Developer Diary


The tasks section is frequently updated to provide information on what is currently scheduled.

The links section contains links to our community discussion threads. You are invited to join us.

25-Dec-2014 — APPROACHING 3.2.ALPHA

01-Aug-2012 — CURRENT WORK

01-Jan-2012 — BRIEF OVERVIEW



19-May-2011 — KEY TOPICS — What's in storeHow can I join? — HALL OF DECISION DESIGN CONTEST



15-Apr-2011 — KEY TOPICS — MotivationVision and designWhy this project can happen


25 December 2014


This is just a short note that work will soon pass the first AI prototype completion milestone. This will be followed by implementing a modest amount of additional code to join the AI with the game, debugging and testing, which will hopefully not take too long.

In the meantime our board at will be down. We will relaunch with the release of the 3.2.alpha on


1 August 2012


Recently a number of inquiries have been made whether this project is still alive and whether work on this or that feature is planned.

First, this project is healthy and moving forward. There are a number of posters who painted a picture of a programmer weighed down by the workload and sorely lacking support (of other programmers). This is far from the truth, the development effort is excellently organised but it is BIG. It is a bit of a pity that in our society you can't talk freely about bleeding edge technological innovations until you have materialized a sound competitive advantage. The work on the artificial intelligence is absolutely fascinating and it would be a joy to share all the advanced concepts it touches. This involves learning from nature how intelligence comes to be, from neuronal response patterns over semiochemicals (e.g. pheromones) to how information is filtered, optimized and interpreted in organic systems. Other areas include finite element analysis, advanced maths and physics, the wave-particle duality turned out to be highly significant for AI design, soft decision criteria, efficient network optimizations and many more. I would love to talk about this and how each element is relevant to the AI design. How you systematically have to investigate each component of the system (and game) and establish its impact on leverage, how you make strategic sense of it.

In short we are developing a fully fledged AI, rather an AI operating system that can adapt to many different contexts (and games). Those people who know a bit about this work will understand that you can't simply commission it. I am confident that a team of top programmers at Blizzard or Valve could come up with a fairly well developed non-cheating AI for any game if they put their minds to it, but developing a universal and capable AI that truly understands a strategic situation and acts according to a plan it has arrived at itself is an entirely different beast. Such a task requires vision and skills, ultimately being prepared to learn whatever it takes, on a level that only comes with true dedication. No amount of money can buy this.

I believe that playing a game with such an AI will be utter fun. This doesn't mean that the work isn't painful. But it is painful rather in the sense of how Richard (one of the protagonists of Terry Goodkind's outstanding Sword of Truth series) is carrying heavy bars of steel without help through the city of Altur'Rang. Something someone would only do in dire need, an expression of unshakable faith.

This development cycle, the creation of our AI is coming to a close now, I'd say it is 99% complete, I am merely honing the algorithms. Hopefully before too long you will be able to try out the AI for yourself.

The reason that I am explaining all this is to give you a glimpse of the big picture. While it might appear that not much had happened in the past half year, in reality a nearly unsurmountable stretch of ground has been covered. This AI is by far the biggest piece of technology in the entire project, it dwarfes all other items on my schedule, be it the migration to our own engine (Blender based), the non-linear strategic world map mode, even the infinite Eternal Essence magic system and the Civ gameplay component.


First of all, have patience.

I am entirely aware that this project is in respect of the community not organised ideally. What we are sorely missing is a good moderator, someone who can act as a second in command and learns everything there is about the project to help with the management. If you consider the scope of the work described above, there is no freedom to fill this part as well, so please pardon my silence.

Regarding all the tasks and potential upgrades for H5 brought to my attention, and looking at what is planned for Eternal Essence proper, there is work for years already on my schedule. This means we must have clear-cut priorities and we have to tackle the issue of funding in order to get much more done.

In general I tend more than ever to leave H5 behind earlier and move to Eternal Essence (a spiritual successor) faster. The reason is simple: once we take this step we will be free from copyright and NDA and have our own codebase (which would allow us to assign coding tasks to different programmers), plus we can build a clean modding interface that supports the addition of new 3D models as we see fit (instead of figuring out the proprietary H5/Maya model conversion).

Working with the H5 code we have is a mixed blessing. Because we didn't have 3.1 proper, I had to implement the entire widescreen functionality and a lot of other features and bug fixes on top of everything else. As many of you are aware, there are still a good deal of the 3.1 fixes missing, upgrades to the scripting and multiplayer functionality broken and so on. I am reluctant to assign these tasks a high priority because it would be nothing more than to fix what had been fixed already (proper access to the 3.1 source would instantly resolve these issues).

The upcoming AI core is a different story altogether, it makes sense to integrate it into H5 and see what a difference it makes. Any work we do here will migrate to Eternal Essence nicely.

Funding is really the critical issue to accomodate both camps. Community purists are aghast at this thought, but if you look more closely at the project you will see that we are not a community mod creator but have already taken the first steps to position us as an indie developer. The reality simply is: with more funds you can do more. We have two funding needs: the first is roughly Euro 25,000. This is simply for me to work without constraints and not critical to make or break the project (it won't). But it can substantially speed up our progress. I have done professional pitching before, and it always consumes lots of time, one or two months at least, mostly for preparing your pitch, not counting the time for travelling. If you read this and have money to spare, this could very well be the best investment you ever made. Think about it. It would allow me an additional degree of flexibility, both in terms of development and project management.

For the next stage, to get the project truly off the ground, we will need something like a Kickstarter in the order of Euro 250,000 to establish a small core team (minimum wages initially) and to pay freelancers, artists and others to realize Eternal Essence as a full spiritual successor. The funding goal can very well be flexible, that more funds will enable more features, content, art, story, voice acting and so on.

This is exactly the point at which I would like to revisit how much we can do for H5, whether it makes sense to keep it alive and support it much longer. My idea is that once we launch our Kickstarter (or equivalent), we will ask the people donating to vote with their pledged amount for features they want to see implemented, e.g. multiplayer, mods, campaigns or other features. At that point I would like to offer H5 fans that we will implement whatever H5 feature gets sufficient funds if it is feasible to do.

Eternal Essence on the other hand might prove more attractive though, because this game will have real magic, it will be built around the first universal AI and this will bring an entire world to life, including intriguing ways how the story can unfold and how content can be added. How you can interact with the world. The big motive behind Eternal Essence is that we explore real mysteries that are on the minds of people today: the emergence of intelligence, the ultimate potential of intelligent beings and humans in particular, where are we truly coming from, what is our destiny and the like.

In a way I would like to build a world that has a depth like The Wheel of Time, and supports gameplay mechanics and ways for the player to make choices that go beyond anything seen before in virtual worlds.

AI is our friend.


1 January 2012


This is not a full developer diary entry. The next one will follow most likely once we have the combat AI overhaul.

But there was a slew of additional important overview material that is scattered across the board, even in the off-topic area. The board also explains what Eternal Essence actually is whereas these pages are mostly related to its origin in Heroes.

Here are the links roughly in order of importance:

New year overview - what Eternal Essence can change

Season's change - overview of the scheduled tasks

Heroes 6 and Heroes 5 - how Heroes and TBS can move forward

There was also an e-mail with some highly relevant questions. Here is the transcript:


Let me continue here from my HCM questions. I don't want to take an endless amount of your time, but hope that you can give me a small impression of what the current situation of the project is - I must admit I find the homepage quite confusing, and I haven't had time to read all of the posts on HC. I don't understand what is actually currently being worked on, nor what the current status is. Maybe I'm missing something - and perhaps the most relevant question is: Is there somewhere I can actually read what is the current status of the project (I need to be guided more specifically than just 'the homepage')?

My own dream for Heroes 5 would be a fleshed out magic system with 4 spells for each school on each level, and a fleshed out skill system with 4 basic perks in each skill that lead to exactly one advanced and expert perk (so a total of 12 perks in each skill, with no branching perk lines, such as Master Of Fire opening for multiple advanced perks). And last but not least, a proper Town Portal spell, plus a softer leveling curve at level 25+ with the level cap possibly increased. Oh yes, and ability to have one more skill slot. :P

Anyway, the important thing here is not the details of my dream, but whether there are actually plans to undertake such things (a modification of skill and spell system), whether it can be done (realistically), and if yes to the first question, if work is already in progress.


First thing, the project website is basically outdated, I haven't had the time to do anything there for the last three months. What is kept current is the community board and the information there. It is scattered undoubtely.

My main problem is time. There wasn't any new patch release for about three months, because all the other tasks eat away at my time. The AI work that is currently scheduled, i.e. the combat AI overhaul, the adventure map AI middlegame plus the integration of both, is highly specialized and you would normally need a team of Blizzard quality AI programmers working on this for a year and more. I can do this in three months because I have already invested years in this project and know the full scope, and are faster than most programmers to boot. BUT, and this is a big but, I have real life issues to deal with, and with all the other tasks I hardly can get any hours in on working on the code, which means the project currently slowly dies because there is no progress. I will not let that happen, but this means other tasks need to wait. I say this just for clarity. Many people in the past were disappointed by this that I asked them to wait with requests (even if they offered to work on it) or that they have to put an effort into reading up on what's going on or talking to other community members.

With this out of the way, the current plan is to implement the scheduled AI upgrades, the framework for the non-linear campaign mode and some other things like the tournament mode Right of Conquest (SP and MP) and the Hall of Decision endgame. Once we complete this we will release eventually a proper 3.2 patch for TotE.

After we have reached this milestone, we will go on to transform the game in a more ambitious way. Actually, the feature upgrades you mention need to be seen in this context.

Skill system: what we will do is much more ambitious. You will be able to define a skill system exactly as you lined out. For this to work, we will implement an underlying core system in the game (tied to the Eternal Essence magic) that lets you define skills and perks freely, whether these are stat increases, new abilities, modifications of others, magic spells or whatever. Additionally we will make this customizable, so that a map maker or player can elect to have only a given skill set of choice active.

Magic system: similar to the skill system, the magic and crafting system will be totally open (only limited by the Eternal Essence mechanics which are infinite), so that you can research (and define as a preset) a near unlimited variablity of combat and adventure map spells. This will also be customizable to the extent that what is possible on a given map can be limited.

If you consider the skill and magic system together, you will be able to do exactly what you asked about. But when this will happen depends mostly on how much work I can spend on the code.

I hope this helps. I know it requires patience and I blame nobody who says they will just wait and see.

Here is a request I have: would you mind if I post your questions and this reply on our board? E-mail has turned out again and again too inefficient for things like this, because this way I would need to explain things over and over again. I rather prefer such questions and explanations on the board for all to see.

I wished things were more easy, but alas these need time. What I hope for is one day to have a moderator on the board who knows all these details and could explain things in my stead. But right now it's a constant fight, to do what is necessary and still to get some work done.


19 October 2011


Originally it was planned to release the new enhanced AI stage for the middle game synchronously with H6. This is a truly big improvement because it will actually give the AI a lookahead depth of six weeks (currently seven days) and the ability to develop actively strategies that will have the AI coordinate its forces and advance towards the most relevant strategic targets while doing its best to defend its holdings. Due to the complexity of the work, it still needs more time.

Nevertheless it was necessary to publish a new improved stable build at H6's launch to say this is not the end of H5 and that both can develop independently as strategy games with a different gameplay focus. The stable build released (3.1i) is a very big improvement already and contains lots of enhancements. The most notable is that the last two remaining cheats Nival introduced have been removed and it is now a truly non-cheating AI that can hold its own against veteran players. Most certainly you will note that the gameplay is now much more balanced, you can observe the AI tackling the same challenges you face to build up its power base. But keep in mind, that with the cheats removed it will perform a bit weaker than in the earlier builds. More optimizations for the base building stage are in the works and once the enhanced AI stage for the middle game is out, this will more than make up for the removed cheats and provide a challenging and equally fair game with lots of interesting strategic nuances.

The work is continuing and the 3.2 release is the big milestone that will see the AI at its full strength. We are also working to set up our own independent production, Eternal Essence, that will share the spirit of Heroes but will move much further into the direction of a refined strategic gameplay. Did you not always desire to be a king and raise your fledgeling kingdom into a magnificient empire? Eternal Essence will allow you to do exactly that with new gameplay elements, an open researchable magic system, and a completely non-linear campaign mode. More info will follow when the 3.2 release is out.

The workload, programming and administrative tasks, is currently just insane. Doing all this with a minimal budget makes it really tough, so if you can please check out our Support Us page.


While the beta (3.1.xx where xx is a number) is normally the most advanced build, this is not true if the latest stable release is more recent. Please keep in mind to check the release dates at our General page.

3.1i contains many improvements over the earlier beta, it is not just the same release without debug logging. If you would like to lift the shroud off the map, like the beta open builds do, it is better to start the game with the beta open build, play one turn, save the game, and then to continue with 3.1i.


The combat AI will receive a complete overhaul. The goal is to develop it into the stellar, mathematical exact combat engine that was one of the hallmarks of Heroes III. I never witnessed it doing something stupid in the battle arena. Once this is done it will also allow for a much better integration between the tactical combat and the gameplay on the adventure map. The idea is to have the AI develop the proper skills and forces to tackle its opponents in the strategically most expedient way. It will also take into account the composition of neutral monster stacks and devise strategies with which forces to tackle them. Compared to what we have now, there is much room for improvement.

The combat AI overhaul will not happen before the adventure map AI improvements scheduled for 3.2 are done. This means, there is not much need to report battle arena issues, because these will most likely be resolved wholesale by the combat AI rewrite. It will replace the combat AI completely. The game mechanics will remain unchanged for 3.2.


Included with 3.1i are a number of new and updated reference maps. The latest addition is Markkur's outstanding Power of Dragons. This map plays actually a bit like the epic maps in Heroes III. Power of Dragons also shows how far our AI project has come compared to the H5 vanilla. Try out Power of Dragons with the vanilla if you are sceptical, and find yourself surprised how big the difference is.

Many thanks to Markkur for the stellar work on our reference maps. Power of Dragons, with all its linked places and heavy load of portals, was also ideal for shaking out bugs of the latest build and has made it much stronger.


This entry's recommendation is the custom map Sanquinary Ascension (available from with difficulty=hard, lookahead=heroic, neutral champions=easy, artefact guards=on, taxes=off, eight skills=on, starting resources=equal, all AIs aggressive.

The experience playing this map is exceptional. It draws not only on all elements the Heroes 5 game has to offer, but gives you a good fair challenge to boot. The story is good, and the amount of attention that has been paid to details is incredible. Be amazed by a crowd that cheers you on in the arena or the animations you find in little places on the map to enhance the story.

1 July 2011


It's sort of a recurring theme that people claim that an AI needs to cheat to be challenging. This is not only wrong because it distorts the gameplay but also the state of the art has advanced with the computing power available on the desktop now to create perfectly balanced non-cheating AIs.

Let's first have a look at what cheating means for a strategy game. For example, consider a cheat introduced in Heroes V ToE vanilla, that gives the AI more creatures to recruit each week to counter stronger human players. Imagine a map where the balance between different factions, each located in their own region, is about even. Let's assume that the human player is strong enough to outperform AI opponents in the battle by a factor 1.5 and that the AI receives a creature growth cheat of 1.5 to balance this. As the game progresses all factions suffer attrition, based on the ratio between the army strength and the neutral stacks encountered, and reinforce regularly. There are neutral stacks that might join an army if it is much stronger, the number of creatues does count here. The factions might capture additional towns with more recruits available as well as external dwellings which also add creatures and growth.

If we would have a map that provides for an even balance of power between equally strong opponents without cheats, will the cheat described above allow the map to play out similarly? The answer is simply no. First, the cheat might allow for different rush dynamics, e.g. attacking early with overwhelming force. Second, as the game progresses the balance also shifts because the AI will perform much better against neutral stacks. Third, additional towns captured might alter the balance differently with the cheat. Fourth, the value of resources will be different for the AI because it has more recruits to hire.

The point I want to make here is that cheats alter the game completely. While the original game design with all its rules does support an intricate balance between all elements, cheats tend to dominate the entire gameplay. It leads to a more roughshod game where the intricate game mechanics matter less. While there might be narrow paths, where such cheats can have the intended effect, for example in the design of a campaign map, where the player is guided by objectives, there are many more scenarios, also in the campaigns, where the player deviates from the intended course or tries things out on a different difficulty level, and sees the gameplay completely distorted.

You say this does not matter much? Then why do we have this intricate rule set and the complex and myriad variations it offers? There are people out there who take pride to balance hero skills and creature traits, and say this works for one reason or another. People like variety if it adds depth and can be used intelligently in a strategy game. Nothing beats a tough strategic contest around an observed balance of power that has the opponents give and take but not making much progress, and then you can find out how you can use a special talent or feature of the map to gain the upper hand. This could be everything, the goldmine in a critical location, a powerful artifact, a solid combination of hero skills and spells, or capturing a crucial adventure map location. With cheats this type of gameplay rarely matters.

What fans really desire is a gameplay that adapts dynamically to the chosen difficulty level and still plays out truthfully to the game's rules. It not only caters for different audiences but it allows players to gain skill and to see the game in a new light with every advancement of their own. Unlimited replayability. Add to it an AI that can make intelligent use of all the game's rules and adapt to a player's strategies, and you have the ingredients for an addictive game. With cheats you will never get there.


I will add a feature to this diary to recommend one map with each entry. Having done a lot of playtesting and having others in the community do the same and point out well designed interesting maps, it's probably about time to give a recommendation for those maps that offer something special.

This entry's recommendation is the stock map The Greed with difficulty=hard, lookahead=heroic, neutral champions=easy, artefact guards=on, taxes=off, eight skills=on, starting resources=equal, all AIs aggressive. The factions are Haven (red), Inferno (green), Necropolis (yellow) and Orc Stronghold (blue).

This allows for an excellently balanced game that has the factions observe the balance of power. It is difficult to gain an advantage. You have to fight to secure the gold mines that in turn help you to build up your faction. Control of the central towns is another key to build leverage against your opponents. If you leave one in the hands of the orcs who are already a strong faction, you will face an uphill struggle. In general, with the new build 3.1f, the AI factions tend to observe the balance of power and try to maintain a strong power base.


We had a good controversy since the introduction of the neutral champions whether they are too strong. The reason I prefer to leave them the way they are, with an equitable selection of skills and a random choice of spells, is that it allows us to see the game mechanics in a new light.

Of course, at first it is a shock to see the neutral champion who supports a pack of Blood Furies cast Armageddon on your forces. But, you can adapt your tactics based on the neutral champion you encounter. You can learn and make smarter decisions. For example, with Dungeon I now tend to rely on Minotaurs and Hydras and don't field my more vulnerable units if I encounter a neutral champion. Make sure to check the champion's stats before you proceed to the next battle phase. The stat boosters on the map are valuable now. It also has become much more important to choose the skills that matter most. While you originally might just have picked a skill you would like to have your hero build up later on, now you have to choose the skills that really make a difference in battle. I never really fancied defense before, but now its magic protection is valuable. Fire protection is good and there are many new strategies to try out.

Do not make the mistake to enter a crypt lightheartedly with a high level hero. The necromancer might cast Curse of the Netherworld and inflict a good number of casualties to your forces that outweighs the monetary gain by far. The crypt is a building that is now really suitable for low level heroes. And your faction matters. For example a high-level Necromancer does not need to fear the crypt because of his skills. On the other hand he will have a very difficult time in a dwarven treasury.

And do not forget: the neutral champions allow you to surrender and keep your forces if things really get rough.


TheBaron has outlined an incredible mature background and culture for a new faction that will be introduced by a new campaign. This will revolve around the powerful and mysterious force of Eternal Essence that will be at the heart of a reenvisioned Ashan.

Many more things will come to light and the game will have substantially more depth. This goes for story, a non-linear campaign mode, and the game mechanics itself.

Check out our new section for The Shifter campaign.


Markkur has now presented revision 7 of the Hall of Decision. This is an ancient and forgotten dwarven city, realized to an extent H5 has not seen yet.

We are now at the point, where the map is fairly advanced and we can begin to focus on the quests and game mechanics in the Hall of Decision itself. Remember that this map plays out in a special mode that lifts the fog of war only in your immediate vicinity, the surroundings you can actually see from your heroes point of view.

This will be much more non-linear than many of you might expect. Essentially it is a race to visit all fountains and then make it to the fountain of victory, but anything can happen in-between. Being a race the map has its own inherent time limit.


We have now a new reference map candidate Axis of Evil. It is special in the sense that it has a secondary town for each faction to capture nearby. Because this town is accessible by a one-way-portal it presents special challenges for the AI and the player. While the current AI still struggles to make best use of this situation, the upcoming improvements with the strategic AI end stage, will see this becoming an interesting scenario.

19 May 2011


I was recently asked what content is planned, particularly what maps and campaigns.

There are three layers of content that will be successively introduced into the game. The first one are our special purpose maps, that have alterated gameplay mechanics and victory conditions. You can think of them like the bubbles of space in the great Planescape Torment, places that could be visited through a dimension door, each probably located on a diffent plane where different rules apply. NWN2 Mask of the Betrayer also used a similar concept where you could shift through a portal to a different plane. Eternal Essence will allow you to scour for hidden passages and doors and look for special objects that open a dimension door to new places to visit. The first one will be our forthcoming Hall of Decision. This is more than a simple map, rather it is the crown juwel where players battle out the final stage of the game, sort of the map that every player of Heroes 5.5 will get to see sooner or later.

Second, I plan to have the ToE campaign fully reworked to accomodate a truly interesting strategic gameplay. One part is to tweak the AI, but the other is to redesign the maps around a balance of power. In the original there are way too many maps that center around a gameplay that is rather simple and in the end is little more than to assemble an army to capture a specific town. What I would like to see is a strategic contest in which different parties give and take, with much more active opponents on maps that allow for depth to challenge your wits. There is a bit of an additional explanation at the beginning of our HC discussion thread, otherwise feel free to ask in the forum and I will explain more. The same is true for the original Heroes V and Hammers of Fate campaigns. It should be possible to make them playable with ToE inclusive cut scenes and dialogs, if the proper games are installed on a PC.

The third part is where things really will get interesting. I have a vision for a completely non-linear campaign mode in which the player will discover a rich story and encounter the various new gameplay enhancements planned for Eternal Essence. Enough for an entire new game that possibly can be produced and released as a full title. Whether we will be able to do that depends on Ubisoft and right now it's way too early to consider it more than a possibility. But if we can get Heroes 5.5 up to speed and can create a stellar strategy game and have 100,000 fans clamoring for more, it can become reality.


The other question that was asked was how can I join the project?

We don't have an internal mailing list or a core group who communicates internally. My own position is that e-mail produces way too much of an overhead and is not reliable (it's neither secure nor failsafe). The problem that I face, as the lead programmer of the project, is that I have to keep the overheads as small as possible to have sufficient time for the critical programming tasks to keep the project moving at a decent speed. One poster put it that way, that I am paying for the tasks I work on with my lifetime, but this is only part of the picture. What matters more, if I would work on nice-to-have and only-important-for-some tasks, the truly powerful enhancements would be delayed for years. This is the reason why I frequently have to reject well meant suggestions, there is simply no other way.

The same is true for our communications. The most effective mode is to have the entire discussion in the open. There might arise a need to have our own forum to give our discussion a better structure in the future. We have enough space on our servers, so it's only a question of organisation. But I believe that we should keep our discussion in the open and learn how this can be done most efficiently. I don't really like to say things in individual e-mails multiple times, so I prefer a question in the forum. This way all people get to know about it, and can reply to it appropriately. Also all answers are shared immediately. In addition I maintain this website here to keep things together.

So, in short the answer is, participate in our discussions. If you are a map maker or a veteran player you can also participate in our recent Hall of Decision design contest. Check out what capabilities the lastest builds have and what's new on this website from time to time.


I would like to invite all talented map makers to a small contest to design the map for the forthcoming Hall of Decision endgame in Heroes V.

This can incorporate all kinds of ideas to offer a truly interesting challenge. I would also like to invite all veteran players who have an indepth understanding of the game to come up with a rough design sketch. If you don't feel you are good at map design but have the right ideas for the map and the gameplay on it, we can accept that and have other people work out and beautify the map itself.

Basics and general description

No towns, the strongest heroes of all players on the adventure map are teleported to a random location in the Hall of Decision at the beginning.

Five differently coloured fountains are placed on the map. The first player who visits all and reaches a specified exit building wins the game. The gameplay is not about heroes building up their experience or armies, but rather about using their existing skills and strength to succeed.

The map will run in a special mode, which limits the vision of the player to the first person perspective of the hero. The hero can not see around corners or through doors. I will also implement special logic for whatever the map designer requests (and is feasible) that goes way beyond scripts. That means, you don't need to write scripts, this task is handled directly here. If you wish adventure objects to work differently or serve special functions just write a note.

My original idea was to have a dungeon with tiny rooms each offering a special challenge and reward, like in the old dungeon master games. Additionally the colour of the fountains can confer considerable boni and can serve as a key to bypass strong garrisons without a fight or use special portals. Here is a very rough sample for all who do not know what I mean with small rooms and dungeons. Ideally these rooms will have proper doors to hide what is behind them.

But anything good is welcome, if you have better ideas than dungeons let your creativity reign free with a focus on the gameplay. The size of the map is your choice, but preferably not bigger than necessary to suit the intended playing style and underground.

What you get in return

Well, we can't offer more than full credits and that your map gets a central role in Heroes 5.5. If there are more good contenders we can also choose to have a set of Hall of Decision maps that are either selected randomly or choosen by the player who reaches the victory building on the adventure map. But the best reward will be the fun to play the Hall of Decision map instead of the boredom to grind down the remaining opponents on the adventure map.


You can submit your map or concept in our design thread at HC or in an encrypted zip archive to h55Maps [at] simsysresearch [dot] com.

11 May 2011


One goal of our project is to improve the game with new victory conditions. There are already two nearly ready to be implemented soon: the Right of Conquest and Hall of Decision. Here is the first concept for the Hall of Decision and Right of Conquest victory conditions.

Both victory conditions will be available once one player reaches a superior position. Other triggers are open to discussion. Once the victory condition is enabled a special building will appear on the map, preferably in a random remote location. The first player who visits the victory building gets to choose whether it's going to be a Hall of Decision or Right of Conquest and receives a worthy bonus in the final stage.


The Right of Conquest is an arena death match. All heroes above a certain power threshold become combatants, with their army and artefacts transferred intact. The matches of the first round are determined by a draw. The looser drops out of the competition while the winner gets a bonus. After each round the winner will be presented with an arena match victory screen on which he or she can select a bonus, receive and exchange artefacts from the opponent's inventory and more. It will also be possible to use the bonus or a part of it to replenish some mana. It's open to discussion whether the player's resources and potential recruits in the towns can be factored in or whether the arena master can offer special rewards. Combat results are announced each round with the player given an option to view the fights. The last standing hero wins the game.


The Hall of Decision plays out on a tiny map, some underground cavern perhaps, no towns. Vision will be limited to the hero's view. That is the player can only observe what the hero can actually see. This allows for some neat dungeon design with different chambers and doors. A bit like the old dungeon master games. You don't know what's behind the door until you open it. There can be guards, rewards and portals.

The strongest hero of each player is teleported to the Hall of Decision once this victory condition is initiated. It's not simply a catch and kill game, but rather a subtle endeavor that will allow also for the underdog to win the day. Imagine five fountains of different color placed in the Hall of Decision. The hero who visits all fountains and reaches the exit first wins the game. Each fountain might confer considerable boni and its color might be a key, for example to pass through strong garrisons or portals.

24 April 2011


What does the magic essence actually do to enhance the game and how does it give more freedom to designers?

First, having everything in the game based on one coherent basis and seamlessly interface opens infinite possibilities that were not there in the first place. Second, the universal building blocks aka magic essence will allow us to design and tune new in-game rules very fast. The game can work like a sand box mode because of its universal adaptability.

The strength of this approach is its universality. Except for the people with a math background or abstract thinking programmers, I doubt hardly anybody has noticed what is truly to come.

Imagine a fantasy world where magic seeps from the land and all inhabitants somehow are affected by it. In a game that gives you the opportunity to interact with this magic, you can expect to be able to modify a myriad of things, you gain control how the land reacts to whatever you do, what creatures and heroes can do and what structures emerge on the land and how towns evolve. Your interactions become magic in their nature. Your interactions and the magic of the land will seamlessly integrate.

What I am saying is that modders and to some extent players themselves can control the way the game works. While you are used to game designers defining the rules what you can do, we will have a game in which every rule is part of a universal magic framework and the interactions within the game as well as all predefined rules and map objects live and interact naturally with each other. There will be barriers what a player can do, or what an individual hero or creature can do. But the designers (think modders) are empowered in their capabilities to change whatever they see fit. Don't think of this as a vast conglomerate of in-game rules, instead it is mathematical in its nature, as universal and powerful as algebra and geometry and analysis, but here in its magic equivalent it is a universal model for everything that lives in the game.

Don't you worry about Heroes V. Will Eternal Essence be a different game? The answer is yes and no.

The first thing we will do, once we have the magic framework to rework everything, is to translate what we have in Heroes V so that it will work in the magic framework just as it had in Heroes V. But there is more. Heroes V has all the game rules implemented hard-coded and rather as special cases handled by special code in the proper places. From this we have inherited tons of bugs and all kinds of limitations. Things like the wrongly predicted damage in combat or that creature perks sometimes do not work, spell effects are not applied correctly etc. All these bugs, whether we know of them or not, will be gone because on the next level we don't have to handle specific cases anymore.

But more importantly we will be free of hard-coded limits. We will be able to add and modify hero skills and perks, what creatures can do, how the town buildings work and what they cost, how spells work and research new ones.

All these modifications will be optional. If a player wishes to play a vanilla Heroes V game, he or she can just do that. But then there will be options to add additional skills, interesting sets of new creatures, new town types and alterations to the magic spell system, or if desired a totally weired and mind-blowing combat mechanic. Not to forget modified and dynamically changing terrain, powerful overland spells and castles that have their regions develop with characteristic features determined by the magic resources nearby.

Developers and the big publishers behind them have not yet implemented a universal in-game system like that. It's not that they couldn't see the benefits that such an approach can have. Instead they see that this has a very huge price tag, particularly if you have to build an AI on top of it, and going to design the in-game rules the traditional way is definitely more economic. But then the game industry is stuck in its way to produce games with ever more mass appeal. Only it has turned out that simplifying a game and that their idea of what creates mass appeal, is not exactly increasing the audience, it rather produces more lookalikes and feelalikes in a market that gets more saturated. It is interesting to note that the big PC games that pushed the boundary with innovations and big success, for example Dragon Age I and The Witcher, paid huge attention to how intricate game rules can work together seamlessly to empower the player and enhance the playing experience.

We need to break out of existing habits if we truly want to break new ground. This is where the value of Eternal Essence and all its empowering features lies for me, as a developer to advance the state of the art in strategy games and AI, and as a gamer to see more innovative and interesting games.

I have my reservations that H6 will break new ground in this direction. It is more of an effort to rebalance the formula to find the sweet spot. If it could, even move a bit in this direction, it would be great, but considering that it is a strategy game that needs a competent AI to be truly interesting, I have my doubts. It will be fun to play the campaigns and to check out the content, I love the fact that they included a map this time and focus on story, and while exploring the game we always hope for more surprises and that the game will offer true replay value.


As it was with Heroes V, I had the same hopes and reservations. As a veteran strategy player a competent AI is absolutely essential for the fun, and setting out to play the campaign my hope was still intact after seeing that the AI was at least doing a basic job. If you have played the campaign, you will have come across a scenario in which you basically had to traverse an Inferno territory without a town base. You recruited creatures from dwellings, high level ones as well to give you angels and arch angels after having visited a hill fort, and sort of had to build a decent force to pass by their castles. These castles had very strong garrisons that were intended to make them unassailable after you had defeated their strong wandering heroes, but as a veteran player and having arch angels with you it was entirely possible to build an ever increasing force. While aware that capturing their castles was no mission goal, I was still curious and captured a castle. It was destroyed of course and useless as a base, but I got a message saying "that's impossible, you must have cheated". Big dash for my hope, that statement clearly said they don't understand what a veteran strategy player can do.

Well, I said to myself, maybe it was just the designer of the map, let's go on. After I had witnessed that they actually spawned heroes and armies on the border of the map, so that you had to make a mad rush to defend your castle, which indeed was challenging but not in a strategic sense, my hope took another big dash. There was no subtle strategic gameplay to be had, and many more scenarios later, it was clear to me, that this game will only get to its true height if the AI was to be improved significantly.


This is an important lesson learned. Again I hope that they get it right this time with H6. But it's just that, hope and reservations, and to look forward to play the game and explore what is possible in this installment. But regarding the sweet spot, something that could make the game the biggest strategy game around, my personal opinion is that this cannot be done without at least a competent AI. Workarounds like the one mentioned above in Heroes V go only so far. Still, this is not a problem for me, now that we can go on to improve the AI in Heroes 5.5 - Eternal Essence. We can always go back and hope that the devs take note and improve the game in the next expansion.

So this is where we are. We improve Heroes V in the hope to make real gains for the franchise and hope the same for H6.

In many ways the universal framework of Eternal Essence will allow us to have it work like a sand box for new in-game rules because of its universal adaptability. We can try things out, and can move in this direction or that to find the sweet spot, the formula that is most fun to play Heroes.

On top of that I am building the most advanced AI, so that the game can be subtle or a rush, whatever you desire but in any case a blast and an experience worthy of reliving. A strategy game of this type can indeed have an unlimited replay value and make it to an all-time classic.

15 April 2011


Originally the AI prototype was written as a potential replacement for the AI if Nival would have done H6.

As a long time Heroes fan and veteran strategy player, it was obvious to me that an improved AI would significantly strengthen the game. As a professional AI programmer, I could imagine what a game it truly could be if we had an AI that can challenge the wits of the strongest player and at the same time adjusts itself to the strength of any player, so that every player has a chance to win the game and an experience that is worth reliving. An AI that is more than just a means to process the turns of the AI opponents, but is a valuable service to all aspects of the game. This includes all types of game balancing and support for the player in various game situations. It will also allow the game designers to make much bolder steps and pursue goals that make the game fun to play in every aspect and includes features like AI assisted map making.

As such it is a project to raise awareness for the state of the AI in the industry and to demonstrate what a fully developed AI can truly do for a game. There is a more detailed explanation on the main page and in the dedicated AI chapter.


The original AI prototype was developed in 2008.

My first task after Ubisoft had approved of the release of an unofficial AI patch for Heroes 5 was to review what we had and to make the step from prototype code to production level code, with the goal to release a fully stable and functional upgrade to the game.

The AI code had to be migrated to its own process to achieve two goals: first, having the AI computation in a different process minimizes the stress put on the original application by requiring less resources. Second, the application can fully recover from any crashes in the AI process without the user noticing.

The results are improved stability, particularly under Windows 7, and faster processing times while the animations still show fluidly. There is a lot of potential for increased concurrency on systems with more processor cores in the future.

Technically, this work involved the refactoring of 35,000 lines of AI source code. Not bloated but highly optimized and fine-tuned code, where everything fits into its right place. Sort of this craftmanship that you can see in the buildings that endured millenia in Egypt, Greece and Rome, and that hardly anybody could afford today.

This work was done in March 2011. Don't ask me how I did this work in a single month. I don't think it's possible.

In crunch mode I am ten times faster than your regular programmer. Add to it the fact that I know every bit of the code whereas a team would somehow need to partition the tasks, and you can imagine how long this work would regularly take.

This work also marks a milestone from the original prototype code to production level code. The base we have now is very flexible and extensible and many more things become possible. This was also the very big chunk of work that needed doing first. Everything else should be downhills from now on and considerably faster.

That said, it was kind of a strain, but now seeing how the animations are playing fluidly while the AI does its processing in the background makes every bit of it worth the effort. If you don't have a multicore CPU, get one, it's worth the investment. Seeing how the game has actually a bit of idle time, makes me think of looking for a way to enable anti-aliasing for the 3D view only.

The refactored AI version is now part of the 3.1.n beta release. This will be improved upon and once it is fully stable and includes a number of scheduled upgrades, it will go into a new 3.2 release that is intended to bring Heroes V to a new level in terms of AI and gameplay.


Let's start with a vision. H6 will most likely simplify a couple of things to make the gameplay more slick, and with the focus on story and content it should be quite an interesting ride. In the long run I would like to see H6 and H5 evolve like brothers who grow up together. What we can do with H5 is to be true to its strategy game roots and excel in this area.

Eternal Essence has at its core magic seeping from the land and driving the life of its inhabitants. Everything depends on this magic life force, the earth itself, the mountains and trees, creatures and heroes. Buildings and towns are made of it. Don't think of this magic as something ethereal only, it has a physical quality as well to take the shape of stone and water and wood and everything is made out of it.

Now, if we are going to create a game on this premise, what we truly need is a universal building block from which to create every object in the game world. Don't think that we would need to recreate the animations from scratch for this. What this really is about is to define everything in the game from a coherent basis, so that we can express any relation between objects in the game world and define how things work. As the most evolved turn-based game of its type Heroes does already contain a sheer insane lot of definitions of how things work. Beginning with terrain movement costs over hero classes and skills that affect movement speed, to resources, costs, buildings and creatures. The thing is, currently all these rules are sort of isolated and only brought together by evaluating actions on the adventure map and combat maps.

What I am doing for the AI is an algebraic factorization of all game rules, sort of a logical digital unification on the basis of digital building blocks. It is only logical to carry over this concept into the game world much more broadly. That is we will have everything in the game from terrain to hero skills definable, so that we can express anything we like. New skills, creatures, buildings, towns and more created as we see fit. There will be no hard coded limit in the game, only the underlying fundamental defining structure shapes what is possible.

What this means for the game is that modders will be able to add any object with unique magic properties, as far ranging as a powerful overland spell or terrain movement modifications or combat effects, and it will work out of the box.

Now, as some community members have pointed out, we have already a game and maintaining what we have should be our first priority. True. We will start out with the game we have, and add improvements step by step.

Everything we modify will be optional, for the player to choose whether to play an original H5 game or to enable extensions. Ideally this should be possible directly from within the game.

The first steps

Modifiable terrain effects will be an essential part of Eternal Essence. A magic that seeps from the terrain and can affect everything, not only movement cost. How buildings work, creatures, heroes and magic itself.

The game includes already everything that is needed to add additional floors (currently we have a top and underground floor). Access to these floors can be provided by special gates that appear based on progress in the game, and these lead to different planes similar to the concept in the great Planescape Torment. This can be our first experimental battleground, planes of fire or air, or deserts, swamps and others with special magic properties and alterations to the gameplay. The same thing is true for the Hall of Decision victory condition explained in the gameplay section, a limited small duel map once the player reaches a certain progress in the game and elects to challenge the remaining adversaries directly. In general this allows for many more victory conditions and scenario goals.

Having played H5 (and also H3) countless times I somehow find that we have quite the selection of adventure map objects but often tend to visit them in a collector mentality only, hardly thinking of them anymore. What I would like to see is to make the overall progress on the map a bit slower paced, a bit like Age of Wonders, so that the events on the map gain more weight again.

We have a land full of magic. Everything is possible. Who says that opening a treasure chest shouldn't open a magic portal that leads to another plane? This can be a plane where the main hero has to fight for his survival and to return to the main map. Game time on the main map for example can be stopped here or expire twice as fast, which makes every step doubly important. We could also introduce new adventure map buildings, or alter how they work, to have monsters spawn not only in a regular fashion at the beginning of a week.

Magic elements and gameplay

The gameplay section contains a description of true magic essence that describes how the magic undercurrent in the game can be designed. Technically, we can express everything that is already in the game by the magic elements, and on top of it have infinite possibilites to move on.

It's a bit more difficult to see what magic terrain properties offer as it is often with time related things. But a subtle control of pacing is what has so far eluded most strategy games. Civ has different research speeds and such but this is a rather crude approach.

Imagine that the buildings in your castle somehow affect movement and combat on its surrounding lands. With the magic elements and research required for spells, we can easily tie specific resources and access to their mines to the progress of a player.

For example, if research into constructs requires lots of mercury, research into ranged spells lots of gems, and so on, the presence of resources in the castle's vicinity can have a profound impact on the players development. For example, Academy with ample supplies of mercury will be a stronghold that produces golems and other constructs, whereas a castle with access to lots of gems can be masters of wind and air spells. Giving them additional research options for overland spells that alter the terrain, can give each region dynamically evolving characteristic features.

The other thing is that the building resource costs are a bit illogical as they are right now. They are sort of balanced that the player is required to collect different types of resources, but otherwise it makes little sense. With a system of magic elements and more stress on resource types, we can make this much more logical. For example all types of garrison buildings (basic dwellings, walls, town halls etc) can require wood and ore, on the other hand all things related to magic can require gems, all things related to forging and refinement might need crystals. This can strengthen the gameplay a lot, with resources becoming truly valuable and worthy a fight for.

I believe that the designers of H6 set out to tackle the same issues (with illogical building costs) with their choice to reduce the number of resources in the game, and give the remaining ones more differentiation and value. While Ubisoft in general has competent and good producers and designers, they don't tend to be technically minded. So it will be interesting to see in which direction they are taking the game, while we on the other hand can build a magical undercurrent into the game that offers technically infinite possibilities. It's particularly challenging to build an AI for it, but this is where our strength is.


The balancing issues will be handled by the AI itself as opposed to manual tuning. Once the AI reaches the stage at which we have algebraic factorization it will serve as a service to measure the power of each skill, perk, creature, town etc. It will then be straightforward to offer the player options to balance overpowered elements or adapt the overall strategic balance on an adventure map.

The difficulty settings will have more detail and will be maintained for each player on a map. Each player will have an individually adjustable resource handicap, an AI difficulty and AI type (e. g. conqueror, defender). The current beta build has the AI modified to be more cautious, the defender AI type. The first tests are promising. We have already a well working aggressive conqueror type.

I am currently looking into how to port the existing mods best into the game. Most likely this will be a step by step approach. If you have not noted it yet, I am not going to implement small fixes to enable existing mods but instead build into the game native support for the functionality required by the mods. The current plan is to start with the RPE mod and then see how this process works and in which way it needs adapting to enable fast porting of existing mods. Regarding the consistency of mods, as a C++ programmer and going to integrate them into the game proper, I have to make sure that everything works consistently. There is no other way. Actually making the step from scripts to C++ code in the .exe proper, should forestall most of the issues. Every modification will go in cleanly via xml (aka .xdb files).

Still, as a community we will need to approve what goes in, sort of giving the "official" mods some form of seal of approval that will be displayed in the proper place within the game where the player chooses the active extensions. This is doubly important if the modders will have the power as the indicated framework implies (every rule being modifiable).


I still believe that the game is beautiful. My favourite one is watching the blue light on the obelisks wandering down in the time between turns. It's also cool that the animations that talented map makers have added on our new reference maps show in all its beauty. With the AI process now running in the background, the game world with its animations becomes completely unfettered so that the view can be changed freely without delay, stats can be reviewed and so on. There are currently still a couple of hiccups from time to time but that comes from the debug features in the beta.

Nival did a stellar job on the H5 graphics. Interestingly, H5 appeared just around the time when graphics in computer games had sort of matured. Judging from the screenshots, H6 will still add a bit to that with its clean and beautiful designs, but not as much as the step from H3 to H4 or from H4 to H5. Given that the machines get faster all the time, what we have with H5 already will serve us well. The animations and visuals Nival created are outstanding, and can be expected to run fluidly on most PCs that are build today or later. In the end this makes the gameplay all the more important.


In general it's good to have fans who think about the game mechanics and try to explain how it affects the playing experience, because this is the line of thought required to improve the gameplay. It's also good to rationalize it, because many people can feel whether the experience is good or not, but often can't express what's the real cause.

My experience with H3 was that the AI was truly competent, and if I did see that with regular means there was hardly any chance, I went out into the fields and dungeons for treasure hunting, particularly the tomes of earth and air magic for the town portal and dimension door spells. These maps were huge, because they contained tons of stuff in quite a compact area. In contrast much space is consumed in H5 by the 3D landscape itself, and the impossible sized maps are probably not as big as the largest H3 maps. And there was true treasure to be gained in H3, I remember the helm of heavenly enlightenment (+6 for all stats) and nothing beats a message like a group of 20 gold dragons offers to join you. It might indeed seem overpowered, but that is where the fun lies.

But we can do more in H5 now, unpredictable and magic stuff of a random nature. Creatures can have additional magic capabilities that affect the land if they stand there long enough, sort of giving them big boni and maybe even additional objects/guards in the combat arena.

Another possibility is to have the AI track the player's performance, and remember against which creatures he had to struggle to maintain the upper hand. These creatures can easily pop up somewhere unexpected, or as additional guards for neutral stacks (in the combat arena only) with a vengeance. Or as ghosts.


I was recently asked, why I have no team of twenty open source developers who could easily be found for this project.

There is a bit of explaining in order, in the hope that this adds perspective for folks to understand what type of effort we have here.

You might have noted that I never confirmed that I have access to the original sources. We have to respect that it is Ubisoft's property, and that they make the rules. Seeing the effort that Ubisoft has put into H5 and now H6 and what we can do with H5 in the time ahead, it's probably a good thing that the Heroes franchise made it into Ubisoft's hands. That they approved of the H5 AI mod release is a huge step that makes them one of the more proactive community oriented publishers in the world. I have to make every effort that all benefit from this project, that means Ubisoft as the owner of the intellectual property and the fans, if we wish that this project goes on.

Regarding the AI source things are different and that is where the benefit for me is. Of course I have full access to the AI source code because I developed it. My stakeholders tolerate that I do work for the Heroes 5.5 - Eternal Essence project because I can reason that it is a very good reference. That's at least what I hope, as a fan you have sometimes to take a sanity check on your assumptions. For me that is, if a project takes more time of your life than your free time, you need valid reasons to do it. This could be the greater good (raising the level of awareness for the AI quality and what it truly can do for games) or good professional or personal reasons. When I was at university, the professor who taught algebra said that you cannot afford a TV if you want to do your (university) work proper. I learned something for life there (I still do not have a TV). If things take more time than you watching less TV or appearing later to parties, think twice.

This puts a lot of constraints on the AI source. If I were going to give it freely away in some open source fashion, my stakeholders would probably start to sue me for damages, rightfully. This leaves hiring staff to do more work as the only commercially viable option. The regular programmer will take a long time and will not be very good at improving the AI. It's not that the code is not well written or difficult to maintain. Just the opposite is true; I realized that while doing the refactoring of the prototype that was developed 2 1/2 years ago, coming back to that code after this time that it is actually fairly easily readable and maintainable. But the AI for an intricate rule base such as the one H5 has is truly complex and I'd say that this is at the fringe of what can be handled with current technology in terms of complexity. You need very bright people to work on code like this and top talent is difficult to come by.

There is of course the hope that one day Heroes 5.5 - Eternal Essence can stand on its own in terms of quality and ingenuity and its own original gameplay and can attract top talent. But before that a lot needs to happen. The central theme is to develop an AI that can provide an incredible service for all aspects of the game and transform the game to something never witnessed before.

And as said before, I cannot do this effort alone. While there are limitations with the source, for a truly original game we need content, yes map making and campaigns, work on the interfaces and lots of testing. In this way it is indeed a community effort with lots of stuff that needs doing. I have no doubt that it is content that drives games, the AI is just a prerequisite that determines how far the content designers can go.

There were also additional reasons to separate the main .exe process and AI process. First, it allows for a clean separation of the intellectual property, second it is now very easy and straightforward to modify the AI to be used in other games.

There is the hope that we can get development studios interested in using our AI. We can offer a very fast, highly optimized and truly advanced AI that can be used to drive games in a short time frame of a couple months. This is equally true for existing ones as newly developed games.

And one more thing, we have our own game project to use the AI in the pipeline. It's a sci-fi space strategy game that will excel in strategic choices and AI and offer a unique gameplay around the exploration of the galaxy. It has real physics and scientific models. True technology, that players can develop their own space ships, bases and industries as well as research advanced technologies for real akin to what we have proposed here with the magic elements in Eternal Essence. But it's to early to say more about that now.


The order of the items listed roughly represents the order in which they are performed. I try my best to optimize the order that the most relevant issues are tackled first. Still, sometimes it's necessary to add something new to the top of the queue or change the order of items.

Please allow for some time for the tasks to be finished. Most involve analysing the source, determining the entry points for the functions to be modified, a proper analysis of the side effects of proposed changes, probably adaptions to other functions that refer the modified ones, plus implementing debugging code to verify that the modifications work, and multiple test runs and cycles. The code base for a project of this type, including the third party modules, is typically around 10 million lines of source code, that you get an idea of how big the tasks really are, and how involved code analysis often is.

The real upgrades also require a good deal of advanced design work and coding. So, if you see me responding slowly to your community input, please have a bit of patience. The answers will eventually be coming.

On the topic of the combat AI, this will be fully reworked after the adventure map AI is in all respects at least as fit as H3. This includes the town portal and dimension door spells, proper support for border guards and of course bug fixes. It doesn't make much sense to apply quick fixes to the ATB or stupidness of the combat AI prior to that.

I know there are issues with the interface usability and animation start-up times. If there will be time available to address these, it will be done, but they don't have top priority in the current state.






BETA 3.1.9