like to see something like that in games leveling up system.
Hope to read more of your work, Saludos.
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
|
posted September 07, 2009 06:35 AM |
|
Edited by dimis at 07:19, 07 Sep 2009.
|
more references ?
As I said earlier, this is consistent with the work that has been done for up to some time ago (may be 9 months or so; I have to check). It is for sure ok complete up to page 9 of the thread, since I refer another guy (Kallen) to the same document. If you want more details, you really have to go through the thread, or post a question there.
So, in particular, the observation of repeating sequences of skills is kind of old.
You will find this particular one kind of interesting (to my opinion) along these lines (appeared in page 4 of the thread).
However, in the meantime, there has been a dramatic improvement, but again you have to read the thread after that post by Kallen. In the document of the link in the previous post, page 12, section 5, I mention that I see two approaches. One is kind of data-mining and what I would indefinitely follow, and the other one is by direct reverse engineering the code and figuring out precisely the skill advancing problem. Well, another guy showed up at some point - AlexSpl; he liked that other approach, and solved the problem. He gives the description of the algorithm in page 12 of the internals thread. You might also want to have a look on his tool; LMOracle.
So, we are missing the re-implementation of the skill advancing algorithm, and of course the burden now comes to me because of all the source code that I have ready already, so that we can re-evaluate certain policies. This is what I have to do soon.
And then there are other suggestions from the past and we should see the performance of policies (strategies on picking skills as they are offered), e.g. Frank's skill selection, maretti's group of skills, plus one or two mine which focus on the exploitation of the so-called "magic" exception mechanism (you can have an idea by the question maretti asked in page 10 of the internals thread and gave the recent "boost" to the whole problem).
That's all for now. I should better write down some code and we can actually give an end gradually to all these. There are also other problems that are going to appear, but these are really minor in terms of the importance of the questions that we ask and we are able to answer.
____________
The empty set
|
| |
|
|
