Articles | Apex Design (2015)

In this article I am going to discuss the various elements behind my Apex design, the core concept that drove the various sc2 campaign attempts I made dating 2012 to late 2014.

The goal of this article is to explore design processes I have matured over the years, establish a foundation of thinking for future projects, and expose a portion of my core game design theology. It is in part self-documentation for the future when I undoubtedly forget just about everything I did the last few years because my memory is like that of a fruit fly.


Throughout my many years creating total conversions for Brood War I often considered campaign concepts but never attempted them. Wc3 saw my entry into attempting campaign projects, but limitations of the game's performance and AI held me back. Starcraft 2 would see a repeat, with the severe AI limitations holding back the Apex line as a whole and ultimately leading to my retirement from modding.

However, throughout my time in all three games, I came to harbor a deeper understanding of what it was I wanted out of a campaign. Although I will not likely ever return to the RTS market as a player or a modder, it may benefit others to understand how I was designing these projects, what I placed value in, and why they failed. At the very least this article serves as documentation for future reference on what I spent so many years fighting to create.

Warcraft 3 - An Era of Silent Dreams

Warcraft 3, as an RTS, was a significant failure compared to Brood War. Its writing forever tarnished by the involvement of individuals like Metzen, and its gameplay designed both to reduce the barrier of entry for casuals and to reduce the skill cap for gamers. However, as far as custom content was concerned, that failure and the subsequent conflicts of interest created a void which many custom content creators attempted to fill. As one might perceive this situation, adversity leads to adaptation. Ultimately Icefrog found the magic formula with Dota but to this day the custom content in Warcraft 3 continues to thrive. Warcraft 3 found a balance in a simplistic and intelligently designed multiplayer platform that only continued to flourish in the face of the comparatively hostile titles Blizzard released later on in time. Testimony of this is echoed by the Tower Defense and Dota clones to flood the indie market thereafter.

Warcraft 3 had a larger ratio of custom content produced for it than Brood War, but ultimately it had fewer large-scale projects and the ratio of objectively good projects actually dropped, a side effect of the kind of people it attracted - people who were adverse to the concept of mods and quality, completed projects. What it fostered instead was a host of pet projects and derivatives of various crossovers and spinoffs, all made manifest by its ease of entry and abundant community tools. Like Brood War, however, Warcraft 3 was troubled with hardcoded limits left and right. The advent of various JASS systems helped offset certain mechanical limitations, leading to the creation of things like pseudo physics engines, but fundamental problems, especially regarding computer AI, held back the game as a whole.

Campaigns for Warcraft 3 were exempt from the sole major AI foray I am aware of - AMAI. AMAI managed to solve many issues with the AI in wc3, but only in a melee environment and with very specific and hacky solutions. As advanced as AMAI was it couldn't solve the fundamental issues. Although AMAI had a system built into it to support mods and new units I, personally, found it very convoluted and overcomplicated to work with. So I cannot say how well it would have helped a campaign.

Wc3's AI proved to be horrific at managing unit pathing and the captain system had boundless troubles. The dreaded half-hour long pathing time between base to enemy only to destroy a single farm and then walk back home, assuming they could path the map at all, was the killer of most of my projects. Not to mention the performance. Generating AI through the editor's built-in AI "editor" was an excellent way to prompt a severe stall when an AI was attacked for some reason. RCX had moderately more success doing it the old JASS way, but, ultimately, one of the only success stories.

Without JASS and anyone skilled enough to tackle the problem my campaign concepts in wc3 remained concepts. Criticism could be rightly pointed towards my inability to learn JASS as it is plausible that certain solutions might have been possible given enough research and skill. However, more aptly exposed is my lack of programming skills as a whole. Had I been a programmer I could have fixed the issues with Brood War by reverse engineering it, and never needed go to warcraft 3 or other titles in the first place. After all, Warcraft 3, once you exclude triggers/JASS, has no advantage over Brood War as a custom content platform for an RTS. Alas, such things were not to be, and I made grand effort over many years to find solutions to my problems with my own devices, however limited they were.

Warcraft 3 summed up in one image.

While I experimented with multiple total conversions, I eventually settled on a dota-styled hybrid of melee and hero gameplay. This was iterated in three portions - Eternity Mark 1-3 - and the Event Horizon project (unrelated to the movie, incidentally). They formed the early concepts of what would eventually formulate the Living World and, finally, the Apex concept. It was through these experiments that I acquired a strong disdain for arbitrary unit spawns and scripted responses, and grew to prefer the living dynamic of an active melee environment. In time, once I fully settled into professional game analysis, reviewing, and technical design, I would eventually learn to formulate my ideas into words. Communication was never my strong suit and, indeed, my time with warcraft 3 illustrated I was unsuitable for public interaction on any level. I did, however, learn what I wanted out of an RTS or RTS/ARPG hybrid. I wanted an active melee environment to build my campaign missions around.

The idea behind the active melee environment is that it becomes tamper-resistant during play by handing responses to an actual AI, and feels more immersive as a direct result. A world in which all of the elements actually do something by deriving them off of existing melee gameplay is more interactive and more meaningful than a corridor-bound series of scripts that you must execute or follow in sequence. It is the same reason why QTE's were created to reduce depth of action games and allow non-gamers to associate themselves with what would otherwise be skill testing media. Assigning the actions of your enemies or allies to timed unit spawns and arbitrary events removes a great deal of gameplay that could be explored, and a great deal of world building that could be actuated through perception alone. After all, AI is the driving force of your world's elements. Once we step back and restore those living elements we create a far more dynamic if not difficult to balance environment.

While I came into sc2 with a very strong idea of what I wanted out of it. The path would be rife with challenges both in terms of design and in terms of game engine support. The very thing I demanded out of the game was one of the most difficult to get in a Blizzard title - functional AI.

Apex saw iterations A to F starting in 2012 and concluding in 2014. Most major iterations saw example gameplay and repeated conceptual pushes into terrain, units, graphics, sound, and actual mission creation. In a way these efforts are a near exact mirror of those I had in wc3, because they all met their end to the exact same problems I had earlier in my modding years.

Building concepts, figuring out limitations, testing performance, and otherwise establishing an idea of how viable various ideas are going to be is critical for my projects. Starcraft 2 failed most of my tests, but, until we got deep into AI work, I thought I could pull a campaign off regardless. With tactical AI, Starcraft 2 deceptively appeared like it was more advanced than Warcraft 3 in this arena.

Almost all variants of Apex also saw a rewrite to the world it was building upon. Apex F, its final incarnation, saw a major revision of all previous script and introduction of previous writing ventures, including elements related to CC community members. As a lighthearted and extremely politically incorrect project it was ultimately intended to be built for a very specific and niche audience. I didn't release my work publicly, so I had a very clear set of design goals in mind and needn't care about how other people would adapt to them.

Apex - Core Gameplay Concept

"I want something extremely difficult that can handle any playstyle and simply performs like a modded melee game, but with emphasis on exposing an entirely new world through the highly effective process of having the player learn and adapt through it."

I am a cave-styled designer. I build projects that have extremely high skill requirements for entry. I build projects I personally find extremely difficult to play. I intended for the campaign to require a minimum of C to C+ objective iccup skill in BW for the tutorial missions alone. At the time of this writing, that was Master to GM level for Sc2. But as it is a comparatively easier game, most players would find the entry challenging but manageable. However, the difficulty would scale linearly per mission, and by 0x05 become extremely challenging in a manner only practicing the missions would help you improve upon. Each mission would contain a minimum of an hour of B&D gameplay and use virtually no ingame cinematics, no prompts or out of dialogue notices, and very little major non-unit dependant triggering.

Fueling the core gameplay was a multi-pronged penetration into total conversion territory. Clearly, Starcraft 2's existing gameplay didn't align with my goal. This wouldn't be some small sweep on numbers, either. I intended to take the designs as deep as I could within the limits of my assets.


I shall list many of the bulletpoints regarding my approach to the campaign in terms of subjects. I believe addressing each major aspect this way can help me communicate what my goals were at the time.


  • 1 - All vanilla content was to be rebuilt from scratch.

Although Apex's designs often relied on very heavy modified Starcraft lore to fill the unfortunate gap in my artistic abilities, I had absolutely no qualms about ditching nearly everything in Starcraft 2 as it was. We were, for all intents and purposes, taking two race ideas, rewriting them, and then going from scratch elsewhere.

I walked into Apex with an exceptional history of not only observing and researching Brood War professional play, but also having applied that understanding to various projects in the past, even the balancing numbers for Retribution's test recordings. I feel that by having pierced the veil between psychology and play that I can build unit design beyond simple numbers in a reliable and dependable manner. To that end my revision for Starcraft 2's existing content was going to see something entirely new rise from its ashes.

My goal was always to step back to Brood War numbers and pacing then evolve outwards - in a very different direction than the one sc2 took. Comparatively fewer "spells" with greater emphasis on how units felt and on their mechanical behavior. I expanded the tech trees well beyond their Brood War limits, taking the responsibility of the diversity and power balancing nightmare that would rise as a result, to suit the lengthy campaign playtime. How the units and their weapons behaved ultimately was the character of the various elements, not necessarily how big their numbers were. With that mindset I could keep the numbers fairly tame between units.

Another major change from Starcraft 2 that may not be readily apparent in its meaning is collision design. Starcraft 2 suffers from excessive AoE distribution and the clustering behavior of the pathing compounds this issue. By increasing unit collision I indirectly reduce the effectiveness of AoE damage and also prolong fight length. This change alone has massive and far-reaching implications, especially when considered in the context of a single-player campaign. This did, however, allow me to once again make artillery and heavy units strong. They were no longer benefiting so heavily from innate clustering so their individual effectiveness could be returned to them. Again, the high tier depth meant I could isolate siege units appropriately and have them favor macro and economic play over some kind of early cheese.

Another thing I considered removing was hitscan, especially for units like Scourge who, with hitscan, are ludicrously overpowered. This lent unfavorably to Siege Tanks who, with retuned stats, are extremely powerful due to avoiding overkill. In a way the tanks in Apex F were double dipping the benefits of sc2's "advanced" hitscan and sc1's "overpowered" numbers. Here is where the designs both converged and then heavily diverged. I ended up giving tanks a much longer cooldown but ultimately I planned to remove hitscan from nearly everything possible. I didn't want the game playing itself except where it really made sense.

Universally units moved slightly slower, had slightly less range, and did less damage per hit while having longer cooldowns. I avoided bonus damage that was any higher than 20% of base when possible. Most units had little or no bonus damage. AoE falloff distance was another thing to take note of. In Brood War I commonly gave units with large AoE fairly deep, curvy falloff, so they could do mild or minor damage at a distance but required severe clustering or precise shots to lay major damage. The same was with units that had inaccurate (e.g. valkyrie) shots. The Artillery in Apex F was super inaccurate and required direct impacts to dealing major damage. The inaccuracy was the determining power balance of the structures, whereas in other player controllable units I would try to avoid inaccuracy so the control remained in the hands of the players. Suffice to say, targeted nukes would have needed careful testing and balancing.

As far as unit control was concerned I didn't want to tie behavior into buttons beyond things like deploy and siege mode. Siege mode was applied rigorously throughout various terran tech trees (as well as transform for certain Gundams), but spells like Strike Cannons and Yamato Cannon were extremely rare. Given that this was a campaign, I had a great more lenience on how I chose to expose ability-based powers and spellcasting. In general I wanted abilities like these to be void on anything that didn't directly make use of it. Thor strike cannons became a siege mode ability for artillery, Yamato Gun ceased to exist, and most new units were void of such button pushing power spikes. The goal here was to avoid "fire and forget" thinking as much as possible, and to broaden the micro demand of control while lessening the need to arbitrarily push buttons on cooldown to maximize efficiency.

The major exception here were races like Armageddon and Protoss who prominently featured spellcasters. As a direct result their skill cap to play rose dramatically, since power budgets and costs were factored heavily into the utility creep of having a sudden explosion of on-demand power budget versus active power budget (psi storm versus marine cluster, for example). These races were then introduced later into the campaign so players had a chance to adapt to the much higher unit control demands over vanilla, else they would never be able to handle both tasks, especially since most spells were utility-oriented. Armageddon in particular was speculated to be amongst the most difficult to play races of any RTS ever created merely due to the micromanagement requirements, with many units featuring 4+ abilities each, as they were normally delegated to the AI. This was, however, the natural answer to the demands for playable Armageddon from Brood War's era. I did not, however, expect to be able to make the bosses manageable for human players due to their exceptional complexity. This meant that the human playable Armageddon race would have to function extremely differently from the computer controlled one.


  • 2 - Upgrades were removed.

In melee upgrades have huge implications but in a campaign they are commonly only used to weight stats against you early on before they level out and then are completely forgotten about. They then become an arbitrary action made necessary by stat inflation and serve no actual gameplay purpose, and therefore have no place in a campaign. I was still on the fence about tech researches like Siege Mode, because I had no intention of invalidating later mission decision making by making researches campaign-wide. I was increasing the depth and cost of tech trees very significantly to compensate for the lack of upgrades and the addition of many new units and structures, but techs remained an unsolved element all the way until the project's ultimate death. Ultimately, the smart decision seemed to be to add a third resource into the economy and tech tree loadout and then ditch tech upgrades entirely. But this would need extensive testing to ensure it felt good to perform as messing with Starcraft's economy system is an extremely slippery slope.

The hots zerg evolutions and wangs of liberty out of mission upgrades were red flags of bad design at the fundamental level. They were poorly thought out and not at all balanced, and had no actual gameplay interaction. They existed solely as inconsequential filler to give the illusion of depth in an environment where no depth could exist. They were there to distract people from the shallow design of a campaign intended for kindergarten-level thinking, accompanied by per-mission gimmicks and gotchas as opposed to computer AI taking the wheel. It is no wonder I would run into so many problems trying to adopt a polar opposite design paradigm.

A side effect of removing upgrades is balancing armor becomes significantly more important. Every individual armor value has tremendous implications, especially since many units attacked in a volley or did incremental damage. Additionally, those large AoE's with different falloffs would also interact with this in potentially unpredictable ways, as the wide radii with small damage values could get soaked by high armor values and diminish AoE value. Luckily, the original AO helped me out in knowing how this would pan out. It offered an alternative means to balance small unit spam (M&M in WoL) versus big unit spam (predictably an issue in a campaign with high tier depth and long mission length).


  • 3 - New mechanics

When I think of mechanics I think of something the player interacts with on a game level scale that is not serving merely as grind. So, obviously, I was removing things like mules and larva inject. The new mechanics I was adding were mostly to fill the gap left behind by the removal of upgrades and helped fill the shoes that grew ever larger by the multi-base gameplay my map design was centric around.

Taking a nod from Supreme Commander, Apex F saw the introduction of Artillery and Tactical Missiles, along with SDI defenses respective to each type. Unfortunately, while the systems were functional, I never got around to making a mission that could demonstrate how well they would actually add to the gameplay, since the AI's performance rendered even the most mundane of maps unplayably laggy.

The idea was to replicate Artillery wars from Supreme Commander while avoiding the many issues shields introduced. By keeping the artillery and tactical mechanics simple I could also ensure the AI would use them reasonably intelligently. Artillery was highly inaccurate and offered even more inaccurate "blind fire" attack modes at a longer range than normal. Tactical Missiles entirely replaced nukes and could be launched map-wide, but traveled like projectiles and could be shot down by special defenses.

Getting the missile travel down was unexpectedly irritating, even with a veteran Sc2 mapper at my side.

Of course, this was just simple Terran stuff. I had to figure out how other races were going to work. I planned to give Protoss Wizard Towers similar to those in Age of Wonders 2. The lore offered reasoning behind this, it would just be a matter of making them balanced and meaningful. Since my Protoss were amongst the most heavily redesigned of the races I planned to make them lean more heavily in sc1's direction of cost-power weighting, and their wizard towers would fill the role of global presence with similarly-designed spells to those in AoW2.

The abundance of new unit designs that were impossible in sc1 and rendered possible by sc2 allowed me to make many unique units that didn't rely on spells to make them stand out from one another. Battlecruisers, for example, had missiles that could fire while moving and had considerable range but would split to up to 4 targets, and did minimal damage. Their railguns, replacing their standard weapons, fired in an extended volley that could hit any ground unit or any Armored or larger air unit. Battlecruiser health was reduced since they were being reduced to mere Corvettes, but their power budget warped in curious ways. They were very strong and their volley was the highest burst of Terran units currently available in the test map, but they overkilled easily and their positioning was all the more important to make use of their full strength. They remained slow, relying very heavily on terrain and spotters to avoid goliaths and other surface threats that could easily out-dps them in groups.

When considering the implications of sc2's power over sc1 and what I could do with the economics, I often toyed with ideas of how I could use the third resource. Originally reserved for Adamantium-based fabrication, the writing concept reached its conclusion and I decided I would not need to use Adamantium for anything besides P2x*. This meant that the Protoss and various Terran factions could easily use the third resource for something else.

The ideas of Supreme Commander's active resource system commonly appealed to me, but I would have a hard time implementing such a system into sc2. Perhaps I could have used auras to mimic various other elements from the title, like associated building strengths (power generators next to structures improving their efficiency or income). This seemed plausible for Protoss, especially since I planned to introduce some nature of mass fabrication late into each race's tech trees. Of course, I had to be extra careful with this because Supreme Commander wildly swings from "too expensive and inefficient" to "free resources forever", due to the compounding nature of perpetual constructed resource income. Supreme Commander, however, had far greater buildspace availability and far cheaper, comparatively, mass fabrication than what I was considering. The smallest mass fab for Gracia was going to be nearly twice the size of a command center, with detonation force capable of killing most units nearby when it died. Of course, the implications of abuse are obvious, so their cost and construction times were equally grand.

If I decided to add building adjacency bonuses I would quickly enter an extremely dangerous arena. One where efficiency was being determined exclusively by numbers. I figured I would keep the bonuses away from things like offense bonuses (firing rates on defenses, unit construction times, etc.) but give them some other nature of bonus. The idea was shallow and scrapped in most considerations, but it was there to be experimented with if I did find what I figured would be a relatively balanced approach. After all, I had removed terran addons, like the reactor and tech lab, as their bonuses no longer fit with the macro design. Perhaps adjacency bonuses could be delegated exclusively to Protoss, and boost things like pylon power radius and shield regen rates.

Another idea was seeding. Early into Apex I decided "minerals" would be largely artificial, created by Terrans through seeding operations to later be used for fabrication. This could have been exposed to the actual gameplay, and particularly Confederates would have the ability to construct seed ships who could then seed minerals which, over a large period of time, could grow to become harvestable. Confederates, being lower in tech to the Gracian faction you were controlling in P0x**, would not be able to create mineral fabricators. However, I still wanted to make them feel industrial and large-scale to fit their theme.

While the many ideas had merit, only mineral fabricators were actually created before the project died, and they were never truly put to the test in terms of gameplay concepts. Perhaps in another lifetime I will get to try out all these concepts and see how they actually played. I suspected they would be either too micro intensive to objectively add depth, or they would still be too easy to abuse. Ultimately, when approaching the concept of economics and macro, we need to look at terrain.


  • 4 - High Terrain Complexity

I came to the conclusion early into my years of modding that balance is a very difficult thing to perfect, and stats and values have diminishing returns in how they actually affect balance. Balance has to be considered less from a mathematical perspective and more from a perspective of psychological interaction. Very rarely is a player of any reasonable skill level going to allow his units to fight enemy units one on one in a perfectly orchestrated attack-move point and click, and even the simplest player interactions can throw academically designed balanced numbers to the wind. Therefore, a designer should first and foremost fit themselves into the head of a player - something most developers can't do as they aren't gamers themselves - and consider the potentials of each attribute when used by a sentient entity.

With little room to hole up your starting location, Players had to rely on external islands for static defenses in this mission, or had to keep a task force on hand to deal with the repeated drop attacks. They could opt to move their main module later on in the mission, or even move it early for an alternative starting location. Such were the options I left open. However, you could not lose your primary module or you would be defeated.

Terrain greatly influences how players behave because their gameplay mechanics, and how they perceive those elements, are directly influenced by the terrain. Furthermore, terrain is more important to a single player campaign than the balance of any given unit, unless we start talking about blatantly mistuned elements. Because we are greatly restricted by our limited access to AI functionality in mods and campaigns, we often turn to terrain as a way to help relegate the gameflow and balancing.


Consider the strengths of melee vs ranged and kiting, and the value of cliff levels even without defender's advantage in that ranged units can consciously decide when to fight. Therefore, the placement of cliffs becomes a clear and decisive means of changing values of units and locales. An easy example even most newbies are well aware of is cliffs above minerals valuing high range units like siege tanks. Limiting buildspace to more threatening locations is another low hanging fruit we can often times consider, especially when we have artillery. Why not make artillery require a lot of buildspace, so abusing it requires existing map control? Therefore, the player can't just turtle in a tight corner and wait for things to die for him. This also meant that fabricators required map control to be safe as well.

My efforts in campaign terrain design were to introduce brood war level terrain complexity, using examples like Heartbreak Ridge as a pinpoint in research to studying how terrain could skew balancing. Heartbreak ridge had many ramp-based ridges that greatly favored defenders standing on top of them, especially units like lurkers. Unfortunately, sc2 doesn't have cliff advantage like sc1, and presented solutions to this were not favorable or flexible at the time. Here I had to make some more compromises.

Defender's Advantage had to be relayed into the strength and durability of buildings, for starters. Things like bonus damage to structures and the comparatively low structure HP versus high damage units of Vanilla SC2 were abolished in favor of much higher hp pools and slower, but harder-hitting attacks, like the Sunken in BW, comparatively speaking.

I also slightly to moderately reduced the sight range of many units to promote more active scouting and sighting, especially since my project featured many high tier units with exceptional range that were otherwise considered too safe if they had an innately high sight radius. Such units became too self-reliant, since they could spot trouble and avoid it, sight choice targets, and make plays all on their own, eliminating the value of synergy and leading to players spamming single units. I didn't want to discourage such behavior by adding hard counters, instead I focused on encouraging diversity and synergy by allowing units to complement each other mechanically in an insinuated manner rather than through numbers. This feels more rewarding to the player as well, since their gameplay is now circulating their own flow of thought, rather than numbers telling them what to do (e.g. bonus damage, auras, etc.)

Inversely, we can think of vanilla Sc2 and how E-sports towers were designed to dramatically reduce the skill cap of scouting. By handing free vision to anyone who put a worker there, the vision game was available both to beginners and professionals alike in exactly the same manner. The goal of our project is to avoid such instances where the power of such an element is exposed so carelessly so that we can promote player evolution rather than regress to the mean and have the game play itself. This, of course, lent to potential design complications with Artillery and Tactical Missiles, which is why the former requires vision to be most effective, and both of them need immense buildspace - ergo, map control - to truly take advantage of. This lent into further weighting around positional play, which otherwise was going to be tough to work into a single player experience.


Positional play was going to be very important in the campaign. While I would struggle to make the AI ever recognize positioning, it was something the player was going to abuse no matter what I did in my design. It's just a thing players do. I wanted to design my maps in a way that allowed the player to make positional plays. Comparatively narrow chokes, exposed expansions, a minimum of 10 expansion locations a map, and plenty of crossing cliff levels allowed for vision control to make more impact.

0x02 - I introduced the player very early to what I called "risk expansions". Areas of high value that were exceptionally exposed. While the player would always be smarter than the AI no matter how far I managed to go with it, I always wanted there to be a sense of danger about your work. I also wanted to reward them for playing well, even if it was being cheesy. In a well-designed game nothing should be considered "cheese".

Of course, E-sports towers and things like that didn't exist and I designed my AI anticipating airspam to be a thing.

Ultimately, what we wanted out of the gameplay was to introduce a larger and more mechanically diverse single-player game world where high level players could, at long last, explore. By having an extremely high skill ceiling new players have something to aspire towards, and high level players have something to perfect. This is a critical ingredient for a good game. Good examples include DoDonPachi, quake 3 arena and brood war.


  • 5 - Teamplay

Almost all of my maps had a minimum of 3 players present. Most of them were at least 2v2, some were big FFA's. Unfortunately, this lead to the death of the project concept. The AI just lags too badly even with 2 players. My maps were impossible in the sc2 engine, at least at the current level of understanding the custom content community had.

Almost all campaigns are universally very easy 1v1 B&D snorefests filled with annoying distractions or A-move "RPG" drudges. I wanted to make something that had scale and unf to it in a way only a dozen Jaedong clones could deliver. Campaigns designed to fill air time with unrelated features, like minigames, were exactly what I was looking at to do the opposite of. I wanted all content and no grind. The critical deciding factor was going to be making each mission stand apart, and for that I needed computer AI once again. The behavior of your opponents and allies, coupled with the tech trees and map design, were going to be the gameplay. That is, after all, how Brood War survives in competitive gaming for so many years. With teamplay I was going to do something very few campaigns only touched upon at the best (such as WoS and Mekani from RCX). With sc2 the potential for greater exploration of this content was already available.

1.5 and hots saw the introduction of a build selector for computers. This was disabled for performance reasons, but we considered rebuilding it as an alternate system to give players the opportunity to change the builds of their allies when applicable. This had great potential for the campaign, but needed to be explored to realize any possibilities. I was considering giving allies specific build types you could encourage them to do, like defense-oriented, offense-oriented, air, surface, support, etc.

Teamplay can potentially be problematic to balance. Most games opt to have allies be useless and require the player to do all the lifting. I wanted allies to be as competent as a skilled player, so there was no tuning back any element of their presentation, be it scale or tech level. This meant that, of course, the player relied on their ally, and their ally relied on them, to succeed, as the enemies were scaled appropriately to deal with you as a team.

Blizzard mission design can be distilled down to a single mission - The Tosh jailbreak mission in wangs of liberty. A timed wave tug of war between Raynor's Raiders and the defenders which highlights two enormous problems immediately - your allies are largely useless, and the player's hero needs to be overpowered to offset this. There is no need to be conscious of allied movement because they are continual and on a timer, and serve little real impact in the game without your interaction. The rest of the map content is all gimmicks and therefore not content at all. The rest of the campaign is merely minor deviations from this concept. While very easy and fast to create it is hardly approaching what is possible with the idea of a campaign, even in the confines of the game's limitations.

Teamplay is a big deal and not easily summarized. Should allies be mandatory? In Apex F, for the most part, they were not going to be mandatory. Because of module networks and command ships allied defeat did not result in mission defeat. It just meant the going would only get tougher. In a world where computers need free money because their economic control is so abhorrent, where does the player fit? Some missions would see you best suited to defense while allies attacked. With customizable build options, however, the metrics of missions could have been even more dynamic. Perhaps you wanted to defend and have your allies focus on map control and artillery. Perhaps you wanted to attack but have your allies focus on defense. Different strategies may have suited different players and missions in better ways. This is a grey area, one that would need actuation to research confidently.

Speaking of which, computer resource income was something that was going to be addressed in a very critical manner if it ever became functional. In sc2, AI's are just handed money. Apex A had the AI handed money so long as it was still harvesting. For my tests I was just giving them free money period. However, I was probably just going to use the resource income multiplier so long as I could get the AI to expand.

Expanding is some black art in sc2. The AI will do it, but not consistently, and not when you ask them to. Workers have a habit of idling at expansion sites for extreme amounts of time, doing nothing, even without an APM cap or anything else going on. Furthermore, constructing buildings at an expansion is a truly biblical task for the game's toasters. It could take 20-30 minutes to see a Terran construct a few barracks at his expansion. While I originally intended for some opponents to build heavy production at their expansions to further encourage player map control this, again, was impossible to do with the traditional AI at the very least. It is a wonder how this junk ever got released in the first place. Which brings us to the big 420 shekel three-expansion subject.


  • 6 - Computer AI

Needless to say, much hinges on the computer AI performing, and the AI did as much as it could to avoid performing.

Vanilla Starcraft 2 campaigns just hand units to the AI and a-move them at you on a timer. There is no defense, no expanding, nothing. The maps amount to the player a-moving through empty bases with idle units effortlessly slaughtering everything in their path. To make maps """difficult""" random gimmicks are thrown in to distract or annoy the player. This is a common theme throughout many game genres, including FPS' (infinitely respawning enemies are a common "trick" to difficulty in Activision titles). Incidentally, the game has a poorly cobbled together melee AI, one the campaign mappers didn't use either because it was an unfinished, untested prototype, or because they were lazy. It is likely both were correct.

I would do everything in my power to avoid introducing anything into a mission that felt arbitrary or grindy. I simply wanted to produce a series of missions that reflected melee scenarios with custom data driving it. As it is, sc2's triggers don't offer the ability to do this. Melee Defense is not even exposed to triggers, since it is all locked in galaxy files that serve as a frontend for extensively hardcoded natives. While many custom sc2 AI projects sprouted up during WoL, very little actual information regarding computer AI exists for the game. Most of those AI projects turned out to be modifications of the existing melee that didn't address the major issues. Changing build orders is not enough, however, for someone wanting to make an AI project in such a title.

A few trigger and galaxy-based projects did exist but, again, they were too early and didn't offer anything near the power I needed. Furthermore, no one had attempted to build an AI for a campaign environment. Hypothetically speaking you can make the triggers do whatever with the AI within the confines of Galaxy's threading and performance limitations, which may yet be why the melee AI performs so poorly. Realistically speaking, such a project would outscale the entire campaign and take years to do. I had no intentions of wasting that kind of time overkilling what should never have been a problem to begin with, especially in this day and age.

Apex A was built upon the discovery that I could use Bullies, pre-placed units that didn't "exist", to construct AI bases. It was indeed that very same Tosh mission that used this, one of the only missions in all of sc2's vanilla campaigns. I put the idea to the test in various ways, and then published my findings.


Bully-based construction with pre-placed buildings and units being built by the AI according to triggers. Bully regions were divided into quadrants that activated with specific conditions, either time-gated so the AI didn't hitch trying to build structures it didn't have requirements to (this still consumed APM and could completely stall the computer's actions for an extremely long time, gogo testing), or when the AI controlled certain regions. In Apex A, a common theme was the AI attempting to bunker rush you or contain you, and when the contain either failed or enough time had passed it reverted back into macro play and gave up the contain.

As a proof of concept it was neat but in practice it was unreliable. The AI did not play out the same way every game, often "forgetting" key buildings. Additionally, Bullies had to be continually activated and deactivated in an infinite loop every few seconds, which created stutters when enough were active at once. Furthermore, Bullies had to be designed individually for expansions, and the thought of building bully setups for every player for every expansion for every map quickly sends the mind reeling. Additionally, bullies are static, which breaks away from our paradigm of trying to avoid pre-scripted behavior. And, still, no defense. I tried hybrids of melee and campaign, but the melee code's very existence causes the performance hits that troubled the project for so long.

Ultimately, at the heart of Starcraft 2, Blizzard didn't want to put in the effort to create an AI. Therefore, the foundations for a functional and performant AI simply don't exist in the game. This was the crutch the Apex design depended on fixing during Warcraft 3 and one that would plague it to modern day.

The AI is the gameplay of a campaign. You could have the most perfectly designed units and mechanics surrounding them and it literally means diddly dick all if your AI can't provide a believable, living opponent. AI is the pillar of balance and design in a campaign. It is what determines, ultimately, how strong units are in how it uses them. The AI builds the immersion for the player, because the AI is behind the elements of your writing and world. Gameplay is derived by the player interacting with the AI in various ways. Without AI you have no campaign. Plain and simple.

The AI absolutely had to expand, build its base, defend its base, attack the player quickly and reasonably intelligently, micro its units, and avoid the innumerable issues sc2's AI had. The concept of super dynamic build orders and other crap was immediately scrapped in favor of a Brood War-styled system, and as it turns out, the things BW found impossible to do all the way until BWAPI were actually the easiest parts of sc2 - the Micro.

Very early into sc2's life I had two things - functional models and functional tactical AI. It was like my time with Brood War had been inversed - all of my difficulty with custom graphics had vanished by now, and suddenly you had an AI that could micro. I thought between these two I had it all in the bag. I got Wilson'd.

JademusSreg produced an example Raven tactical AI that was far more aggressive and dangerous than the default. In the hands of a raven spamming AI it made an impressive opponent in the Apex A 0x03 demonstration map. A claustrophobic island map with limited resources and construction space, the map was one of my first major explorations into fusing elements of Starcraft with Supreme Commander (Mass fabricators were an element I pursued in this map, including their volatility) with an inner community meta (mass raven spam). The results were very exciting, but the map's other elements rendered Apex A a backburned concept.

The goal of the tactical AI for this project was not to create an opponent with perfect micro but, rather, simply an opponent that could match a player's aggressiveness and acted in a way players would expect. It had to avoid obvious damage, make clear and obvious micro decisions, use its spells somewhat intelligently, and do so at a speed that kept players on their toes. Perfect marine stutter stepping is unnecessary to making our game hard - but ensuring they used their marines intelligently was. The fact we had the power to give the AI perfect stutter stepping amongst other things was something I could never dream of prior to sc2.

Even the most basic tactical AI changed the game. Infestors that abused burrow and ran away when they burned up their mana, making room for more casters to move in. Commandos who dropped mines and then retreated back to max engage distance. Ravens who aggressively dropped turrets and explosives. Units that would be feared in the hands of a player but were laughable in the hands of a computer were now fearsome once again. They demanded respect and counterplay, something that had never before been done in a campaign.

Many of my campaigns died very early, but I don't let it bother me much. Growing pain is a part of getting old. I'm just not as smart and not as resourceful as I once was. Nothing is final until it's in the hands of my players. It's expected it will take a long time to get to the stage I can give them something, and it's expected a lot of my ideas just aren't going to work. That's all a part of experimenting, conceptualizing, and trying to find your way.

However, the tactical AI would be all we had to show for our time with the game until late 2014. JademusSreg often claimed he could "fix" the melee AI that was giving my work so much asspain. For years he attempted the idea off and on, but ultimately had nothing to show for that time. Motivation drifted, and the dream of seeing everything come together quickly died. He later told to me that the dream was basically impossible.

While this would discourage most people, another individual I knew, one who would be best described as the polar opposite of Jademus with no experience in sc2 whatsoever, did manage to address a chunk of AI problems and produced a far more functional alternative in two week's time.

This is not your average deathball.

By this time I had invested 5 years of time into researching and testing Starcraft 2. While I had nothing to show other than a few remastered ports and a lot of demonstrations, I had accrued a deep understanding of the game's limitations. Only now was I beginning to realize just how screwed I really was.

The true extent of the AI's dependency on hardcode only exposes itself once you get into it. Conjuring alternatives in Galaxy is only so effective, because galaxy as a whole is limited and half-baked. Compared to the languages it was based on, Galaxy is a bastard child with very little returns in terms of performance or flexibility. Virtually every single mapper I've ever talked to, even those who could be described as fanboys, described Galaxy as being by far the worst element of all of sc2. Coupled with Blizzard's extreme resistance in communicating with custom content developers and you had a big problem.

The problem at its heart was that the AI performs like shit. Standard melee with AI is terrible. It's not just terrible because my i7 is starting to age, either - newer systems with double the single-threaded processing power don't fare much better. Sc2 is a monolith and the AI does all it can to destroy its performance.

It isn't just about FPS, either.

The AI must run through some kind of buffer it gradually fills with shit or something, because it slows down over time, amounting to 33% slower every 10 minutes. Its execution speed just gradually slows over the course of the game. It doesn't respond very well to commands, like expansions. While we managed to gut many ugly pieces of Galaxy from it - like the 500ms sink that is the unreliable Clear Obstructions wave - and we did gather back some frames, it was ultimately an unsolvable equation all other designers had long since jumped ship on.

Consider that an FFA on exotic gardens (W&M's map) in standard melee actually brings me to 0fps before the first attacks even launch.

The actual unit counts of the map had nothing to do with the performance loss. Whether it was 20 or 200, the AI could still drop you to single-digit FPS if given the chance. The Zerg test map once started with 5 zerg. I condensed them down to 2. I quartered their unit counts. The FPS remained unchanged. To say I had tried to solve the problems the Zerg map presented would be to say I tried to breathe. I tried all kinds of different ideas. Merging melee functions with triggers, triggers calling melee functions, hybrids, nothing really worked. At the end of the day I had wasted an entire year on trying to fix the game's performance alone. Enough was enough.

Even at this stage I am not going to say the AI is entirely unsolvable. It would just require a solution no one knows yet. Probably as simple as bruteforcing it with an overclocked CPU. But now, getting into 2015, I've started Unreal 4 and I can't even think about touching sc2 ever again. To me that is inconceivable with the state the intern refuse blizzard calls an editor is in. I see sc2 as a resource in which I learned some stuff... and that's it. Hopefully my ideas will help someone out in some silly design ethical way if they're still sticking it out.

Fixing the AI might change my perspective a bit. But that seems impossible and no one has risen to the challenge of proving otherwise. I would very much like to see a positive end to this road. I'm still waiting to hear such a thing exists.


  • 7 - Audio

Audio is a massive part of my four-pronged Game Design philosophy. I consider the four elements equal - Gameplay, Graphics, Audio, and Computer AI. Without one of these pillars a game is only a shadow of what it could be at best, a disaster at worst. You can also see it as a game requiring absolute synthesis between its elements.

Sc2 and Sc1 both make extremely heavy use of commercial assets. Blizzard is one of the companies most reliant on commercial assets I have seen to date. But sc1 has something sc2 doesn't - audio engineers. At least, to an extent. It might be better to say that sc1 had a vision and sc2 didn't. Sc1's audio work is considered masterful even though a chunk of the sounds are 1:1 off of CD's. This is because the sounds are all identifiable, all have character, and are all very readable. Sc2 is very much the opposite.

A general way of looking at sc2 is that a large chunk of Terran impacts and explosions use the exact same explosion sounds - sounds with very high compression that tend to "flood" the audio channels and mesh together in a sea of noise. Just like graphics, audio readability is a big deal, because players are instinctively looking for audio cues for events. Sc2's audio is forgettable at the very best of occasions or otherwise lost in a sea of noise. Blizzard is much more interested in loudness war tactics to, once again, distract players rather than provide an objectively fulfilling auditory experience.

It isn't enough to "use sc1's sounds instead". Sc2's engine is capable of far greater frequency playback, and sc1 doesn't have many sound variations. While I was replacing a tremendous amount of sfx and eventually planned to re-voice all units (in addition to standard dialogue), the SFX needed special attention beyond just throwing crap in.

If I were to say, take the Z mutalisk sound from sc1, I would need to create around 6-7 alternate new variations for it. I can do this with very subtle compression and pitch shifts in Audition 1.5. The key is subtle. You want clear variations but not so varied they sound dissimilar.

You want attack sounds to be louder, more concise, and more pronounced than impacts. An example is my siege tank replacements. I had a set of sounds from Planetside 2 - some long, "smashing" muzzles, and some "thump" impacts. I used the impacts for the siege tank muzzle and the muzzles for the impacts. Why? The impacts were more easy to identify. Their sudden burst of presence demanded attention, much like the tank of BW. Very different in actual sound but very similar in their prose.

I often looked to infinity engine games for sound inspiration as well. Although also a heavy user of commercial assets, every spell and effect had its own character and sound to it.

I made special note to avoid the "growl" of heavy compression (seen in sc2 terran sounds a lot) because the loudness war does little for gaming besides blind the player's auditory senses from what basically amounts to random shit. Loudness War tactics are used extensively in Blizzard media - Metzen's voice in various WoW trailers, Kerrigan's voice in hots, etc. all have their lower frequencies gutted so they can be compressed into a wall of sound. They are intended to steal attention away from other elements, but have lost critical portions of their audio data, coming out flat, mangled, and unnatural. Such frequency crunching has become common in sound effects as well - hots' intro cinematic has shrill, painful audio that sounds extremely hosed on most audio hardware. The only conclusion I have here is that Blizzard was listening to their audio through laptop speakers.

Comparatively speaking I consider the Protoss "OK", I was going to keep a few of their sounds. Zerg, however, would be the hardest to reconfigure and seem nearly polar to other decisions in Sc2. They are minute, indistinguishable from any other sound in the game alongside each other, and are all virtually identical. The ultralisk is using a commercial arrow sound for its attack. I mean, what? Comparing the sc1 ultralisk to the sc2 ultralisk really emphasizes the difference between having a vision for your game and then just not caring at all. I didn't just want sc1, though. I wanted to expand on what made sc1 identifiable and charming while taking advantage of what a newer engine had to offer.

I needed to amass a few hundred Bear sounds for the Mother Russia faction that would inhabit this New Siberian map at some stage. The Bears would need distinctive features to easily tell them apart mid battle. The mechanized bears could have a muffled, metallic sound to their voice, for example. But they needed something. Something is better than nothing.

As a voice actor I would have no problems redoing nearly all voices in the game and doing all of the campaign dialogue myself. Only the female voices would have been a wild card. I'm used to that kind of audio workload so it wouldn't have been an issue, really. Community members would have voiced themselves. It was going to be hilarious because almost none of them are actors and half of them have these overly calm, monotone stoner voices.


  • 8 - Music

For music I was looking heavily into Japanese inspiration on two fronts - Castlevania and various Anime. The campaign story was seen as sort of a reflection of DBZ and Fist of the North Star, with ridiculous scale and silly circumstances. The music had to be hot-blooded and fit the mood of the moment. For the Russians a lot of Red Army seemed obvious, but other factions were more troublesome.

Many new factions and races existed in the concept from a revised Dominion to the Gracians, the Federation, Russians, French, Robo Hitler, and the Confederacy, all just for Terran alone. All had their own style and sound whenever they were presented, even if it was only the one time.

Music selection is an art. It is something I pride myself in taking a huge amount of time to do right. The praise of music selection in my projects has continually pushed me to make more and more effort into delivering a good audio experience.

Each mission was going to have unique music. A series of intro tracks for specific phases, if applicable, and then bare minimum 3 or so tracks to be looped between after the intro for those phases. This simple design allowed me to pace the mood and scale of music to the perceived intensity of the gameplay.
I had a huge amount of LCS-level pocket picks for my music selection I couldn't use in the Black Sun projects.

Unfortunately, a potential trouble spot awaited me in the form of cinematics. Since sc2's triggers are heavily restricted you cannot make use of either their save game or play movie functions, amongst a few others. Playing pre-rendered cinematics, the only way I was sure I could make anything decent, through fake textures and then playing the audio separately was a great way to get desyncing. Trying to time music to a triggered in-game cinematic seemed unnecessarily irritating. This was fucking 2014. Being unable to play my own pre-rendered cinematic was absurd. I was actually just hoping I could convince a friend to reverse engineer the game so I could bypass the restricted triggers and just play my own video files for cinematics. Else I was going to have no real cinematics, just talking faces like sc1 with mundane running around like ingame sc2 stuff. Pacing music to that would still be very hard.

There are several key fundamentals to the way I choose music. The pacing of the music pieces all must fit each other, and they preferably must be styled in a manner that identifies them to a race or character, a theme that can be repeated throughout the project. My Protoss used Japanese traditional instrumentation a lot, while I used a lot of Persian techno and pop music for the French. The Russians were easily identified by boisterous choirs, and Armageddon used a lot of Yasuharu Takanashi.

It's a copout to say "pick music that fits, duh". I dry acted dialogue to music to ensure it sounded like it fit. Pacing, of course, was important. Was the map a 3v5 against a ton of Zerg with infestors and vipers grabbing and spewing all over everything that moved, or was it a grueling artillery duel? While planning for player dynamic would be troublesome, you could definitely pace the music for what you expected the player to be facing. Fitting pacing builds energy and helps set rhythm in the presentation, and the voice acting also plays into this. Dramatic music alongside calm vocals doesn't quite set a mood.

I did look into getting my music composition setup rolling again but it's been almost a decade and I don't think that will ever happen. Were the AI to get fixed and I to resume sc2 work, I would probably focus on graphics first.

  • 9 - Graphics

Graphics would both be a mix of traditional modder "grab anything you can and make it your own" and entirely custom hand-made textures, particles, and meshes.


I'm a modder. I'm accustomed to doing whatever it takes to get a job done. Often times, as a modder, you can't make your own shit. I spent 15 years trying to learn how to make my own shit and I basically got nowhere useful. I still can't flesh out a TC of sc2 with my own content. I have to sort of get halfway there and make ends meet.

While I possessed the skill to model most vehicles and structures I would need for the remastering of the old races and the new assets for the new ones, I can't UV or texture. I also cannot model organics. I can, however, make particle effects and I can make fumeFX textures for those particle effects.


I owe my skills with particle editing to my extensive Black Sun work. Transitioning to sc2 was quite easy despite the extremely different systems.

These skills lead to an unresolvable hole - I don't have infantry graphics or units for my demon races.

However, as a modder, I possess exceptional skill in not only pulling content out of games but also remastering them so they function - and fit - in other games.

Many people port World of Warcraft models to Starcraft 2, and basically none of them actually update the models to support sc2's shader engine, so they look like shit. I do that, and more. Models like this are changed extensively from their home game, featuring new textures their previous incarnations did not have. I used MindTex extensively to produce normal maps for me, and I found it to be far more reliable and effective than nDo for this task.

I could port demons easily enough, but infantry remained an unresolved variable.

I will be honest. My work with graphics for this project would have been a truly biblical scale, even for someone with my experience. The first segment, only around a dozen missions, still could have been 3-4 years of work very easily - and that's if I could focus. Focus isn't really my thing. I'm kind of brain damaged, actually.

Only a few of the near dozen rock sets I ported from Tera. Each set had over 60 rocks. The sets were defined by their textures. Thankfully for me, my heavy reliance on Tera assets was greenlighted before my time working with them, thanks to the Tera developers saying they didn't care if people used their shit some time before. Certain other companies I had to be a bit more wary of, should I ever had desired to have released actual media regarding the project. Thankfully, I never intended to publicate anything much less release it.

I was not happy with sc2's graphics. Overly glossy, tons of UV issues everywhere, inconsistent resolutions, poorly kitbashed units, over-reliance on 2 certain textures for their particles, bad animations, just about everything about the game's particles irked me.

While I stressed over how I was going to fix the vanilla units and structures without a good, functional 3ds max import script, I was researching ways of non-destructively remastering the tilesets.



I was experimenting with a means to re-generate normal maps for the terrain tiles. Sc2 owes a lot of readability issues and overall ugliness to the terrain. While one could use splats to stack high rez textures, splats need manual alpha channels for transitions, and mapping dozens of 256x256 maps with splat textures just seemed absurd. A critical part of my design is to avoid having to do as much work as possible, thus the reluctance to manually trigger every action for the AI - an absolutely absurd notion that was the only feasible way I'd get the AI to "work". And, much like splats, the solution was only theoritical in its effectiveness. Both would have to be performed to see how effective they would really be, extreme time commitments for unguaranteed results.

While it is true I can replace terrain textures it is significantly less work if I can just make existing ones look less glossy and weird. By recalculating their normal maps I can try to bring out sharper definition but keep overall noise down. Believe it or not but sc2's grass textures actually look pretty decent - the engine is just so bad at handling their specularity, and DDS is so bad at compressing normal maps, that they come out looking like vomit.

My initial experiments proved promising but I gave up on the project before pursuing them further.

Consistency is a huge deal and one I would have to approach in passes.

While I planned to simply revise the textures for existing units like the Zerg and Terran, with proportion rebalancing in some places, work on anything I modeled myself and somehow textured or something I ported would be far greater. Indeed, most of the new assets I was pulling into the game were simply too high detail. I also had to ensure that when I updated the old assets and got rid of their glossiness that I did so in a manner that aligned them all together in how they appeared. Zerg, for example, are very noisy and difficult to focus on because of their abundance in sharp specular highlights. If I were to soften these highlights I would need to ensure all other similar organics followed a similar route of change. For Terrans I planned to introduce cube maps to give them a metallic look - I would need to do this for doodads as well.

In Apex A I planned to convert Zerg to an organic race called the Blood. Changing their unit textures was proving troublesome due to my lack of photoshop skills, but the new creep - complete with custom animations - was turning out very well. This screenshot highlights how easy it is to fuck up color balance - the units needed something else that helped define their shapes if this was the path I was going to choose. Then 1.5 deleted the project and all of its backups because Blizzard decided to silently purge game-related directories. Blizzard would repeat this in patch 3.0, deleting untold amounts of user data with no warning. This is how "professionals" do business. Wonderful.


Coming out of Black Sun: Retribution my work with 3d scenes had given me hope that I could fill a lot of sc2's problem spots - such as the really bad cliff tiles - with custom meshes. I planned to create a lot of custom doodads that would serve exclusively as detailing in maps, something like the Xerus and Ice cliffs.

These ice cliffs prompted me to consider building new cliff objects for every tileset so I could hide the extremely ugly standard cliffs. No doubt why Blizzard made them in the first place.

However, I was considering going even further than that. I had alpha-masked plating textures I could overlay as flat planes on top of terrain to sort of reach a halfway point between having a decal that acted as a terrain tile with high-rez textures and mapping terrain that way, or having no new terrain at all. By using alpha-masked objects I could keep existing terrain and then just place detail meshes on top of it to give it more pop.

Of course, you had to be careful with this. Too much pop and you can easily worsen the already troublesome reads of a 3d RTS. A part of my terrain philosophy was to place details in mostly unpathable locations. The exception would be if you were making an TPS hybrid like ET3. Then you could go berserk with high level detailing.

I did a lot of mapping for wc3, but I always avoided excessive reliance on doodad spam to build custom objects - a habit many wc3 mappers grew far too attached to. I figured if I needed more than 10 doodads to make an object that object was better off being modeled from scratch.

Mapping and graphics were expected to be iterative things. I didn't know how I was going to drop detail on ported assets - simply dropping texture resolution wasn't enough. The critical part was learning how sc2's animation system worked - a task rife with troubles of its own - getting functional particle examples into the game, making new stuff work with existing gameplay elements, then addressing what exactly I would need to make and how I would make it. Remastering the final chunk of that work, that is to bring them into a detail level fitting of the overhead camera, would need to be heavily researched before attempted. Nearly nothing about sc2's current graphics as a design would survive my work.

If I was to name one thing that was the core, fundamental issue of sc2's graphics, it's that the game is 3d. Sc2 gains virtually nothing from being 3d. As an engine, it's still 2d. The movement of projectiles using a Z axis is faked, as evident by how they interact with cliff levels and how weird their Mover system functions. The physics is client-side. The pathing, AI, collision, and everything related, is still 2d. But the game is inherently more difficult to read and more performance heavy for no real reason besides hipsters thinking everything needs to be 3d just because. You can't blame Blizzard too heavily for this, though. So many people would cry if the new Starcraft wasn't hurr definition 3d garfuks.

Really, something 2.5d would have been just fine, keeping the "3d" benefits of particles and movers but using sprites otherwise. After all, you can give sprites normals for lighting definition.

But, pickers can't be choosers, or some crap.


  • 10 - Mission Design

I had comparatively little ambition for mission design. I wanted a campaign that played out like big, long melee games with high difficulty and a lot of custom shit. Anything beyond that was extra.

My mission design was mostly insinuated. I used terrain and resource layout as a driving element of map design, not "lava sometimes appears" or "there is an arbitrary timer here". Maps, at most, had a boss that would spawn when certain conditions are met. Some maps had phasic transitions, usually involving the activation of one or more new enemies or enemies gaining more tech.

It was expected that the computer AI and its tech availability in the missions would dynamically build mission structure using the terrain and its elements as a spine to grow outwards from. I approached each mission with a set of ideas on how I wanted it to play based on my knowledge of how different Brood War could end up just because of base positioning. I made a mock gameplan in my head and tried to map around it. Usually the ideas just took off from there. Sorry. I'm not a planning kind of guy. No scribbles or excel shit here. Just good old fashioned yolo.

I had way too many mission ideas I wanted to fling at the game just to see if they were at all viable, but until I had a functional AI that performed well, I just could never actually try them out. This stagnated my work on mission theory overall, so compared to other elements, I don't have much to say other than "I wish I knew how it would have played". You can only predict so much. Too much of what I wanted to do relied on the AI to pan out. It was a balls-deep ordeal.

AI as a driving element opens a lot of things up in mission design provided you can control it, especially with Tactical AI. Ricky's Frenchies would expand everywhere like a plague, the Russians relied heavily on doom drops and artillery, and Armageddon units played like warlords and tactical masterminds, using their abilities to undermine you with a fraction of your units. Loyalists slow pushed and Confederates focused on siege. You could bring all of these characteristics to life in how the AI behaved. Special units with specific Tactical AI nodes that granted them increased reflexes and positioning alluded to more advanced infantry, recklessness alluded to Summoner madness (especially with Ricky), how many units they attacked with, how they spaced their buildings, so on so forth. Much of this is trivial to do in any game engine but was virtually unexplored in any campaign simply because designers never look to computer Ai to be doing design work even though it is, ultimately, what is presenting their design.

There were going to be no cinematics except between missions and very rare transitions. No help indicators or prompts were going to appear. No heroes or bosses would have big red indicators telling you to stand out of fire. I am a firm believer of a game requiring the user to pay attention to play it. All gameplay elements complex enough to require a tutorial, like the new Artillery, was simply going to have a bit of dialogue from the characters surrounding it. Everything like boss abilities needing cues would use real-world audio and graphics to convey their importance. Of course, all abilities and weapons were going to have unique particles. I'm a big fan of how Asians do particle effects. Lineage 2, Tera, Final Fantasy 14, these are my big inspiration for particles and how to design graphics about particle reads.

Basically, I look at how blizzard did bosses and I see that as an example of how not to do bosses. You wouldn't even get an HP meter. The bosses would be actual units that roam the map and hunt you and your allies down. I think of bosses like I think of shumps and DMC. Super fast, tons of shit flying around, and not a huge wall of HP to slowly grind down while the boss performs XYZ in sequence. I wanted you fighting a boss. Not a script.

Realworld balancing would likely lead to huge design changes. A few of the complex base units did get created in part by Xenon, but their incredible workload lead me to believe that the data editing portion of the project would be the second most involved aspect. Comparatively speaking, getting graphics for 90% of the new stuff would have been trivial. Sounds would have been trivial. Mechanically making them work in an editor as slow and user hostile as sc2's was going to be a nightmare. As Armageddon would appear at the end of the first segment they were of the lower priority to get conceptually functional compared to the base AI and other gameplay elements. They were, however, likely to be the hallmark of the campaign due to how many new assets and crazy shit they were going to present, just like the first Armageddon Onslaught.

I looked to games like arcade Shmups where, if you took your eyes off the screen for even a second, you could easily die. I wanted that kind of intensity in an RTS. I aimed to inject that kind of high intensity, high aggression into the AI and therefore the mission designs as a whole. It was going to be tough to pull off no matter the AI's performance.

I had never given up the desire to bring back an old BW project into sc2. Apex F would have been that opportunity had things turned out differently. With sc2 I could have done things I never could dream of elsewhere, and the existing proof of concepts provided by Xenon showed many ideas were plausible, if not extremely difficult to pull off.

The writing regarding the dialogue of the missions will probably not be released, since I may yet adapt it to something else, but those who have read it have told me I did a good job making a huge clusterfuck. I oh so wanted to turn it into something playable. Really, it was the writing that kept me going with this game for so long. I saw it as the only means to execute on those ideas.

I generally wanted missions to last an average of 1-2 hours. I figured there was no point having short missions because every map was going to be huge, and I wanted comparatively fewer missions per segment than sc1/sc2 had per segment, since the actual gameplay content was going to be much heavier and much more difficult to tune. Apex F's first segment was slated to only have around a dozen missions, but two of the final missions accounted for at least half the workload of the entire project up to that point. It was projected that if the first segment was successful that the resulting foundation would lead to having a far less painful time making the remaining segments. I had written dialogue and mission concepts for up to 4 segments, but the volatility of the project greatly hurt devotion to creative pursuits.

The first segment's mission loadout looked something like this.

  • 0x01 - First tutorial map. This was something like a 5-way Terran FFA with a bunch of subfactions. This map features a really nasty bunker rush and a lot of cheeky shit. It's designed to break new players and otherwise weed out casuals from people who will actually have a chance to make progress in the project.
  • 0x02 - 2v2 against Frenchies. A lot of map control and anti-drop stuff in here. The second tutorial map. The two French factions have synergizing armies so it's important to get between them and prevent them from merging attack forces. Your ally needs to stay alive.
  • 0x03 - Final tutorial map against Comrade Vlad and Mother Russia. 2v3. Features first boss and a lot of nukes and artillery. Big emphasis on controlling three major choke points and managing SDI buildspace.
  • 0x04 - Flavor community map. 2v2v2v2.
  • 0x05 - First Zurg map. 2v5, later degraded to 2v2 during my performance tests. This map killed the project. You were supposed to get trolled by mass infestor/viper/baneling and lurker swarms while a zerg amassed a heavy air army. First map you get your own artillery. First "difficult" map.
  • 0x06 - Followup zurg showdown. In this map you need to destroy the Giant Cock with the help of George Washington. 3v3. Big boss fight against the Giga Overlord. Big emphasis on sustained pushing to keep Zurg from rebuilding.
  • 0x07 - First protoss map. A 2v1v1v1 that serves as a bridge between Terran and Protoss tech. As a result, less difficult than the vs Zurg maps earlier.
  • 0x08 - Extremely difficult map against the first 2 tiers of Armageddon and Protoss Illuminati while you amass enough resources to build a Convenient Stealth Shuttle from a capturable Quasargate. Also a boss fight against an Illuminati Gardenship. Very high emphasis on extremely quick decision making, precision micro early-game, and high level map control.
  • 0x09 - This mission was float and could easily have been scrapped.
  • 0x10 - 5v3 against Iron Devils. Big boss fight against the Arclight Emperor and the Machine God. This is one of the few maps the enemy starts with having totally fortified it. Tons of drops and gundams. Very easy to lose early on from pressure.
  • 0x11-0x12 - Conceptually these levels got scrapped.
  • 0x13 - Fight with the Illuminati and Armageddon tiers 1-3 2v2. Duel with dragon boss. Introduction to enemy Wizard Towers (you get yours in later segments).
  • 0x14 - Duel with Phantom Startouched 1v1. Note that this is the first 1v1 map of the campaign and also the hardest mission to date, counting as endgame. Hidden boss offering one of the only achievements for the campaign if you can defeat her.
  • 0x15 - 5v3 against Necropolis and Sethis. This map would have been extremely performance intensive even if 0x05 performed fine. Sethis is a 5-phase boss that consists of much of the map time with global fight mechanics. Endgame difficulty.
  • 0x16 - Recreation of AO in which you can control any faction - including Armageddon. Something like 8v1. 10 tiers, each with a major boss associated with it. Just this map alone would be the biggest custom content project attempted in Starcraft 2 thus far... but I've already done it, just not for sc2, so I was confident that, in time, it could be done. Only the AI and performance really stood in my way, followed by the crazy data editor logistics. Endgame difficulty.



  • 11 - Dialogue-Driven Storytelling & World Building

Without the ability to play my own video files, the limits of what I could do with Starcraft 2 in terms of cinematics were extremely binding. I wouldn't get very far in an environment that required map switching for different environments, custom animations for close-ups, and extensive triggering for even the most mundane of choreography.

Plus, cinematics were going to be bootstrapped specifically to plot progression. I am a firm believer that gameplay elements should never be introduced or demonstrated to the player through cinematics. Not only does this completely shatter immersion but it inhibits the learning process of players, breaks the pacing of the environment and the gameplay, and is an incredibly cheap cop-out for what amounts to laziness in gameplay and map design. Furthermore to this point I am a firm believer that any kind of "glowing indicators" or other elements that directly tell you what to do to accomplish an objective or goal render that entire element superfluous and undeserving of a player's attention.

Therefore, ingame cinematics were intended to be an extreme rarity, with the majority of cinematic workload being offloaded into the area between missions.

The campaign had two major functions. The first was to produce an objectively difficult single-player experience - something nearly unheard of in modern gaming outside of arcades - and the second was to deliver an extensive dialogue-driven story. The two elements had potential to conflict; though missions could be long, their gameplay was very intense and demanding. Long-winded spouts of dialogue were not generally going to mesh well with high intensity gameplay. Here I had to, again, make somewhat of a compromise.

As mentioned previously, the campaign's writing took many turns throughout the various Apex iterations. An audio documentary regarding world building was released during the project's experiments, telling of the iterations and their evolutions. Ultimately, Apex F settled on a very complicated and difficult to portray melding of worlds combining community members with various pet projects of earlier writing endeavors.

I will go in-depth of major iterations of the project's writing.

Apex A

Apex A saw Alkhazir as a Starborn primal who was chosen as a guardian for the Manifold, an extremely powerful device that sees repeated use as a plot device throughout Apex's various iterations. Ultimately, Alkhazir was intended to eventually die and pass on the Manifold to the Creator's heir, Abel. This concept is also a driving element throughout some of the future variants of Apex.

As a campaign, Apex A saw the player taking up the role of Alkhazir as he, and his sidekick Vazera, rampaged across Gracian space building up resources and eventually making entry into the Apex. Alkhazir is not aware of his history, and Vazera, being very weak physically, acted as Alkhazir's technician and managed his nanite-controlled Necropolis as well as his flagship.

Apex A established the initial rewrite of the Starcraft universe. Notable changes include the year being far further ahead than Starcraft normally is (which is a cataclysmically massive plothole in Blizzard's design of the world in general, seeing as Terran advancement and growth proceeds at an impossible rate and at extremely weird intervals to meet poorly thought-out plot demands). The prison ship thing never happened with the Terrans, but factions like the Dominion and UED still exist, just their roles are different. Gracia is a faction making repeated appearances in the world's rewrite, taking up the role of what the Dominion eventually became in sc2 (a leading power by splintering). Gracia is universally viewed as less advanced than their parent faction (UED) but having strong leadership and large numbers.

A core feature of the Terran rewrite is that nanites are used in excess for cloning and fabrication to justify Build & Destroy missions, which otherwise struggle to make any sense at all in nearly any kind of writing. The concept of just building entire fortresses on a planet's surface quick enough before planetary bombardment removes you is troublesome enough to quantify in a realistic and believable fashion, although I address this by making big ships often having bombardment-oriented weapons so that you can live out your exterminatus fantasy one way or another.

Ultimately, the plot takes Alkhazir through the Apex and back into realspace. He brings a human, Annah, later tied to the beastskin Annayus, because she is "close to the stars", psionically touched by the essence of the Starborn. Annah enters the Apex via a mindlock and ultimately dies, but her echo manifests and draws the Blood back with it. The Blood exist only in Apex A's concept, and were a complete logical replacement for the Zerg.

Alkhazir - Hmph. Learn to look past the skin, Vazera.

Vazera - ...Okay.

Alkhazir - We'll deal with it later. Status?

Vazera - And Annah?

Alkhazir - Dead. Status?

Vazera - I could yell at you, or perhaps lecture you, but perhaps it would be best if you just saw for yourself.


Alkhazir - The Manifold blew out the entire ship.

Vazera - Drives, weapons, defenses, most of the nanite hives. You have screwed us rather dandily this time.

Alkhazir - So, repair it. Even one hive is enough to make repairs.

Vazera - You've been gone for hours, Alkhazir. I already held off one Dominion attack. They'll be coming back, and soon. I can't handle both tasks, and you need to recover.

Alkhazir - They want me. I'll be happy to share my gratitude for their contributions to our project.

Vazera - Your vessel is not perfect, Alkhazir. It will break down again eventually. Even controlling the nanites is dangerous for you right now.

Alkhazir - I am fine. Bring NATCOM online.

Vazera - I can deploy what modules we have, but you're going to be massively outnumbered. This wasn't just some expedition fleet like earlier, this was an entire battlegroup. How they got their forces in here - and why they're so intent on keeping us out - I don't know. But I hope you have the answers.

Alkhazir - I found nothing of value.

Vazera - You're joking, I hope.

Alkhazir - I need time to consider what little I have experienced. There is one sensation... I cannot understand.

Vazera - We don't have time.

Alkhazir - I'll make time.

Vazera - To what? Destroy the ship again?

Alkhazir - We knew this would never be easy.

Vazera - Yes, but I didn't know we were going to be gambling with our lives in the process of merely stumbling around -aimlessly- in the dark. Do you even know what you're doing?

Alkhazir - What is life worth living without risk?

Vazera - I want to learn the mysteries of the Manifold and the Apex as much as you do, Alkhazir, but not at the cost of my own existence!

Alkhazir - Think better of me, brother. I would never put your life in danger willingly.

Vazera - We keep the Manifold offline for now.

Alkhazir - Fair enough. I'll set up a perimeter and establish what I can. Where are we, exactly?

^ Apex A 0x05 - A portion of dialogue that comprises the beginning of mission 0x05, after Alkhazir leaves the mindlock with the Apex.


Vazera - Stop!! Stars of the deep, stop! Stop the ship!

Alkhazir - It's stopped.

Vazera - Damnation... still alive. Thank the Stars.

Alkhazir - Was it that upsetting?

Vazera - Alkhazir you fool! Do you have any idea what you have done?!

Alkhazir - I took us out of the Apex. Empty road in either direction. What more do you want?

Vazera - Oh, I don't know, how about an answer as to what the hell those things are and why they are after you?

Alkhazir - From what it felt like, they weren't individual entities. Just... a single object, in fine detail. So to speak.

Vazera - From what it felt like. Right. What else did you feel? Before you tried to kill us?

Alkhazir - You don't listen very well sometimes, do you?

Vazera - Look. I can't commune with the Manifold. I don't "get it". So, please, bring it down to my level.

Alkhazir - Fine. I couldn't communicate with the Apex.

Vazera - Okay. And?

Alkhazir - I couldn't get it to "listen", I guess. It was there, but it wasn't answering. So. I thought maybe Annah could.

Vazera - You think just because she can survive an encounter with its psi waves that suddenly she could communicate with the Apex?

Alkhazir - Perhaps. She did something, anyways.

Vazera - She did -something-.

Alkhazir - I think she may have given it the finger instead of saying hello.

Vazera - And she died... how?

Alkhazir - Well. Long theory short, I'm pretty confident she managed to pierce the veil between echo and reality.

Vazera - Echo... and... reality.

Alkhazir - She opened a hole in our dimension and stuck her head into it. Actually, I think that Echoes may be real after all.

Vazera - And... those things...

Alkhazir - -That- thing.

Vazera - ... Is the echo?

Alkhazir - Well. My current theory is not so much that they "came out", but rather that they were "drawn" out. I suspect that because Annah was so afraid, or perhaps so unaware, of her presence in the Manifold, that she kind of drifted out and pulled things back in with her before she decided to just... let go.

Vazera - Okay. Let me consider this for a moment.


Vazera - So. Aside from the reckless disregard for the effects of using something as overpowering as the Manifold like a blunt object instead of a surgical tool, while not knowing how to use it, exactly, you brought in a human for the sole purpose of poking it even more violently then what you were previously capable of doing, seeing as how the "first try" it "didn't work". Better yet, you decided to do this inside the Apex, probably without telling her what exactly you were doing. You just... threw her at it.

Alkhazir - Yes. Something like that.

Vazera - And... in doing so, she contacted...

Alkhazir - The Blood.

Vazera - Blood.

Alkhazir - Do you have a better name for them?

Vazera - Al, you really, really fucked up this time.

Alkhazir - I get it.

Vazera - No, you don't get it. The ship is trashed, Al. The modules. The Infinity Drive. I could be repairing this for days. Weeks.

Alkhazir - I get it, Vazera.

Vazera - You. Don't. Get it! This isn't the first time! The second! What will you do next? How far will your insanity take you? How far will you go?

Alkhazir - So I screwed up! Get over it! If you don't want to get burned, get out of the fire! I'll take the Ogorum and unfuck this universe alone if I have to!

Vazera - Really. You're barely holding the vessel together, Al. The second it breaks, the Manifold goes berserk again. What will come out this time? What will you do when-

NATCOM - Alert - subspace disruption detected.

Vazera - Oh look. It's your new friends, come back for sloppy seconds have we?

Alkhazir - Fix the ship.

Vazera - Oh sure. I'll get right on that, commander. Go on, Al. Wrestle the demons of your dreams. Struggle some more. Fight the waves.


Vazera - What are you going to do? This fight is already over.

Alkhazir - It is not over.

NATCOM - Material analysis inconclusive.

Alkhazir - Don't bother. It's just a piece. A fragment forced through subspace. Oh yes. I sense your connection to the Apex, little one. You're still whole inside... just not outside.

NATCOM - Detecting phazon particles.

Alkhazir - Of course. The big boys bring the big knives. But you should have brought a gun.

^ Apex A - 0x06

An interdimensional organ host, the Blood use subspace gates to funnel their manifestations through the Apex and into Realspace, pursuing Alkhazir until Vazera conjures a plan to seal the conduit by trapping it in an artificial singularity. This ultimately leads to Vazera attempting to seal Alkhazir along with the conduit, fearing his recklessness would only destroy his own chances of discovering who and what he was. Alkhazir proves too strong, even for his scheme, and uses the Apex to escape, where he eventually manifests in Protoss lands, and enters a Protoss Khalai plot to re-unite the tribes by seeing his appearance as an omen from their Gods.

A trend of the dialogue-driven writing is character interaction. Apex A was the first major script writing project where I was actually encouraged to make character interaction the sole driving force of the story. Although I had hoped to give much of the dialogue visual elements I was at a tremendous disadvantage thanks to Starcraft 2's limits.

My writing is universally dialogue-heavy, particularly what was once my life's work, Throne of Armageddon. I extensively used dialogue to build character because I find writing descriptions very challenging. My shaky grasp of the English language renders concise but fulfilling descriptions that avoid repetition extremely difficult. Of course, in a Sc2 project, I was limited even further by my inability to make custom graphics and my inability to pre-render appropriately paced and synced cinematics. This lent all of the pressure into the writing and voice acting - something that ultimately reached fruition as a stand-alone project in Black Sun's radioplay.

However, at the time of Apex A's writing, Retribution did not exist, and I still had a lot to learn about how I wanted to write dialogue.

My writing for scripts is very barebones. A great deal of implied diction is lost on text, and those that have read my scripts have alluded to this heavily.

The challenging part of Apex A was figuring out what to do with the Protoss, how Alkhazir would eventually meet Abel, and what exactly the Apex and Manifold really were. I never completed this writing in concept.

Apex C

Apex C was the next major iteration to manifest as an attempt at a project. It is based exclusively on the Starfallen and Demon World concepts I had been scraping together for a time previous.

The role of Alkhazir was shifted immensely. He became the Shadow of the Throne of Stars, to be awaken at a later time and to be used as a Big Bad. The player character was now Lazarus, and his advisor was Talpharaxx.
Apex A and C both saw a huge amount of practice voice acting, but C less so than A.

The writing of Apex C is far too extensive to summarize. Its world ballooned in size and far outscaled the previous campaign concept. Almost all of that writing persists to this day in the form of Apex H.


Talpharaxx - I will not ask what that was all about, but I will ask why you bothered to demonstrate it to me.

Lazarus - Very well. In short, the Starfallen houses are moving. They hope they can one day activate the Throne of the Stars, open the Apex once again. Return home, if you will.

Talpharaxx - They have always been 'moving'. It does them little good.

Lazarus - A while ago, Kalkatha came to me with an interesting proposal. I thought it was odd for him to ask me to become involved with any of his matters, given our... rough past. But the Fireborn clans think the Starfallen have the potential to upset the balance of the planes. Something has changed within them, so to speak.

Talpharaxx - Why do we care? Sorry, why do *I* care?

Lazarus - I think it will open of its own will and no other. I wanted to see how quickly house Odin would track me down, though. As it turns out, rather fast. They may be brash, but fools they are not. This was a test. They want to see how strong the Fireborn have become since the war.

Talpharaxx - And what is why you wanted my Gracian modules.

Lazarus - Correct. You have the same puppets they have. We beat them at their own game. We learn a little bit about them, they learn nothing about us - except that we are still one step ahead. Logically, their next move will be much more heavy handed. Odin knows well enough I am aware of their renewed efforts to breathe new life into this world.

Talpharaxx - Should he succeed in doing so, he will awaken only his own death.

Lazarus - Also likely to be correct. We don't know anything about the Apex or its full capabilities, or what exactly the Throne even is, or how the Starfallen came from it. All we know is that they are dying off, one by one, and the few that remain cling to the fantasy they can return to it.

Talpharaxx - Kalkatha, like any good warlord, wants to keep such a dangerous device locked down.

Lazarus - Half correct. He's more worried about the Stormborn than the Starfallen. If another war breaks out, the Stormborn will choose a new war king to lead them against us, purely for sport. Kalkatha seems to think that they have vested interest in allowing the Starfallen to open that door again, just to see what monsters come out of it.

Talpharaxx - Hmph. There are plenty of monsters here already. More than you or Kalkatha know about.

Lazarus - And this is why I found you, old friend. Your eyes are always looking where I haven't had the time, or the need, to look. Plus, you're pretty handy as an engineer. Should that door open, what happens to your little slice of peace, hmm? Ah yes, it burns along with the rest of us.

Talpharaxx - Very well. I will guide your steps, Lazarus. If only because having the Fireborn warlord indebted to me is worth the effort of ending a few Gods.

^ Dialogue from Apex C. Universally, Lazarus begins entry as a player character around his interactions with the Starfallen and their human clones, as they are a major player in the universe and act as a way to ground players in recognizable locales despite the extremely far-fetched nature of the world and writing.

Apex C took the "human clone wars" to new levels and was an entirely post-apocalyptic setting in which sentient humans didn't really exist anymore. More Starfallen than the handful in Apex A existed, forming into clans known as "Houses". House Odin, the primary antagonist for much of the introduction of the campaign, serves as Lazarus' heartthrob and acts as your punching bag for a time. It becomes evident eventually, however, that their resources are immense, and they are more dangerous than they appear.

Apex C's writing is integral to Apex H, so it will be gone through in detail during the developer series at a later point in time.

Apex F

Apex F was an accumilation of an immense amount of writing data and became my largest writing project ever conjured since Throne of Armageddon's world. It outscales even Loladins of Legend, a project I spent over five years building the world for, in multitudes.

Combining and malforming the demonkin worlds of Apex C, refined elements of the rewritten Starcraft world, and then a vast amount of writing circulating pet projects about community members and the OPERATION: MANCATCHER slash fic, Apex F became an unholy, but surprisingly wieldable, amalgamation of highly offensive and at times extremely matured writing.

Apex F represents some of my greatest dialogue writing and has an incredible amount of opportunity for voice acting practice despite the project's death. ... lackking-2 ... rrator3x00 ... darkprimus ... romagnus-2

While I'd love to go more into the project I don't plan to. There is the chance I may yet try to turn its writing into something else, and the voice acting I've done has given more than enough away about some portions of it.

Apex F was intended to have at least five segments, numbering an average of a dozen missions each. Voicing even a tiny portion of dialogue amounts to many hours of recordings. I'd have had a project easily outscaling Retribution's audiobook on my hands, but it would have been far less challenging to make. Nearly all voices besides the Female voices would have been trivial for me to make, and I didn't have to hold myself to quite as high a quality standard as I did Black Sun, due to Apex F mostly being about humor.


  • 12 - Deviations

I considered making RPG-centric designs as this would allow me to dodge the performance problems of the AI and focus much more heavily on individual interactions. At first glance this seems like a great opportunity to use all those high rez assets in a up-front environment.

However, it is actually impossible to justify working with something as heavily restricted and as user unfriendly as Starcraft 2 for something like an RPG when you've had the UDK, and now Unreal 4, available for years. Making an RTS project in sc2 seems viable because that is a comparatively much more difficult project to get off the ground independently, but resources for other projects are extremely abundant.

Do you want to explore this as a level? In Unreal, it's not that hard at all. In sc2 it's entirely impossible. Sc2 inhibits even my skills, and I am as far from an artist as it gets in my line of work.

Sc2 plainly doesn't support other game types very well. While many people are content to deal with the half-baked results they get by forcing it into other roles, I am not really a half-baked kind of guy. For me it's either ham or ff at 20. I've wasted way too many years in engines capable of only doing half a project. That doesn't mean I can't learn from it, though. Every experience, no matter how minute or how grand, is an experience that reshapes who you are. Men are built by their experiences. My experiences have made me extremely cautious and I have huge trouble with commitment as a result. I didn't commit too heavily to sc2 despite my time with it, and so I come out of Apex a little wiser and maybe a little salty, but not without benefits.


  • 13 - What I Learned

Collision is a huge deal. Most games I worked with prior to sc2 don't have a lot of flexibility in collision or it doesn't make a big difference. When dealing with something as touchy as sc2 and AoE damage I was really surprised by how balance shifted to bio just by making marines a little fatter, even with a broken light tank doing 40+ damage a hit in an aoe with no falloff.


I learned about physics. Although my experiments with sc2's physics only lead to boundless rage and frustration and days of wasted time, the skills I learned in dealing with PhysX helped me get functional hair and cloth into Unreal 4. I still haven't figured out how the fuck to get physics colliders for hair and cloth into unreal that actually work yet, but it's still something.


I learned the ins and outs of different shader models (Sins -> Sc2 -> Unreal 4) and how the different systems improved over time. My transition between those engines made my understanding of Physically Based Rendering, an otherwise complex subject, very easy to get into.


I learned that color balance is a huge deal in mapping. Environmental mapping is not something I am accustomed to and I am going to have to improve a lot in my understanding of it if I am going to make massive zones in Unreal 4 that look and feel real. It's not just about having assets, it's about how you use them. Avoiding amateur pitfalls and bad habits and all that garbage.

I learned how difficult Starcraft 2 can be when you take off the limiters of an AI and tell it to drop 5 bunkers in front of a person's command center. How that can suddenly turn things around and make an otherwise bland experience exceptionally difficult. How that one experience, that challenge behind it, can drive you to become better. How challenge can drive an entire design. How and why I demand such challenge from my projects. It's a name to the face of my concepts for all the years I never was quite familiar with. That's what I want to make because that's what I crave. Through this I learned more about myself and why I have come to despise most modern games even if that opinion is not a popular one.

That's what it's all about. Even if I never finish something again, even if I don't find my place. Knowledge is all a man can really ask for. Life is like a box of dicks. You don't know what you're going to get, but you know it's got teeth this big and they're going right for your throat.

Related Links

I released the few particle effects I made for sc2 here and if you want the max files for them, you can have them as well.

I released a chunk of my FumeFX textures a time ago, useful for building particle flipbook sheets, here. Please provide credit to me (IskatuMesk) if you use any of them.

Special thanks to Xenon, Ahli, ArcanePariah, hixy, JademusSreg, Ricky, and AA7's surging pecs.

Article versions

1.0 - Initial publication
1.1 - Script-related stuff, some polish
1.2 - Some elaboration here and there
1.3 - Early draft polish sweep

1.31 - Gameproc Audit

1.32 - Gameproc Beta Draft (Oct 12 2015)