RTS games need to convey continues streams of highly detailed information in realtime to the player. Sometimes, fancy graphics and design gets in the way of that. Take a look at this screenshot from the RTS game 0 AD:

No doubt the scene is impressive with high amounts of detail and lots of action - but I'd venture to say this high level of detail is detrimental to gameplay.

Consider the following questions: How many units are fighting each other? How many units are on each side? Which sides are fighting? Which soldiers are melee and which are ranged?

Your brain needs to answer these questions in a matter of milliseconds. Even if it takes 100 milliseconds longer to mentally process, due to the visual complexity, it has a profound effect on the experience of playing an RTS game. If you have difficulty believing that, try playing a game in which user input is processed after a delay of 100 milliseconds. We've all experienced what that's like in an online game with moderate latency. Visual complexity, too much of it, creates a sort of mental latency.

Now take a look at a scene from the age-old classic AOE2:

I'd argue that while the AOE2 graphics aren't nearly as impressive as 0 AD, they are more conducive to gameplay. The design, however modest, serves its purpose. Here we see a bunch of archers on horses and trebuchets attacking buildings. At a glance, the scene is so much clearer and easier to understand than the screenshot from 0 AD. This of course, is not by design, but a happy accident: the makers of AOE2 were limited to pre-rendered 3D assets (using 2D sprites) and pre-rendered terrain, given the limitations of computing at the time. The simpler graphics just happened to work better for gameplay. 

Now take a look at a screenshot of units attacking buildings in AOE3:

It takes a moment to understand what is going on here - doesn't it? If I were an alien from outer space, and had nothing to go on but the screenshots from AOE2 and AOE3, I might think AOE2 was the newest game in the series. Poor adoption of 3D realtime is a technological backpedal from using 2D sprites.  

Because of its continued success, Microsoft is still making expansions for AOE2, almost 20 years after it was first released. The original sells so well, you can still buy a copy at walmart. AOE3, despite being made many years later, and with better technology and funding, maintains a fraction of the following AOE2 has.

Similar examples can be drawn from the likes of C&C and Red Alert. As soon as these titles made the switch to 3D realtime rendering, the gameplay suffered. I don't think anyone would argue that RA3 was better than RA2 - or that AOE3 was better than 2. The sales numbers alone would disprove that notion. Whatever happened to the Dune franchise - the one that started this whole RTS thing? Dune 1 and 2 were released in 1992, Dune 2000 in 1998 - to great success. But after Emperor: Battle for Dune was released, the first in the series to use 3D realtime, Westwood never made a game again.  

This phenomenon is not observed in other genres, such as FPS, where even the most primitive 3D realtime implementations are better than 2D with a 3D viewport (e.g. Doom classic vs CounterStrike 1.6). This is because the human brain is much better at processing 3D information from the first-person perspective. After all, your brain processes first-person perspective in everyday life. You can however, observe this effect in other genres that use the isometric camera - e.g. Roller Coaster Tycoon 2 vs Roller Coaster Tycoon 3. Why did these franchises peak at the height of pre-rendered 3D tech and flatline just after the transition to 3D realtime? Hmmmm. 

In this forum discussion on none other but the 0 AD forums, the poster talks about edge unison vs. distribution and what it means in visual ergonomics. He came to the same conclusion I have, but in different words. 

I do not mean to say 3D realtime has no place in RTS gaming. I only want to suggest that it is much more difficult to make a proper RTS game with 3D realtime. Starcraft 2, a game rendered in 3d realtime, has absolutely none of the visual problems 0 AD has. But Starcraft 2 is expertly designed by the people who wrote the book on RTS - it succeeds visually, in spite of the difficulties of 3D realtime, not necessarily because of it. The high poly 3D models, the level of contrast between terrain and units, the overall visual distinctiveness of game objects and anisotropic filtering all work together. No wonder it took 5 years and 12,000 builds to make. SC2, and the original Total Annihilation, are the only two 3D realtime RTS games that have stood the test of time - in my book, at least.  

Other common mistakes made in RTS game design is one of scaling. In the late 90s - the RTS golden age - developers couldn't implement zooming because of the limitations in computing at the time. Nowadays, you're hard-pressed to find an RTS game that doesn't have some absurd level of zooming. The effect on gameplay is catastrophic. Supreme Commander might as well be called Icon Wars, because you are forced to play the game completely zoomed out - at least, if you want to compete online.

While giving players the ability to zoom out this far sounds good on paper, it makes for very boring gameplay. The emotional climax of any RTS game takes place while watching and participating in battles. If the meta requires you to be zoomed out to play effectively, so that the battles are tiny dot wars, well, that pretty much sucks the fun right out of it.

Note that you can only zoom in, and not out, in any Blizzard RTS game. Blizzard didn't fall into that trap. Blizzard knows better. 

Somewhere in this podcast series about the making of Warcraft 2, Bill Roper, one of the lead designers, discusses the decision to not show too much information on the mini map, so players would not be tempted to play the game through the mini map. This set a precedent in RTS games. And the same concept is applicable to zoom: showing too much causes the player to not be able to experience the game as it was intended.  

Let's take a look at a screenshot from the most recent incarnation of Total Annihilation: Planetary Annihilation. 

Here the units models are quite good, and visually distinct from the ground underneath. But the scale, dynamic lighting and map curvature obfuscate the information to such a degree that the information isn't so easy to digest, at a glance. 

I understand the pressure game developers feel to innovate - it certainly looks impressive, and no doubt, they sold a lot of copies of PA on the strength of this visual style. You can build and move to different planets? Win! Zoom all the way in or all the way out to see everything on the planet? Win!  But I question whether these innovations ultimately lead to more fun and engaging gameplay.

No zoom out and a flat map might sound boring on paper, but at least we know those things work. My goal with Strike Tactics is to use what works, and not fall into the trap of over-innovating. Just because you have the technology to do something, doesn't mean you should. Game developers must exercise restraint - and constantly question whether a feature harms or strengthens gameplay - regardless of how appealing it seems on paper. After all, what good is any of it if the gameplay isn't there? 

Strike Tactics is a top-down pre-rendered 3D game. The pre-rendered part is by design, but also, by necessity. Many of the CPU limitations RTS designers had in the late 90s are also limitations I have in the browser. But because all game assets are pre-rendered, the units can get away with extremely high poly counts, which serves to give them visual distinctiveness and contrasts them from the terrain.

The fact that the game plays in the browser has limitations, but it also removes the temptation of rendering everything in realtime and upscaling everything to an absurd degree.

I have a lot of theories as to why the RTS genre has lost the popularity it once had. The unsuccessful transtion from 2D to 3D is just one facet - there are many more. At some point after the RTS golden age, because they had the technology, developers started making enormous Michael Bay-style RTS games, advertising bigger, badder and MORE, but neglected the fundamentals. The fact is, the human brain can only process so much information before everything on the screen becomes a meaningless blob of lasers, smoke and metal.

 

Game developers must exercise restraint.

Top-down vs. Isometric

Originally Strike Tactics was to be isometric as the engine was created from a fork of Feudal Wars (my previous project, now on hold), which is isometric. However, after a bit of experimentation, I decided to break away from the mold of the isometric/dimetric perspective. My reasons are purely practical. I can get away with smoother rendering and freely rotate my graphics without having to worry about perspective. It also completely solves the problem of unit occlusion: if the camera is pointed down from directly above, no game objects can block other game objects (unless they are underneath another game object).

Unit occlusion has been an age-old problem in the RTS genre. It's normally solved by showing unit silhouettes, on the objects behind, which looks tacky and can be taxing on performance. I solved this problem with Feudal Wars by reducing the opacity of any object that was blocking another object. Note the castle and tower has reduced opacity to show the units behind them. I thought this technique was at least better than using silhouettes. 

In a medieval game like Feudal Wars, a camera tilt is a necessity, because human beings - the primary unit in these kinds of games - look horrible from top-down. Like so:

Mechs, aircraft, tanks, however, look great, if not better, from top-down:

This is because tanks and aircraft are generally designed to be aerodynamic, which makes the side facing the direction in which they move take up less space on the vertical plane, and the side facing the camera, take up more space on the horizontal plane. As a result, you see all of the detail of the object when looking directly from above and less detail, the lower the camera is on the horizon. 

Top-down is the biggest risk I took regarding game graphics, because there isn't any real precedent to lean on. There aren't any well-known (if any) top-down RTS games, even going all the way back to Dune. RTS games have always had a tilted isometric, dimetric or similar camera, and for good reason. A tilted camera allows you to see more angles of the object. So I knew it would be a challenge to get everything looking pretty and interesting from directly above. Nonetheless, thanks to some extremely skilled designers, the visuals for Strike Tactics, I feel, are in a good spot. I'm happy with the aesthetics and feel going top-down was the right move.