|
|
markkur
Honorable
Legendary Hero
Once upon a time
|
posted October 22, 2011 10:37 PM |
|
|
As rpgyay said Hotseat works. I play a lot myself.
Any chance you could try the game on a Lan? I've yet to hear if anyone has yet.I think it should work the same, just looking for verification.
Anyway, really just wanted to give you a...Welcome.
|
|
LarkinVB
Known Hero
|
|
posted October 22, 2011 11:08 PM |
|
|
Quote:
Anyway, really just wanted to give you a...Welcome.
Oh, thank you. I stopped playing H5 about 2 years ago. I already tested H6 demo a bit and it seems nice but buggy and has a weak AI.
Since H5 is inexpensive these days I look forward checking this mod. Shall I play as is or are there other mods which are recommended in combination ?
I can't test LAN, I just own a single computer.
Cheers and thanks again for the welcome and your efforts.
|
|
markkur
Honorable
Legendary Hero
Once upon a time
|
|
posted October 22, 2011 11:39 PM |
|
|
Quote:
Shall I play as is or are there other mods which are recommended in combination
You may not know that vamilla H5 was not ever truly completed, in that bugs still persist that were never addressed. Mods that come along can have conflicts with this state of things. "Q"s efforts had to go back to ground zero because of this and even now things are yet uncovered. It is best if you have only the latest A.I. (J) Also, keep the maps in the "Maps" folder under TotE 3.1 empty of all but the map you play each time.
What I found out the hard way and many others have also, is the the game and Editor bring all mods and changes to bear at once.
Whenever I make the change to the "latest" I also remove old saves from earlier versions in the "profile folder under mygames". This just makes sure there is no chance for conflicts. That way any reporting I do I am assured that what I will share will be accurate.
A lengthy response but I know we are getting many new players and thought I'd go into more depth.
Quote:
Cheers and thanks again for the welcome and your efforts.
Quantomas, aka "Q" is the Wizard behind this work and deserves the thanks. I'll pass it along to him the next time he comes out of his arcane lab...or, he may poof in at any minute
|
|
fulano
Known Hero
Can I link to my own avatar?
|
|
posted October 23, 2011 12:10 AM |
|
Edited by fulano at 05:03, 25 Oct 2011.
|
Will it work in LAN? I assumed the coding just hadn't been done but I can try a LAN game if you need LAN testing.
I haven't had a problem with hotseat so far but I noticed it seems to force you to use seven skills, not eight or six like single player gives you the option to do. Is that how it's supposed to be?
EDIT:
Also, this focus on a good AI is great. I've seen great ideas for game projects but many have died, partially because they never did much to make a good AI.
EDIT2:
It seems the seventh skill was a bug that happened when a Barbarian visited a witch hut with six skills already, but again, with the beta I can't reproduce the bug.
|
|
rpgyay
Known Hero
|
|
posted October 23, 2011 01:46 AM |
|
|
|
I did some testing with LAN and besides not being able to choose the advanced AI options (scrambled interface) it did work though I experienced a bit of a lag when trying to move my hero or do anything in my town, I uploaded my my map, save and logs in another post and I believe Quantomas reviewed it.
|
|
Jantz
Tavern Dweller
|
posted October 23, 2011 01:53 AM |
|
Edited by Jantz at 02:04, 23 Oct 2011.
|
Speaking of the thanks. I follow this fantastic project almost from the very beginning, and I'd like to thank Quantomas for the hard and near work he's done and that he keeps on doing now! Your time and effort is greatly appreciated. U simply revived my most favorite game ever! All u do about the AI is just magic! Many thanks!
Well, concerning the fact that your AI used to cheat before version 3.1.15. To be perfectly honest I was aware of it too. And to put it like that, it cost me all I knew not put in a word in support of Selvaria when everyone maintained there were no cheats. Because I think even the AI we have now (though, beyond all doubt it's 300% stronger than it used to be in the vanilla!),imho is still unable to give enough of a challenge for an experienced human player in most cases. So, this brand new option with a slider to set an AI rush discount is a really great decision! Also pleased we've managed to persuade you to get the AI visit treasure vaults, always thought it was a must. 
Markkur, your new map "Power of Dragons" is also quite a piece of art! I'm sure creating this took a very lo-o-o-ong time. Playing it now..
|
|
LarkinVB
Known Hero
|
|
posted October 23, 2011 10:57 AM |
|
|
Quote:
Quantomas, aka "Q" is the Wizard behind this work and deserves the thanks. I'll pass it along to him the next time he comes out of his arcane lab...or, he may poof in at any minute
Of course my thanks are also and foremost to him.
Is there some background info on Mr 'Q' ? I found that he was posting in a german forum and his location was displayed as Hamburg, which is a city only 100km from where I live.
If I understand correctly he was also the coder of the vanilla HoMM 5 AI and has access to the AI source code ?
He is doing all the coding and design on his own or is this a community effort ?
|
|
Quantomas
Responsible
Famous Hero
AI Wizard
|
|
posted October 23, 2011 06:18 PM |
|
|
Quote:
Because I think even the AI we have now (though, beyond all doubt it's 300% stronger than it used to be in the vanilla!),imho is still unable to give enough of a challenge for an experienced human player in most cases. So, this brand new option with a slider to set an AI rush discount is a really great decision!
One day people will believe, maybe only after the AI has chased and defeated them in their last castle, but they will. 
The main reason that a human veteran player has the upper hand against the present AI is that a veteran can typically outperform the AI in combat by a factor 1.5 - 2. Additionally the human player has still the advantage to grasp the overall picture more easily and devise a strategy with the planned development in mind.
There are three important tasks to rectify this:
Combat AI: full overhaul, that we get something top-notch with the mathematical precision of Heroes III.
Middlegame: full six week lookahead for the AI to develop actively strategies and coordinate its forces.
Opening: further optimization of the building phase.
Once these tasks are done (3.2) there will be no need for cheats anymore. Once the human player will no longer be able to outperform the AI in combat, and the AI goes for the most important strategic targets while coordinating its forces to keep its holdings, plus a good optimization of the building phase, the AI will be a match for the veteran strategists. Of course this also requires fine tuning of its artifact use and its skill selection to be adaptive.
But make no mistake, these tasks are big and require lots of work. I have already the design for the middlegame and are implementing it, the other two tasks are comparatively more easy but momentous as well.
Quote:
I found that he was posting in a german forum and his location was displayed as Hamburg, which is a city only 100km from where I live.
That's right. This brings me to another topic. I had to stop posting in the German forum because I cannot afford the time for this any more. By now there are so many small tasks, writing posts, e-mails, webdesign, readme, patch assembly, publicity and so on that I didn't get to write code (except for the 3.1j upgrade, the AI rush discount) for more than two weeks by now. So, I have to get back to the lab proper, after all it's most likely the world's most advanced AI project, and not something you can do in-between.
Still there are more administrative tasks.
Do we have someone who could help with setting up our own studio in Ireland? Someone who could help with the legal and administrative tasks?
Yes, you heard this right we plan to move forward. We hopefully also will move soon to our own community board at www.eternal-essence.com. More info what we plan with Eternal Essence and how the relationship with Heroes V will evolve will follow once we are ready.
Help with the administrative and management tasks is more than welcome, because as it is right now this is what keeps me from working on the code.
Regarding work on the vanilla AI, no, this was done by Nival.
That said, I need to get back to the lab. The current tasks are: getting the kamikaze attacks working, tracking down some bugs in the lookahead opposition tracking (the odd behaviour of heroes remaining in the castle or leaving it unattended at some times), the optimization of the build up phase, and of course the strategic AI stage for the middlegame.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
::Back to tons of work in the lab::
|
|
Pavijan
Adventuring Hero
|
|
posted October 24, 2011 12:00 PM |
|
Edited by Pavijan at 12:27, 24 Oct 2011.
|
I have just saw that you added "Allowing for AI difficulty independent resource handicaps -
Initially choice of equal, half or double resource amount of the player for each AI in the .xdb" in Scheduled Tasks on EE website. I don't know if it was there before, but that efficiently removes my suggestions for ANY form of AI discounts in final version. Full support!
One little suggestion: I don't know if you have sen this -
No random ATB patch!
Maybe this small exe change can be implemented in future versions of AI patch and EE?
Mainly because that random ATB bonus provides absolutely NOTHING for enriching the gameplay, just brings some illogical frustrations, and makes strategical planning harder...
P.S. This project seems becoming even more ambitious as the time is flowing... I have feeling that something relly great is waiting for us at the end. 
____________
|
|
SKPRIMUS
Promising
Supreme Hero
The One and the Prime
|
|
posted October 25, 2011 04:17 AM |
|
|
Quote:
...The main reason that a human veteran player has the upper hand against the present AI is that a veteran can typically outperform the AI in combat by a factor 1.5 - 2. Additionally the human player has still the advantage to grasp the overall picture more easily and devise a strategy with the planned development in mind.
There are three important tasks to rectify this:
Combat AI: full overhaul, that we get something top-notch with the mathematical precision of Heroes III...
@Quantomas, I'm not sure exactly how you'll go about this & apologies in advance if you are already thinking about this because some of your words already imply that you may be: In combat, the hero build is kinda like 50% of the game.
If you improve which hero skills/perks the AI picks & then actually have the AI hero cast mass light/dark spells first when they have them before casting single target spells (unless for example the atb shows a great opportunity to cast PM), then it would go a long way towards making the AI more difficult in combat.
Hope you find someone able to help with the legal & administrative tasks!
Quote:
...random ATB bonus provides absolutely NOTHING...
@Pavijan, I'm not sure about this statement, but that's another subject outside the scope of the AI patch/mod really.
____________
Hope defeats despair - "a blatant clue"
too many idiots in VW
"to lose is to win, and he who wins shall lose"
bashing orcus 
|
|
Pavijan
Adventuring Hero
|
|
posted October 25, 2011 10:29 AM |
|
|
Quote:
but that's another subject outside the scope of the AI patch/mod really.
You are probably right, but it's just that it requires hex-editing of .exe file at different place than it was in original 3.1 one and we don't know where to change it in new AI exe... If we could maybe get a version of at lest final 3.2 patch without Random ATB, for those who wan't it, that would really be nice...
____________
|
|
Mort
Tavern Dweller
|
|
posted October 25, 2011 01:43 PM |
|
|
Quote:
@Quantomas, I'm not sure exactly how you'll go about this & apologies in advance if you are already thinking about this because some of your words already imply that you may be: In combat, the hero build is kinda like 50% of the game.
If you improve which hero skills/perks the AI picks & then actually have the AI hero cast mass light/dark spells first when they have them before casting single target spells (unless for example the atb shows a great opportunity to cast PM), then it would go a long way towards making the AI more difficult in combat.
I agree this would help the AI a lot. Another thing I think the AI needs to improve is that it should have a high priority on casting dispel/magic immunity on creatures that have been frenzied or puppet mastered. I feel I can get away with using these spells a lot without worrying about the AI countering, in a recent battle a Zealot decided to shoot at my creatures doing pathetic damage rather than cast cleansing on the frenzied stack of angels next to it. If the AI could use predict even as far as what the next few creatures might do it could avoid things like this.
|
|
SKPRIMUS
Promising
Supreme Hero
The One and the Prime
|
|
posted October 25, 2011 03:58 PM |
|
|
cleansing doesn't work on frenzy but I understand what you mean.
____________
Hope defeats despair - "a blatant clue"
too many idiots in VW
"to lose is to win, and he who wins shall lose"
bashing orcus
|
|
rpgyay
Known Hero
|
|
posted October 25, 2011 08:53 PM |
|
|
Quote:
You are probably right, but it's just that it requires hex-editing of .exe file at different place than it was in original 3.1 one and we don't know where to change it in new AI exe... If we could maybe get a version of at lest final 3.2 patch without Random ATB, for those who wan't it, that would really be nice...
If Quantomas were to do this he would need to take time off from working on the AI so if he were do this it most likely would be low priority and of course an optional change unless an overwhelming amount of people were in favor of it.
|
|
Pavijan
Adventuring Hero
|
|
posted October 25, 2011 10:45 PM |
|
Edited by Pavijan at 22:49, 25 Oct 2011.
|
Quote:
If Quantomas were to do this he would need to take time off from working on the AI so if he were do this it most likely would be low priority and of course an optional change unless an overwhelming amount of people were in favor of it.
Well it's just changing one float constant value at specific adress, and being that Quantomas would probably know where it is in his own exe file, that would require minimal time to execute.
P.S. Of course it should be optional, I didn't even think otherwise.
____________
|
|
fulano
Known Hero
Can I link to my own avatar?
|
|
posted October 25, 2011 11:03 PM |
|
|
Quote:
P.S. Of course it should be optional.
I love optional!
|
|
PyroStock
Adventuring Hero
|
|
posted October 25, 2011 11:07 PM |
|
|
No random ATB could be an option later, but it's not a priority. No random ATB also shifts the balance more in favor of the high initiative nations/units so it does change the game's balance.
Quote:
Another thing I think the AI needs to improve is that it should have a high priority on casting dispel/magic immunity on creatures that have been frenzied or puppet mastered.
I was considering uploading some battle replays to the 'Heroes 5.5 - Eternal Essence - Battle-Replays' thread, partly because they included:
1>AI Hero removing my puppet master by casting magic immunity on her unit.
2>AI Hero removing my puppet master by casting vampirism on his unit, which made his 100+ paladins much more scary.
In another battle, the AI casted puppet master on my raja cats, but my magic mirror shifted it to my gremlins (who had armor crushing). This prevented my 113 battlemages from shooting so my battlemages attempted to cleanse/dispel it, but failed. Later I calculated they really only had 20%chance of success against a L25 enemy hero.
That aside, it would be impressive to see an AI hero teleport a frenzied unit so the AI uses your own frenzy against you or other tricks.
On another note, 2 towns the AI could use much better are:
Haven = I don't think the AI uses training, which really hurts AI Haven armies. In big maps the AI should have huge mobs of paladins not huge piles of gold.
Academy = AI only uses rank 1 mini-artifacts so the AI Academy special ability is only at 1/3rd it's potential.
These directly lead to the battles themselves being easier later. The AI doesn't use Dungeon's sacrafice, but that isn't that valuable and is too expensive after a certain point. The other towns the AI either uses the town's full abilities or I cannot tell. Some are so minor it isn't important. The enemy AI heroes which have been the most challenging for me have been Fortress heroes due to rune magic. Stronghold heroes are usually the easiest for me due to their big vulnerability to dark magic.
____________
|
|
magnomagus
Admirable
Legendary Hero
modding wizard
|
|
posted October 25, 2011 11:11 PM |
|
|
Quote:
Well it's just changing one float constant value at specific adress, and being that Quantomas would probably know where it is in his own exe file, that would require minimal time to execute.
P.S. Of course it should be optional, I didn't even think otherwise.
Changing the values seems to be very simple, but making it optional means you have to make another menu-option and change the layout of the ui, which is a lot more work. You cannot keep making optional checkboxes for every little balance tweak. I would personally go for a compromise and make it less random, but not completely predetermined.
|
|
fulano
Known Hero
Can I link to my own avatar?
|
|
posted October 25, 2011 11:38 PM |
|
Edited by fulano at 23:38, 25 Oct 2011.
|
Quote:
Changing the values seems to be very simple, but making it optional means you have to make another menu-option and change the layout of the ui, which is a lot more work. You cannot keep making optional checkboxes for every little balance tweak. I would personally go for a compromise and make it less random, but not completely predetermined.
I like that idea more than getting rid of the randomness completely. Maybe only randomize to a certain range. But this really isn't AI related right now...
I also wanted to say that I like it when a game has a mess of options to customize the game before you play, making it very different depending on how you feel like playing. Civilization IV comes to mind, it had over 20 options I could change before starting a game. It makes it more fun to replay the game because each game would be different if I wanted it to be.
|
|
Pavijan
Adventuring Hero
|
|
posted October 25, 2011 11:40 PM |
|
Edited by Pavijan at 23:45, 25 Oct 2011.
|
Quote:
making it optional means you have to make another menu-option and change the layout of the ui, which is a lot more work.
I had actually in mind a separate .exe file with ATB change executed that could be shipped with patch in optional folder. That way Q doesn't have to change anything in GUI or anywhere else, and I seriously doubt that anyone would complain about 12MB (or less) bigger file to download.
____________
|
| |
|
|
This thread is pages long: 
