Journals
- Journals
- Posted by
CPripyatUit
In order to escape from the hell that is medical bureaucracy, I'm making yet another attempt to return to level design. After deciding that all my passion projects are more ambitious than I can currently manage, I figured I'd start with something smaller, and I decided something with challenging puzzles would be a cool map.
Which raises the problem of puzzle design in Half-Life².
The campaign doesn't have that problem: people playing it for the first time are usually new to the game, so the challenge is figuring out what the puzzle actually is and learn the design language: blue barrels float, wooden beams can be broken, gas canisters can detonate, etc. A battery can be carried around to be plugged in.
For any sort of seasoned HL2 player, that's not a challenge. They know the usual puzzles and the design language, and most of the usual puzzles have become mere chores. Carry A to B, stack X on Y, put L under M to lift it up. The "bring A to B" is the most basic form of puzzle and usually results in copious backtracking and can make the player feel like an errand boy.
So what I want are three things:
Elaborate physics puzzles are one possible approach. Those either are too obvious ("a wooden beam supporting a precariously balanced ramp that I could blow up and walk down? how convenient!") or too elaborate ("what am I supposed to do, build a pulley?!").
Circuitry puzzles à la Portal seem more promising, but teaching the player an entirely new design language may take more time than a single map can accomplish.
Number locks are simply trial and error.
I keep thinking about it, but it's hard coming up with enough learnable, challenging, non-menial puzzles to fill a map.
Which raises the problem of puzzle design in Half-Life².
The campaign doesn't have that problem: people playing it for the first time are usually new to the game, so the challenge is figuring out what the puzzle actually is and learn the design language: blue barrels float, wooden beams can be broken, gas canisters can detonate, etc. A battery can be carried around to be plugged in.
For any sort of seasoned HL2 player, that's not a challenge. They know the usual puzzles and the design language, and most of the usual puzzles have become mere chores. Carry A to B, stack X on Y, put L under M to lift it up. The "bring A to B" is the most basic form of puzzle and usually results in copious backtracking and can make the player feel like an errand boy.
So what I want are three things:
- puzzles that are challenging, in that the solution has to be deduced by the player
- puzzles that aren't menial, in that the execution is trivial and doesn't amount to pointless labour
- puzzles that are learnable, in that players don't get frustrated and feel like they were set up to fail
Elaborate physics puzzles are one possible approach. Those either are too obvious ("a wooden beam supporting a precariously balanced ramp that I could blow up and walk down? how convenient!") or too elaborate ("what am I supposed to do, build a pulley?!").
Circuitry puzzles à la Portal seem more promising, but teaching the player an entirely new design language may take more time than a single map can accomplish.
Number locks are simply trial and error.
I keep thinking about it, but it's hard coming up with enough learnable, challenging, non-menial puzzles to fill a map.
That's another thing that's hard: limiting the space the player can move in. It's easy if the level is indoors, all corridors, like Nova Prospekt or the likes: just lock the doors. Nature's limit on level size.
But outdoors? It's hard to put impassable barriers anywhere without making the gameplay feel stilted, awkward, forced. Artificial.
You're in a part of the city where every exit has conveniently been blocked by collapsed buildings? That doesn't make sense, then how did you get in there, and how is Gordon gonna get out there once the screen fades to black?
Or put up traffic cones and riot barriers and cover them in toolsinvis? That's even worse. Gordon can shoot down helicopters, but a traffic cone is enough to stop him? Nah.
Separating the playable environment from the backdrop that's just there to liven up the scenery is pretty hard, at least in my experience. Guiding the player naturally, so they try to go down routes they can't as little as possible and can tell easily where they are supposed to be and where not is even harder. Campaign maps don't have this problem; each level transitions into the next, so the questions of entrance and exit are already answered, and having only the exit ramps blocked while the highway remains open is much more plausible than walling the player in on all sides. But standalone maps have to have a beginning and end that makes sense (at least that's what I want for mine), and that's tough.
I haven't really found a good answer for it yet. Entrances are easier than exits – just some drops that the player can't climb back up –, but it's something I struggle with immensely (and the way my brain works, I can't get to work on the rest of the map until I've solved that problem – it sucks, really).
But outdoors? It's hard to put impassable barriers anywhere without making the gameplay feel stilted, awkward, forced. Artificial.
You're in a part of the city where every exit has conveniently been blocked by collapsed buildings? That doesn't make sense, then how did you get in there, and how is Gordon gonna get out there once the screen fades to black?
Or put up traffic cones and riot barriers and cover them in toolsinvis? That's even worse. Gordon can shoot down helicopters, but a traffic cone is enough to stop him? Nah.
Separating the playable environment from the backdrop that's just there to liven up the scenery is pretty hard, at least in my experience. Guiding the player naturally, so they try to go down routes they can't as little as possible and can tell easily where they are supposed to be and where not is even harder. Campaign maps don't have this problem; each level transitions into the next, so the questions of entrance and exit are already answered, and having only the exit ramps blocked while the highway remains open is much more plausible than walling the player in on all sides. But standalone maps have to have a beginning and end that makes sense (at least that's what I want for mine), and that's tough.
I haven't really found a good answer for it yet. Entrances are easier than exits – just some drops that the player can't climb back up –, but it's something I struggle with immensely (and the way my brain works, I can't get to work on the rest of the map until I've solved that problem – it sucks, really).
Every map has to be about something. There has to be a goal, a reward for pressing buttons, shooting enemies, solving puzzles. At least that's how I feel: I want to feel like I've achieved something beyond the sum of the tasks the map gives me. I've been trying to formulate some goal or other for a map I've been planning, and that's really been the hard part.
A campaign map doesn't have that problem. There's an overarching story to guide me, the goal is "make it through the level to the next plot bit". There's voice acting and cinematics to reward me, Alyx talking to her father, Kleiner broadcasting to the city, etc., all leading to the game's climax. But in a self-contained map, I have to provide the goal. I have to tell the player "this is what you're working towards".
My idea was, okay, there's a Combine installation that needs to be shut down or destroyed. That's easy enough. That's a common goal.
What kind of Combine installation, though? What does it do?
After all my puzzles and combat encounters, the player is gonna barge into that Combine installation and blow it up or flick the off switch or whatever. The actual act will be just as basic as the rest of the map: press a button. So that act of pressing the button needs to have meaning. What did I achieve when I blew up that Combine fortress?
Will this allow the rebels to mount some large offensive? Was it producing weapons? Churning out troops? Jamming communications? Conducting horrible experiments? Housed a superweapon that could obliterate entire city blocks?
And I have to make that decision before I start building it, because its purpose will inform its design. Form follows function. A secret lab full of torture chambers will have a different design than a weapons manufacturing plant or a troop garrison. So I can't just build a generic outpost and pencil in its purpose later, not without major, major revisions that may as well be a complete rebuild.
And coming up with that sort of purpose or goal is hard, harder the more I want the narrative to make sense. The map I'm planning is set in an urban environment that's largely accessible to regular citizens, so any sort of super secret, super access restricted installation is out. I've written myself into a corner before I've laid down the first brush.
Writing it all out like this helps me focus, so that's nice, but the problem doesn't fully go away.
(This is the next logical step after last week's journal about planning.)
A campaign map doesn't have that problem. There's an overarching story to guide me, the goal is "make it through the level to the next plot bit". There's voice acting and cinematics to reward me, Alyx talking to her father, Kleiner broadcasting to the city, etc., all leading to the game's climax. But in a self-contained map, I have to provide the goal. I have to tell the player "this is what you're working towards".
My idea was, okay, there's a Combine installation that needs to be shut down or destroyed. That's easy enough. That's a common goal.
What kind of Combine installation, though? What does it do?
After all my puzzles and combat encounters, the player is gonna barge into that Combine installation and blow it up or flick the off switch or whatever. The actual act will be just as basic as the rest of the map: press a button. So that act of pressing the button needs to have meaning. What did I achieve when I blew up that Combine fortress?
Will this allow the rebels to mount some large offensive? Was it producing weapons? Churning out troops? Jamming communications? Conducting horrible experiments? Housed a superweapon that could obliterate entire city blocks?
And I have to make that decision before I start building it, because its purpose will inform its design. Form follows function. A secret lab full of torture chambers will have a different design than a weapons manufacturing plant or a troop garrison. So I can't just build a generic outpost and pencil in its purpose later, not without major, major revisions that may as well be a complete rebuild.
And coming up with that sort of purpose or goal is hard, harder the more I want the narrative to make sense. The map I'm planning is set in an urban environment that's largely accessible to regular citizens, so any sort of super secret, super access restricted installation is out. I've written myself into a corner before I've laid down the first brush.
Writing it all out like this helps me focus, so that's nice, but the problem doesn't fully go away.
(This is the next logical step after last week's journal about planning.)
Every time I try to make a map, I force myself to try planning it beforehand instead of building away willy-nilly. And every time, sooner or later, I sit in front of a stack of badly hand-drawn maps and am out of ideas. Stuff I draw doesn't fit, doesn't work, I had a better idea afterwards, the proportions are off, the page is too full… you name it.
I tried different approaches. Floor plan design software, for one, though it's tough to find any that is free, works offline or without an account, and lets you save in some useful format.
So last night, I thought: what about writing?
I know I can do that, so what if I wrote descriptions of the maps I wanted to do? It can only get better, compared to drawing and sketching...
I tried different approaches. Floor plan design software, for one, though it's tough to find any that is free, works offline or without an account, and lets you save in some useful format.
So last night, I thought: what about writing?
I know I can do that, so what if I wrote descriptions of the maps I wanted to do? It can only get better, compared to drawing and sketching...
Every few years, after playing around with other games and engines, I touch Source / Half-Life² / Hammer again, and every time I go in with grand ideas about what kind of things I'd like to do. And every time, invariably, I quickly run into the limitations presented by the I/O system.
This time, I've come back to HL2 after doing some stuff with Arma III and BI's Real Virtuality Engine, namely the 3DEN Editor and the SQF scripting language, and the difference is night and day. The two engines almost perfectly complement each other; what Arma is lacking, Half-Life provides, and what Half-Life fails to do, Arma makes a breeze to achieve.
Whenever I was working with 3DEN, I would usually want to make some missions themed around urban warfare (or story-heavy missions), and the bottleneck would usually be level creation. The 3DEN Editor doesn't facilitate much in the way of creating new environments; beyond dropping new buildings or props into the landscape, the maps are immutable. New maps can be made, of course, with different tools, but anything like creating new houses instead of choosing existing models is a fairly big endeavour, and any sort of indoor scenes are usually close to impossible to stage due to the AI's limited-to-nonexistent navigational capabilities. Cinematics of any sort are hard to create, due to a limited set of NPC animations and very imprecise navigation that's largely oriented to squad-level movements on an expansive battlefield. What 3DEN and Real Virtuality excels at, though, is scripting highly-adaptable missions with any number of varied outcomes, custom dialogue, custom AI behaviour, custom gameplay features, the whole nine yards, due to its fairly trivially-learned scripting language and near-unlimited possibilities it offers in customising UI, NPC behaviour, interactions with world objects, and accounting for as many divergent player behaviours as the scripter is willing to anticipate. This makes it possible to make highly nonlinear missions.
Back to Hammer, it's the complete opposite. Building new things is trivial. The geometry tools are right there; short of vertex limitations, nothing stands in the way of creating any building the mapper can imagine. An entire city can be built from scratch, if so desired. Cinematics can be much easier, in certain ways; NPCs can be directed to walk anywhere with a nearly pixel-perfect precision, an extensive system for scripted sequences exists, and animations flow together much more smoothly due to the story-oriented singleplayer origins of the engine, as opposed to the MMO combat-oriented gesture system of Arma III. The navigation mesh is hand-built by the mapper, so indoor navigation can be as good as the author is willing to invest time and effort into; NPCs running into walls, failing to see doors, or being unable to navigate around trivial obstacles is usually a non-occurence. Compared to Arma, whose physics engine is virtually nonexistent, the Source engine allows liberal use of movable props, physics puzzles, destructible levels and objects that are easy to manipulate, and packing custom content with the maps is trivial as well.
Polar opposites. Taken together, the two would form a near-limitless engine.
And here comes the bottleneck, where my HL2 dreams are concerned: no scripting.
Half-Life² is an incredibly linear game. At no point is the player asked to make any sort of choice, at least none that matter beyond throwing cans at cops. There are no alternative routes. There are no side quests, optional objectives, no ways to fail partially without failing entirely. Reduced to its core, Half-Life² is a tube; what goes in at one end must come out the other end, following the only path available.
And that works fine for the game's campaign. It's much more of a ride-along movie with puzzle and combat interludes than an interactive narrative, it doesn't try to be anything else, and it does what it is very well.
But by these engine limitations, attempting to create any sort of nonlinearity in custom maps is very, very difficult.
I've been thinking a lot about projects like that lately, about what I'd like to make once I've reacquainted myself with the engine and tools sufficiently. Drawing on Warren Specter's famous quote about wanting to make games that are "an inch wide and a mile deep, rather than a mile wide and an inch deep", the idea of making small levels, perhaps the size of no more than a city block, that offer a multitude of ways for the player to engage with them, I've thought about ways that could be achieved in Source.
But doing that with the I/O system? A map that needs to keep track of, and adapt to, countless variables and changes in their values? It seems impossible, secondarily due to the entity limit, but primarily due to the sheer workload and the ever-increasing possibilities for increasingly hard to track errors to occur the more complex the I/O network becomes.
The one beacon of hope here is Mapbase, which to my knowledge implements VScript, something I have yet to learn. I've always been a little intimidated by Mapbase and largely dropped out of RTSL mapping tournaments when they switched to routinely mandating Mapbase, simply because at the time I couldn't figure out how to install it and was just glad Hammer worked for me at all after an extended stint with perpetually broken game configurations for mods (maps not updating, maps failing to compile, etc. etc.). I ought to take another look at that some time.
Idk. I just needed to get all that out. I'd still love to make less linear, more freely approachable HL2 maps and try to bring more of that New Vegas, Deus Ex, Mankind Divided feel over to the Source Engine. Of course the obvious solution would be to simply switch to a game and engine that offer more support for the kind of maps I want to make, but it's a labour of love: I love Source / Half-Life², and I want to make and play the kinds of maps I dream of in the game I love playing so much I'm still doing it in 2024.
P.S: Two additions I forgot earlier:
One, perhaps the best way to compare working with NPCs in Arma III vs. Half-Life² is this: in Arma, it's easily possible to create NPC behaviour that will dynamically, react to certain map/story events, in variable order (within the boundaries of what you prepared), but very difficult to exercise precise control over an NPC. In Half-Life, it's easily possible to manipulate an NPC with minute detail, but very hard or often impossible to set up any sort of behaviour that will continue working once you take your hands off the reins.
Two: one more limitation of HL2 is that custom dialogue or custom scripted sequences are very hard to implement purely on a map basis, because they require things like recompiling scenes.image, editing game-wide files like language files for subtitles, etc. In Arma III, custom dialogue relies on its own, separate subtitle and script files, much like HL2 custom materials and models can be packaged into the map.
This time, I've come back to HL2 after doing some stuff with Arma III and BI's Real Virtuality Engine, namely the 3DEN Editor and the SQF scripting language, and the difference is night and day. The two engines almost perfectly complement each other; what Arma is lacking, Half-Life provides, and what Half-Life fails to do, Arma makes a breeze to achieve.
Whenever I was working with 3DEN, I would usually want to make some missions themed around urban warfare (or story-heavy missions), and the bottleneck would usually be level creation. The 3DEN Editor doesn't facilitate much in the way of creating new environments; beyond dropping new buildings or props into the landscape, the maps are immutable. New maps can be made, of course, with different tools, but anything like creating new houses instead of choosing existing models is a fairly big endeavour, and any sort of indoor scenes are usually close to impossible to stage due to the AI's limited-to-nonexistent navigational capabilities. Cinematics of any sort are hard to create, due to a limited set of NPC animations and very imprecise navigation that's largely oriented to squad-level movements on an expansive battlefield. What 3DEN and Real Virtuality excels at, though, is scripting highly-adaptable missions with any number of varied outcomes, custom dialogue, custom AI behaviour, custom gameplay features, the whole nine yards, due to its fairly trivially-learned scripting language and near-unlimited possibilities it offers in customising UI, NPC behaviour, interactions with world objects, and accounting for as many divergent player behaviours as the scripter is willing to anticipate. This makes it possible to make highly nonlinear missions.
Back to Hammer, it's the complete opposite. Building new things is trivial. The geometry tools are right there; short of vertex limitations, nothing stands in the way of creating any building the mapper can imagine. An entire city can be built from scratch, if so desired. Cinematics can be much easier, in certain ways; NPCs can be directed to walk anywhere with a nearly pixel-perfect precision, an extensive system for scripted sequences exists, and animations flow together much more smoothly due to the story-oriented singleplayer origins of the engine, as opposed to the MMO combat-oriented gesture system of Arma III. The navigation mesh is hand-built by the mapper, so indoor navigation can be as good as the author is willing to invest time and effort into; NPCs running into walls, failing to see doors, or being unable to navigate around trivial obstacles is usually a non-occurence. Compared to Arma, whose physics engine is virtually nonexistent, the Source engine allows liberal use of movable props, physics puzzles, destructible levels and objects that are easy to manipulate, and packing custom content with the maps is trivial as well.
Polar opposites. Taken together, the two would form a near-limitless engine.
And here comes the bottleneck, where my HL2 dreams are concerned: no scripting.
Half-Life² is an incredibly linear game. At no point is the player asked to make any sort of choice, at least none that matter beyond throwing cans at cops. There are no alternative routes. There are no side quests, optional objectives, no ways to fail partially without failing entirely. Reduced to its core, Half-Life² is a tube; what goes in at one end must come out the other end, following the only path available.
And that works fine for the game's campaign. It's much more of a ride-along movie with puzzle and combat interludes than an interactive narrative, it doesn't try to be anything else, and it does what it is very well.
But by these engine limitations, attempting to create any sort of nonlinearity in custom maps is very, very difficult.
I've been thinking a lot about projects like that lately, about what I'd like to make once I've reacquainted myself with the engine and tools sufficiently. Drawing on Warren Specter's famous quote about wanting to make games that are "an inch wide and a mile deep, rather than a mile wide and an inch deep", the idea of making small levels, perhaps the size of no more than a city block, that offer a multitude of ways for the player to engage with them, I've thought about ways that could be achieved in Source.
But doing that with the I/O system? A map that needs to keep track of, and adapt to, countless variables and changes in their values? It seems impossible, secondarily due to the entity limit, but primarily due to the sheer workload and the ever-increasing possibilities for increasingly hard to track errors to occur the more complex the I/O network becomes.
The one beacon of hope here is Mapbase, which to my knowledge implements VScript, something I have yet to learn. I've always been a little intimidated by Mapbase and largely dropped out of RTSL mapping tournaments when they switched to routinely mandating Mapbase, simply because at the time I couldn't figure out how to install it and was just glad Hammer worked for me at all after an extended stint with perpetually broken game configurations for mods (maps not updating, maps failing to compile, etc. etc.). I ought to take another look at that some time.
Idk. I just needed to get all that out. I'd still love to make less linear, more freely approachable HL2 maps and try to bring more of that New Vegas, Deus Ex, Mankind Divided feel over to the Source Engine. Of course the obvious solution would be to simply switch to a game and engine that offer more support for the kind of maps I want to make, but it's a labour of love: I love Source / Half-Life², and I want to make and play the kinds of maps I dream of in the game I love playing so much I'm still doing it in 2024.
P.S: Two additions I forgot earlier:
One, perhaps the best way to compare working with NPCs in Arma III vs. Half-Life² is this: in Arma, it's easily possible to create NPC behaviour that will dynamically, react to certain map/story events, in variable order (within the boundaries of what you prepared), but very difficult to exercise precise control over an NPC. In Half-Life, it's easily possible to manipulate an NPC with minute detail, but very hard or often impossible to set up any sort of behaviour that will continue working once you take your hands off the reins.
Two: one more limitation of HL2 is that custom dialogue or custom scripted sequences are very hard to implement purely on a map basis, because they require things like recompiling scenes.image, editing game-wide files like language files for subtitles, etc. In Arma III, custom dialogue relies on its own, separate subtitle and script files, much like HL2 custom materials and models can be packaged into the map.
So I had a train of thought the other day, while playing Borderlands 2, a bit out of the blue, which I kinda wanna write down and since I wouldn't know where else, I'll just do it here.
Basically what I thought was - Half-Life 2 plus episodes isn't really big on optional objectives or the likes. In fact, no Valve game I've played is. Side quests aren't really a thing (Lambda caches don't count) and being very linear is pretty much a staple of the whole series. And I wondered what it'd be like to try and build some optional stuff into maps... like, go make this boom, if you're willing to take a detour, you can make it easier, I dunno, blow up Combine machinery and if you take a detour first to take out their radio relay comm thingy, it'll be easier because they won't send reinforcements. Stuff like that. Or, for instance, if along the way in Nova Prospekt, you had the opportunity to liberate incarcerated rebels and vorts, who would subsequently serve as allies during combat situations. Etc etc.
Now that's all kinda heavily influenced by my current playing through Metal Gear Solid V, where I usually, before sneaking into an outpost, start by planting explosives around their comms, power supplies, anti air guns etc, but also some through Borderlands 2 - if you bother with the sidequest of getting the power back online at station whatsitcalled, it'll unlock the shops, garage and spawn point there. Etc. I figured stuff like that would be pretty cool in Half-Life 2. Or (didn't think about that as much) Black Mesa. Fulfill objective X in area controlled by grunts, if you turn on the power for the conveyor system, you'll get supplies from the conveyor belts. Or take some time for secondary objective this and that to turn off the AC so you can sneak through the then-unlocked vents. Etc etc.
So yeah. Bit of an odd rant here... just had to write this down to revisit once I'll have time and energy to maybe try and make use of that, at some point in the distant future... do with that as you will, I'd love to hear some thoughts from yall maybe
Basically what I thought was - Half-Life 2 plus episodes isn't really big on optional objectives or the likes. In fact, no Valve game I've played is. Side quests aren't really a thing (Lambda caches don't count) and being very linear is pretty much a staple of the whole series. And I wondered what it'd be like to try and build some optional stuff into maps... like, go make this boom, if you're willing to take a detour, you can make it easier, I dunno, blow up Combine machinery and if you take a detour first to take out their radio relay comm thingy, it'll be easier because they won't send reinforcements. Stuff like that. Or, for instance, if along the way in Nova Prospekt, you had the opportunity to liberate incarcerated rebels and vorts, who would subsequently serve as allies during combat situations. Etc etc.
Now that's all kinda heavily influenced by my current playing through Metal Gear Solid V, where I usually, before sneaking into an outpost, start by planting explosives around their comms, power supplies, anti air guns etc, but also some through Borderlands 2 - if you bother with the sidequest of getting the power back online at station whatsitcalled, it'll unlock the shops, garage and spawn point there. Etc. I figured stuff like that would be pretty cool in Half-Life 2. Or (didn't think about that as much) Black Mesa. Fulfill objective X in area controlled by grunts, if you turn on the power for the conveyor system, you'll get supplies from the conveyor belts. Or take some time for secondary objective this and that to turn off the AC so you can sneak through the then-unlocked vents. Etc etc.
So yeah. Bit of an odd rant here... just had to write this down to revisit once I'll have time and energy to maybe try and make use of that, at some point in the distant future... do with that as you will, I'd love to hear some thoughts from yall maybe