I'm watching fheros2 progress for a long time (I wish a had more time to
contribute instead), but didn't really ever tried to play it seriously, like
dedicate an evening. I finally did, and would like to share my impressions. In
general - yes, some maps are pretty playable, mostly those involving
exploration and fighting against pre-existing monsters and heroes. When it's
more like development from scratch, AI of course much weaker than in original
heroes, and you can beat it like a child.
I'd like however to invite who played a specific map with specific fheroes2
version, to describe that combination in more detail, hopefully that will
useful as high-level feedback and testing, to help authors improve game beyond
small bugfixing.
r2476:
Alteris (Large, Hard) - Was mostly ok regarding playability, mostly an
exploration level, with some forces-gathering to fight a pre-existing "boss"
hero.
Alteris2 (Large, Expert) - A bit more challenging (I didn't know if I'd have
to re-start it), involved multiple assault attempts on multiple pre-existing
heroes. This was additionally challenged by the lack of heroes meet\t/exchange
features (which was reported and fixed in r2478).
Both levels showed that AI has big problems with boats - not only it didn't
sail across to kick my ass, it didn't even bother to explore across the see.
AI's usage of boat was to jump into boat, sail for ~5-7 cells, get out of
boat, even if that was a "trap" cell surrounded with mountains.
AI's land movement wasn't too active either - there were lot of example of
going back-and-forth within 10 cells, ignoring resources and artifacts.
All in all, AI clearly plays defending role, and the gameplay consists of you
exploring the map, gather forces, sail to AI's place to attack it.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
r2476:
King of Death (Small, Normal) - That was easy, I beated AI like a child.
Pandemonium (Small, Hard) - This levels starts with your heroes being
attacked, then defeated and eliminated from the game. That's not how it starts
in original Heroes, where you can make terrain moves. Apparently, fheroes2
implements wrong turn sequence - in original, human gets first choice, but
fheroes2, somehow AI starts. So, this apparently needs fixing.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I am going to start learning up on AI soon, so hopefully in the future I can
improve the AI. However I would think that AI is very difficult to program,
especially when you don't have access to the original H2 AI code. Any players
with AI experience and who wants to improve the fheroes2 AI would be very
welcome! =)
Best regards,
Steven.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
PS: I have made a high priority Feature Request to be able to set the Turn
Order for games, so games can be customised to the human player/s liking, and
which would have the best in all worlds. Later on, maybe certain maps could be
edited to be set up with certain turn orders specific to that map.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I do not know afletdinov's skill at writing an AI suitable to fheroes2 and its
many Options, but if he was working on AI for a long period of time he would
not be able to do much else. Most likely, it would be better for others to
chip in and have a go, which I intend to do sometime in the future when I've
had some practice. =)
Best regards,
Steven.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Note, we would still need some preliminary AI for when the Wait battle Option
is on though, just to start the ball rolling with reducing the advantage the
player has over the AI which never waits.
Best regards,
Steven.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I do not know afletdinov's skill at writing an AI suitable to fheroes2 and
its many Options, but if he was working on AI for a long period of time he
would not be able to do much else. Most likely, it would be better for others
to chip in and have a go, which I intend to do sometime in the future when
I've had some practice. =)
Exactly, it was my idea, and I didn't try to actually play fheroes2 for a long
time because I remember times when there was no AI at all, so I kept thinking
"it must be not ready yet for real play, it must be not ready". Having played
few maps now, and after quick look at ~100Kb of source code for AI, I can say
that afletdinov did a great job - even if not every existing map, but some are
rather playable. Maybe hardcore Heroes players would still call it too easy,
but for casual player for me which plays 2-3 evenings in few months it's just
ok ;-).
It's true that writing balanced AI is quite hard. I'm afletdinov actually
tested pretty well his code, and made it less "rigorous" in some places,
because otherwise, it would be rather hard on human ;-). AI code of course
needs further development, in particular, to make it more proactive and having
higher-level goals than just "patrol around". And I hope, this topic will host
ideas on this improvements, as well as actual maps to which they would be
applicable, to help with actual testing.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I see a problem with testing particular maps and optimizing AI with them as
long as we don't have any maps in the distribution. Of the mentioned maps I
only have "Slugfest". So please mention a link to the maps you use if
possible.
As a start on AI improvement I suggest reading the h2 strategy guides out
there. I wouldn't mind having a too strong AI for now as I think it should be
easy enough to make it go slower and wait longer (if you turn down difficulty)
and as I have played against the far too passive AI for too long now.
Would be glad to help with testing as soon as there is someone working on it.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Would it be possible to have AI that was harder without increasing the
starting resources or having bonus free resources per turn like the way it is
standard H2?
Best regards,
Steven.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I think it would be better to point users to the demo to take the data files
from as it can be downloaded without legal problems. For example from https:/
/www.pcworld.com/downloads/file/fid,4153-order,4/description.html
But this demo version has only one map in it (broken alliances).
Another thing is that we could add the maps we distribute to the translation
engine to get them in different languages.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Valid point. So, maybe we should start a separate subproject of collecting
Creative Commons licensed maps which could be redistributed with fheroes2. I
don't know how many of such would exist, so maybe also negotiate with authors
of maps to release them under compatible license.
And well, if you just know good freely available maps suitable for play-
testing, just post links to them!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I'm watching fheros2 progress for a long time (I wish a had more time to
contribute instead), but didn't really ever tried to play it seriously, like
dedicate an evening. I finally did, and would like to share my impressions. In
general - yes, some maps are pretty playable, mostly those involving
exploration and fighting against pre-existing monsters and heroes. When it's
more like development from scratch, AI of course much weaker than in original
heroes, and you can beat it like a child.
I'd like however to invite who played a specific map with specific fheroes2
version, to describe that combination in more detail, hopefully that will
useful as high-level feedback and testing, to help authors improve game beyond
small bugfixing.
r2476:
Alteris (Large, Hard) - Was mostly ok regarding playability, mostly an
exploration level, with some forces-gathering to fight a pre-existing "boss"
hero.
Alteris2 (Large, Expert) - A bit more challenging (I didn't know if I'd have
to re-start it), involved multiple assault attempts on multiple pre-existing
heroes. This was additionally challenged by the lack of heroes meet\t/exchange
features (which was reported and fixed in r2478).
Both levels showed that AI has big problems with boats - not only it didn't
sail across to kick my ass, it didn't even bother to explore across the see.
AI's usage of boat was to jump into boat, sail for ~5-7 cells, get out of
boat, even if that was a "trap" cell surrounded with mountains.
AI's land movement wasn't too active either - there were lot of example of
going back-and-forth within 10 cells, ignoring resources and artifacts.
All in all, AI clearly plays defending role, and the gameplay consists of you
exploring the map, gather forces, sail to AI's place to attack it.
r2476:
King of Death (Small, Normal) - That was easy, I beated AI like a child.
Pandemonium (Small, Hard) - This levels starts with your heroes being
attacked, then defeated and eliminated from the game. That's not how it starts
in original Heroes, where you can make terrain moves. Apparently, fheroes2
implements wrong turn sequence - in original, human gets first choice, but
fheroes2, somehow AI starts. So, this apparently needs fixing.
I am going to start learning up on AI soon, so hopefully in the future I can
improve the AI. However I would think that AI is very difficult to program,
especially when you don't have access to the original H2 AI code. Any players
with AI experience and who wants to improve the fheroes2 AI would be very
welcome! =)
Best regards,
Steven.
PS: I have made a high priority Feature Request to be able to set the Turn
Order for games, so games can be customised to the human player/s liking, and
which would have the best in all worlds. Later on, maybe certain maps could be
edited to be set up with certain turn orders specific to that map.
I do not know afletdinov's skill at writing an AI suitable to fheroes2 and its
many Options, but if he was working on AI for a long period of time he would
not be able to do much else. Most likely, it would be better for others to
chip in and have a go, which I intend to do sometime in the future when I've
had some practice. =)
Best regards,
Steven.
Note, we would still need some preliminary AI for when the Wait battle Option
is on though, just to start the ball rolling with reducing the advantage the
player has over the AI which never waits.
Best regards,
Steven.
r2478:
Slugfest (Small, Normal) - less challenging than original, but pretty
balanced.
Exactly, it was my idea, and I didn't try to actually play fheroes2 for a long
time because I remember times when there was no AI at all, so I kept thinking
"it must be not ready yet for real play, it must be not ready". Having played
few maps now, and after quick look at ~100Kb of source code for AI, I can say
that afletdinov did a great job - even if not every existing map, but some are
rather playable. Maybe hardcore Heroes players would still call it too easy,
but for casual player for me which plays 2-3 evenings in few months it's just
ok ;-).
It's true that writing balanced AI is quite hard. I'm afletdinov actually
tested pretty well his code, and made it less "rigorous" in some places,
because otherwise, it would be rather hard on human ;-). AI code of course
needs further development, in particular, to make it more proactive and having
higher-level goals than just "patrol around". And I hope, this topic will host
ideas on this improvements, as well as actual maps to which they would be
applicable, to help with actual testing.
I see a problem with testing particular maps and optimizing AI with them as
long as we don't have any maps in the distribution. Of the mentioned maps I
only have "Slugfest". So please mention a link to the maps you use if
possible.
As a start on AI improvement I suggest reading the h2 strategy guides out
there. I wouldn't mind having a too strong AI for now as I think it should be
easy enough to make it go slower and wait longer (if you turn down difficulty)
and as I have played against the far too passive AI for too long now.
Would be glad to help with testing as soon as there is someone working on it.
Would it be possible to have AI that was harder without increasing the
starting resources or having bonus free resources per turn like the way it is
standard H2?
Best regards,
Steven.
Well, we have no datafiles in the distribution, right? fheroes2 requires
datafiles from the original Heroes, all maps I refer to above are from it.
Since that I also completed "Ludicrosity" (Medium, Hard). Quite playable, but
also feels like Normal of original Heroes.
I think it would be better to point users to the demo to take the data files
from as it can be downloaded without legal problems. For example from https:/
/www.pcworld.com/downloads/file/fid,4153-order,4/description.html
But this demo version has only one map in it (broken alliances).
Another thing is that we could add the maps we distribute to the translation
engine to get them in different languages.
Valid point. So, maybe we should start a separate subproject of collecting
Creative Commons licensed maps which could be redistributed with fheroes2. I
don't know how many of such would exist, so maybe also negotiate with authors
of maps to release them under compatible license.
And well, if you just know good freely available maps suitable for play-
testing, just post links to them!