A Visual Timeline of Steel Hunters

From tech/vfx art sandbox to now… Kind of.

A game’s visual fidelity is important as a complement to its tone, atmosphere, and gameplay style. Beautiful games with hollow gameplay might as well be paid screen shots.

I love and always have loved graphics programming, technical art, visual effects work, and general scene composition. I mean, I wrote a book on 3D terrain programming (back in the day when that wasn’t a super trivial thing). I wasn’t a nerd at all.

My first real attempt at establishing gameplay through color keys was an XNA game called Asplode!. And, for your viewing pleasure, my first completed game, as I had been working exclusively on graphics demos and “engines” for years and years before I ever attempted to make an actual playable game. So, darken the lights in your room, sit back, and view: Asplode!:

That said, I’d point to SPACE COLORS as the first real game where I very intentionally used discrete colors to set the tone of the game, discern what’s going in heated combat encounters — a top-down space shooter on iOS was not the easiest thing to manage. All things considered, it was a fairly simple game, but I was pretty pleased with how it turned out for a fairly short turnaround and as a top-down shooter on iOS that was actually playable. Here’s a rough video of SPACE COLORS:

The Visual Language of Steel Hunters

(Very) early on into my work on what would become Steel Hunters, I actually wrote a bit about color and composition. Keep in mind: I’m an English major and haven’t taken nor read any sort of proper information on color theory or what-not, so try not to laugh too hard. I just like colors; I mean, this is my apartment (it’s not normally quite so red, but apparently the Hue lights felt angry — or filled with love, who knows).

Anyway, even my attempts at a focused topic and blog post yield tangents.

Aesthetic Iteration One: Blandtown v1

(Blandtown was the name of the map file up until… Well, you’ll see).

When I first started on Steel Hunters well before I ever thought it would become something I’d pursue for a year and a half and write an article in which I say I’ve been pursuing Steel Hunters for a year and a half, the game’s fidelity, style, visual effects, and composition were just about all I was playing around with. This is one of the early images of the sandbox; I had the idea that it would be really interesting to have this very high-contrast/high-saturation world with detailed mechs (the mech in the image being “detailed” is entertaining compared to what I’m using now) and a flat plane with a grid material for the “terrain”.

All things considered, I do still enjoy that aesthetic. And I certainly would have had a vastly reduced content overhead. I could even model meshes myself. THE THOUGHT.

Aesthetic Iteration Two: Blandtown v2

I wasn’t quite ready to give up on the simplistic buildings, but I did want to add a bit more detail and personality to the world, so I added an arid desert landscape with mesas (except for that most prominent one, which is unlike any mesa I’ve ever seen, but I digress).

I quickly started warming up to the idea of a more detailed sandbox — and you can definitely see where the red/orange-centric palette focus originated. Technically, there’s not a whole lot to speak of here; the terrain material was just a simple height map with about three texture sets all based on terrain masks generated along with the height map. There may have been a macro mask tiled across the terrain for some basic color variation, but I can’t remember the specifics.

It’s probably pretty noticeable by now, but I’ve basically used trueSKY for sky rendering from the very beginning. I first started using Simul’s middleware for that back in Unity on Joy Machine v1 (which was my first attempt at a project and started back in Austin, Texas; I could not make that project fun no matter how many prototypes I did), and I’ve essentially relied on it ever since. Eventually, I’d end up doing contract work with Roderick (bossman at Simul) too — and, I must say, he’s one of the most brilliant people I’ve ever worked with.

Aesthetic Iteration Three: Blandtown v3 (and Prototype 1)

Back on track: once I got my hands on some real visual rich additions to Blandtown, I had to keep going. I grabbed a Desert Town environment pack from PolyPixel and quickly put it to use to deck out what would become the centerpiece for screen shots for the next eight-nine months (and you’ll see how laughably small an area it is a bit later).

I adored the results so much. So much. And at this point I had initial gameplay prototyped and was able to move around (even had a booster attached to the mech to get my verticality on. It was around this time that I really started thinking about the directions that I could take this sandbox going forward. It definitely wasn’t “I’m going to spend as much of my spare time and more of my personal money than I should for the next year with this as my biggest priority”, but it was… Well. It was “hey, I want an action-oriented third-person mech shooter. This could be great.” So, yeah, not quite as intense.

Aesthetic Iteration Three: Blandtown v4 (and Prototype 2)

Over the course of the next month, I found a mech mesh on CGTrader that I really liked, so I did what I could in Maya at the time to separate the mesh out into mech pieces and assemble it at run-time and walk around and shootybang objects to create explosions.

It was also at this point that I first started thinking about the concept of the Behemoth — a giant, complex, hulking mech that would take 10–20 minutes to take down. And, subsystem by subsystem, attacking the behemoth would slowly disable its ability-set. None fo that was implemented in this prototype, but I wanted to do a quick scale test to see what kind of scale I had in mind.

At this point, I had resisted Joy Machine, LLC in East Lansing, Michigan and decided I wanted to take Steel Hunters (which I wanted to call Steel Harvest (based on my love of Dashiell Hammett’s Red Harvest), but barely a day later I saw the first story on Iron Harvest and revised the name a bit. Steel Hunters was already #2 on my list of names, given the the importance of “Hunters” in the game — and somewhat of an homage to my love of Monster Hunter.

I also was invited to show off the game at Michigan State University’s Meaningful Play conference. The game was in no state to play, but I showed people around the world and described the game to the surprising amount of folks that game by to hear me ramble.

And while my mind was 95% made up that I wanted to pursue this as a full-on project, the conversations with the folks at that conference was what sealed the deal.

Aesthetic Iteration Four: Blandtown v5 (and Prototype 2)

It was this screenshot that really made me fall in love with what this project could be capable of, even with me as the sole developer/artist/designer/everything/etc.

It’s way too heavy on the orange, sure, but this is one of my favorite stills from the development of Steel Hunters thus far.

The mood of the screen just felt right. The player’s mech (which is basically just a silhouette in this lighting setup) walking (well, shambling) in a clearly-makeshift manner towards a hulking placeholder behemoth in the distance through a ghost town. A ghost town that somehow still had electricity to power that “Gas” sign. Hush, that’s not important right now.

Some day, regardless of how Steel Hunters ends up looking, even if there are screenshots which are far, far more polished and good looking, this is the image I want as a poster.

Aesthetic Iteration Four: Blandtown v5 (and Prototype 3)

And this is the last screenshot I took before I flew out to San Francisco earlier this year for GDC. It was the final prototype iteration that I took for the game. And, really, this prototype was purely a test of alternate ways I could setup the core systems than an actual change in gameplay from prior prototypes.

This has been the “go-to” image for showing Steel Hunters for a while now since

Aesthetic Iteration Five: Nevada v1 (work-in-progress Greenlight Demo)

I’ve been writing the entire game’s codebase from scratch since then. As well as working on more detailed materials, making engine changes (such as adding proper volumetric lighting using a different algorithm from the one that UE4 ended up adding a version or two later; I still prefer and use mine), making a new environment inspired by Monument Valley and Red Rock National Park with some “artistic license” to make the entire map on the outskirts of Las Vegas (three hundred years into the future).

This environment is one of the five intended for launch, is 4km x 4km (16 km²) in size, and is the home of one of the five human factions intended for launch — with each faction having its own unique style of mech parts and form language. These styles largely evolving from whatever environment and natural resources the humans in these areas had access to.

And remember all those screenshots with a road in between some small buildings like a gas station, motel, and leaning tower in the distance? That entire area is the section of assets a bit to the lower-right of center in the following image.

This is why the launch target is five environments and not a dozen or so. That and each environment has a series of unique properties, dangers, possible complete mission-altering natural events, a substantial amount of land to explore and chase your quarry around as it gets increasingly wounded and/or crafty (Behemoths have adaptive AI that take into account the kind of damage being done to it by you and your up-to-three other co-op players, as well as what types of subsystems your group is focusing on).

I’d much rather take the time to fully realize five completely unique environments than a higher environment count with less detail, less unique properties, more obviously-reused content, and just… Huge amounts of land that aren’t as carefully-crafted as I’m aiming for here.

Aesthetic Iteration Six: Nevada v1 (work-in-progress Greenlight Demo)

This is the last environment screenshot I’m posting for a while now; I was trending back towards a more modern-day normal overcast blue sky as well as easing up on the saturation/contrast a bit.

Things have… Changed since that screen shot. I won’t say how, but there have been a significant number of changes to the landscape style, world and atmosphere style, the lighting setup, materials across all objects, a completely new level layout, and so on and so forth.

One Last Thing

The style, shading, and variation in mech metals is something that I revisit on a pretty regular basis to try and improve upon. One of the last iterations I did that I posted publicly (like I said, saving newer screenshots for a while) was the beginnings of a pass with some alternate shader tweaks and a different BRDF model. By default, Unreal Engine 4 uses a lambertian BRDF model, which I’ve since changed to be an Oren-Nayar diffuse model applied to the diffuse calculations for both my standard shader as well as the diffuse component in a metallic clear-coat shader. These screen shots were fairly early on in this change, but I modified the UE4 core material code to support the Oren-Nayar model in addition to the lambertian model (the Oren-Nayar solution is slightly more computationally expensive, but far from prohibitively so). The difference is incredibly subtle, but overall I’m happy with the change.

I’m also working on an alternate fresnel shading technique to apply to metallic materials which is still a work-in-progress. I’ll write a more technical post in the future when I have more time to go into the specifics, but it’s not as hard a change to implement as one might think.

So, I’ll leave on this note:

For more on Steel Hunters, check out http://steel-hunters.com/ as well as the base Joy Machine site. Hugs.