How difficult would a fanmade modular open-source engine for MTW style campaign be?

[Vantek]

I was just thinking, that the MTW campaign itself should be really pretty easy to simulate, shouldn't it? Even graphically. I mean all you have is a 2D map with 2D icons, and the gameplay consists of completely trivial calculations. You could also build the engine step by step, starting out with something crude that just simulates the game mechanics without looking particularly good. Then add the capability to replicate MTW graphics and sound (probably using original MTW data files). And then maybe ultimately if the campaign is working so smoothly that people see the potential, finally add a module for 3D tactical battles.

Seeing as there are various fan projects around for various other games that seem much harder (TA Spring comes to mind) and MTW community has been pretty productive, why has this never been made? Has there even been serious talk about it? Am I missing something about why it would be harder than it seems?

Of course now, with the community fading away, at least for M1TW this will never happen, but just wondering.

[caravel]

Reverse engineering is possible, but extremely difficult and time consuming. It would require a very experienced C and assembly programmer - highly experienced in disassembly and decompilation - even for just the campaign map.

Most of the fan made open source engines have been for fairly simple games (as with OpenTTD) or have come about through the developer actually releasing the source code. Once source code is released, the whole engine can be rebuilt as required and ported to different platforms (i.e. MacOS, Linux, BSD etc).

[drone]

I don't think he is talking about reverse engineering. Most of the game mechanics, especially on the campaign map, have been dissected over the years and could probably be implemented from scratch in a new engine. The kill probability was published in the official guide. Most of the unknowns would be AI-related I imagine. The battle engine would take the most time to recreate, even with 2D sprites you still have terrain, weather, control, and missile physics/collisions to deal with.

Getting the source code would be awesome. Not going to happen, but still awesome. And for the same reasons we aren't getting the code, I imagine an open-source project to recreate it would be faced with an army of high-valour lawyers.

[Garnier]

I imagine an open-source project to recreate it would be faced with an army of high-valour lawyers.

:D

I am doing something interesting here. It's not MTW, but it is a risk-style campaign game with online multiplayer, which has a relatively easy map/scenario module system. I haven't considered making it open source. But I'll be looking for beta players at some point. ;)

[Vantek]

Yeah I'm not talking about reverse engineering, just building a new engine from scratch (but one that could use original MTW graphic files and sound).

Most of the fan made open source engines have been for fairly simple games (as with OpenTTD) or have come about through the developer actually releasing the source code.

Most? Maybe. But there are some fanmade open source engines (see below) that are in no way less complex than a full blown commercial game. And the campaign part of MTW seems actually really simple to me. It's the 3D battle engine that'd be a bit tough to recreate. But you could make the 2D campaign and enjoy it even if there never was a 3D battle engine.

And for the same reasons we aren't getting the code, I imagine an open-source project to recreate it would be faced with an army of high-valour lawyers.

I don't think that's true. First of all I'm not even sure they have a case. Secondly even if they did I doubt they care about the game enough to waste thousands upon thousands of florins to pester a few poor peasants.

For a couple of examples:

http://freeempires.yoll.net/GG/index.html
http://en.wikipedia.org/wiki/0_A.D._%28game%29
http://en.wikipedia.org/wiki/Ta_spring

All these projects started out as direct mimicking of a certain game or engine. As for Spring, it directly used Total Annihilation game files (but it wasn't distributed with Spring obviously, you had to have TA beforehand to be able to play TA in Spring). I don't think there ever were any legal obstacles to something like this being made for MTW. Guess the community just didn't happen to have enough of the kind of people interested in and capable of doing it. But I'd find it really surprising if it has never been even discussed much.

[Gaiseric]

How do you build a game engine from scratch? I imagine it is very difficult for complex games that would aim to be like MTW. I defiantly want CA to have some competition in the exact same genre as the TW series. The only game that I can think of that came close to being a TW competitor was Imperial Glory. Imperial Glory with mods beats ETW any day of the week, although it didnt do very well in sales.

Maybe some of these TW mimics would be better to copy or get the source code from. Not very many of them did well at the market place, but that may make them easier to access. With proper attention and a few tweaks, they could become great games.

[drone]

As an old school programmer, the campaign map, with it's turn based action, reads like an exercise in data structures. There is nothing in real-time, everything is menu-based apart from the music. AI decisions, troop movement, battle results, incomes, etc. are computed during the end-year phase. The first step would be writing file parsers to read in the data files like the startpos, unit and build prod, LUK map, etc. Then you would need the campaign start menu (chose campaign, faction, difficulty, era), fill in the province/faction data and start points. Most of the campaign game mechanics are fairly well known, so the end-turn calculator would be easy. The GUI, campaign map graphics, unit and building menus add more complexity (user input always does), but again nothing is real-time. The AI decisions for recruitment, building, troop movement, and diplomacy would be the most difficult task, which is usually why the AI stinks in most strategic video games.

The battle map, egads. I don't even want to think about that.

[Gaiseric]

You make it sound easy, although half of the stuff you just said went right over my head.

Of all the TW games I'd say that MTW AI is the most competent. I think a lot of that has to do with the style of campaign map used in MTW compared to the newer versions. Its probably easier to program the AI for the 2D map too. A 3D battle map will be very difficult to program the AI for. Hats off to CA. Up to ETW, battles were their forte. ETW however had a great strategic\campaign map. Someday someone will get the combination right, but they will have to be a computer wizz, borderline autistic, for the programming needed.

[caravel]

Personally I think any open source engine for STW/MTW should do away with the perspective plane and sprite scaling approach (the 3D appearance of the map and the scaling of the map pieces) and just go for a basic flat top down map with draggable pieces (yes something along the lines of what Garnier has developed or the paradox games). This should be fully playable without the battle map component, i.e. if the battle map is not present the option to fight the battle yourself is greyed out - autoresolve being the only option. The campaign map game need only run windowed, but apart from appearances it should function exactly like the STW/MTW campaign. The engine could still use the same MTW map textures, pieces and other graphics and sounds - so in essence kind of like a flattened MTW with no DirectDraw, running only in a window (I would put together a mock up if I had the game installed and if I wasn't at this moment at work).

The battlemap is a whole other project. That's the real "total war" and where most of the code probably is. There may be existing open source 3D engines that could be adapted to recreate it - all that would be needed would be the textures and models from MTW, which the player would need to buy. In these types of ports a lot of improvements are possible, such as 3D models on the battlefield instead of sprites.

[Gaiseric]

What would be really cool, but maybe not possible for a source edited game, would be to have a campaign map like you described above, with a similar number of large provinces like MTW has. Instead of battle taking place on that map, the game would zoom in and have each province hexed out, with appropriate terrain features. Armies will appear on the hexes of the province they are invading from and have to capture objectives, cross rivers, gain terrain advantages, and defeat enemy armies occupying the hexes. A battlemap similar to some of the SSI games that were released in the early 90's would be very cool and probably not as hard to develop as the 3d battlemap of TW games. It would be a great complement to those types of games.

I like games as you described above, but they always seem to end in a slugging match between the AI and the player. In these slugging matches the player has no control over what happens militarily, and is usually doomed to defeat if they have a smaller army. A zoomed in province map with hexes, or even something with thousands of provinces like Hearts of Iron, would give the player more tactical control. This might mean a lot more programming, but it is the direction I hope to see some of those games take.

[Vantek]

As an old school programmer, the campaign map, with it's turn based action, reads like an exercise in data structures. There is nothing in real-time, everything is menu-based apart from the music. AI decisions, troop movement, battle results, incomes, etc. are computed during the end-year phase. The first step would be writing file parsers to read in the data files like the startpos, unit and build prod, LUK map, etc. Then you would need the campaign start menu (chose campaign, faction, difficulty, era), fill in the province/faction data and start points. Most of the campaign game mechanics are fairly well known, so the end-turn calculator would be easy. The GUI, campaign map graphics, unit and building menus add more complexity (user input always does), but again nothing is real-time. The AI decisions for recruitment, building, troop movement, and diplomacy would be the most difficult task, which is usually why the AI stinks in most strategic video games.

The battle map, egads. I don't even want to think about that.

This is exactly what I feel is the case. Given that the campaign map portion of the game is honestly pretty damn easy to program, and actually really enjoyable by itself alone, without the battle map portion of the game... Isn't it strange that recreating it has never been even discussed, let alone created, considering how much amazing work MTW community has already proven to be capable of?

And once you already have the campaign map portion of the game fully recreated as an open source project... Recreating the battle map will start to look a lot less impossible. You could begin with something much more crude than the original MTW. At this point, I think the project would gain a ton of interest from fans who were perhaps fed up with certain bugs or design flaws or compatibility issues. This would attract more and more potential developers... And who knows, eventually there could be a battle map even better than the original.

Or, well. All of this would have been true if I had said it 5-8 years ago. Now it's just too late. Although as another favourite game of mine, Seven Kingdoms, reminds us, there is always hope... (cliffs: the game's source code was released to the fans >10 years after its release, and is now being worked on)

Personally I think any open source engine for STW/MTW should do away with the perspective plane and sprite scaling approach (the 3D appearance of the map and the scaling of the map pieces) and just go for a basic flat top down map with draggable pieces

I agree that it should start out with less fancy graphics with the intention of recreating the original later, once you already have something functional in place. But are you advocating a basic flat top-down map as the ultimate goal? If so, why?