Last year German developer Mimimi Productions made a splash with Shadow Tactics: Blades of the Shogun, a real-time tactical stealth game deeply inspired by older works like Desperados and the Commandos games.
Shadow Tactics sported an intriguing design (check out this deep dive into its dynamic detection mechanic) and became something of a standout success for Mimimi -- though the company almost went out of business shipping it. Read more about how that happened (and what lessons were learned from the experience) in this frank postmortem of the game, written by Mimimi cofounders Johannes Roth and Dominik Abé.
Nearly a year ago, Mimimi wasn’t known to many people. Our previous game, the 3D action adventure The Last Tinker: City of Colors, did okay, but never reached a significant audience.
By end of December 2016, our biggest dreams came true: Shadow Tactics: Blades of the Shogun, our love-letter to Commandos and Desperados was finally released and loved by both players and press.
At the same time, we were nearly bankrupt and had already signed terminations for most of the team. There wasn’t any money left to survive for another month.
In this postmortem, the two founders of Mimimi Dominik Abé (Creative Director) and Johannes Roth (Managing Director) will dive into what went right and what went (awfully) wrong.
Making games and crying about it since 2011.
What Went Right
1) Going all in, achieving great quality
Roth: Shadow Tactics was released on December 6th, 2016 – the day before the German Developer Awards, where we won in as many as four categories, including “Best German Game” and “Best Studio”. Roughly at the same time, the first reviews came in. Scores and verdicts exceeded our wildest dreams: an 86 rating from GameStar, 92 points from PC Gamer, even Rock Paper Shotgun and Kotaku praised us. Streamers and Let’s Players recorded hundreds of hours of gameplay. That month alone we sold over 60,000 units, double my highest estimates. On Steam, we are currently at 98% positive reviews, which feels absolutely surreal.
The whole team, including all contributing interns and freelancers, winning awards at the German Developer Award 2016 (picture by Jennifer Pitton / www.expo-motion.com).
The year before, in December 2015, we couldn't possibly predict such a result after developing for only twelve months and just hitting the Alpha milestone. Alongside creating all content, we had been continuously working on our vertical slice – the level that would later become mission five “Killing Lord Yabu” and had always been as polished as possible, serving as a quality reference for the rest of the game.
When hitting the Alpha milestone we had realized that the vertical slice was actually much more beautiful than we had pictured the game in the beginning – and therefore also more time consuming to create. We simply learned a lot in these first twelve months, be it about our tools, about the game itself, or about how the graphics make their best impression.
Internally recorded gameplay footage, original pitch
Internally recorded gameplay footage, vertical slice after 21 months
We realized that it was not an option to go back to the pitched graphics, but that applying the new level of quality to the whole game obviously wasn’t part of our original schedule. At this point, we started investing all our company’s financial reserves into the project to extend the production timeline by three months. Daedalic Entertainment, our publisher, was able to extend the timeline, allowed us to cancel all planned DLC-levels, and fund another, fourth, month. As a side-note, Daedalic’s background as a well-known developer (e.g. for the Deponia series) helped a lot in that situation, because they’ve faced similar issues many times before. Adjusting the schedule not only meant improving the visuals but also heavily improving the liveliness of every level.
Back then, when we made the decision to finish the game in the best way possible – both internally and with Daedalic completely unanimously – , I told the team very clearly that we would have to go “all in”. In case of failure, it would have been “Game Over” for Mimimi. Still, there was no other way for us, being driven by passion and such. And while that investment paid off greatly, including scoring the highest Metacritic score for a German game since 2011’s Crysis 2, it also took a great toll on the whole team and is something we never want to experience again. More on that in the, you guessed it, “what went wrong” section of this post mortem.
2) Reviving a dead genre
Abé: Remaking an all time favourite game is certainly a dream of any game developer. Back in 2008, when I started my game design studies, I talked to Moritz Wagner (now Lead Game Designer at Mimimi) in the hallway about Commandos being one of my favourite game franchises (the other one is Little Big Adventure; I’m ready to take the contract). Moritz did some cool rap music videos where he was dressed like a ninja at that time, so naturally the picture of a ninja doing Commandos-like gameplay popped into my head. I told this to Moritz and we joked: “Haha, one day when we are big and successful we will make this game”. But it came out not being a joke. In 2014 we pitched the idea of “Commandos with Ninjas” to Daedalic and soon signed a publishing contract. Now we had the real chance to the revive the presumed dead real time tactics genre.
In the following I want to explain why I think we succeeded:
I’m a huge fan of the genre. As a fan and designer, I craved a spiritual successor to those games: the feeling of being in a strangely realistic place breathing with life; sneaking among countless enemies and taking them out one by one with the most clever plans, the most hair-raising moves or sometimes through sheer luck and fast reflexes. I was fairly confident about being able to deliver a game that other fans would love and often during production I would have a gut feeling whether a feature would fit the game or not. Of course that never meant I was ultimately right. Communicating those feelings to other team members and translating them into game design decisions could often be a difficult and time consuming process. But all in all, knowing exactly what the game should feel like in reference to its spiritual predecessors helped a lot compared to developing something completely from scratch. It should also be noted that Daedalic, while regularly giving valuable feedback, never had any creative control over our game – we basically were free to do whatever we wanted.
But enough talk about feelings; let’s look at the analytic side of things. In pre-production we analyzed the evolution of the genre and searched for concepts that are still unique today. We often compared the Commandos series entry title Commandos: Behind Enemy Lines with its own successor Commandos 2: Men of Courage, the most critically acclaimed representative of the genre.
Pyro Studios' 2001 game Commandos 2: Men of Courage
Both are stealth-oriented real-time tactical games staged in WW2 where you control a group of highly specialized soldiers taking on sabotage missions in enemy territory. Commandos: Behind Enemy Lines introduced us to to a cast of playable characters with very diverse skillsets that depend on each other and forced the player to combine their different powers. It feels a bit like chess, a game that has a very limited set of actions and strict rules but leads to countless emergent possibilities for play. Later, Commandos 2 had a more realistic and open approach and gave the player way more skills and possibilities. It removed artificial restrictions like for example only a certain character being able to use a specific weapon and added possibilities like looting enemy gear etc. So with Commandos 2 also being the impartially better game based on user and press reviews, it seemed obvious to take it as the main inspiration for a spiritual successor; but we didn’t.
For us, Commandos 2 watered down some of the series' uniqueness established with its first entry. Advancing it from a chess like game to a more realistic simulation, more complexity had been added. For example, by putting way more actions at the player's disposal they eventually made the game harder to grasp. This was not the way we wanted to create more depth in gameplay. In consequence, we defined our design pillar “Creating depth without adding complexity” which we aimed to uphold for all aspects of our game design. The “realistic” approach taken in Commandos 2 also made all player characters a bit too powerful, eventually harming the unique team play aspect created before by restricting realism in favour of gameplay balance.
Let’s have a look at an example that made it possible for players to even abuse the newly introduced systems: The Diver from Commandos 2 has a very powerful knife throwing skill which is balanced by having to retrieve the knife after throwing. The Green Beret also has a knife which he cannot throw however. Due to the inventory system introduced you were now able to transfer the Green Beret’s knife to the The Diver, making him an overpowered dual knife throwing killing machine while the otherwise mighty Green Beret is more or less downgraded to a bystander.
In Shadow Tactics we focused again on a limited, distinct feature-set per character with no “transfer options” and designed them in a way that every character skill was strong in combination with skills of other teammates but also quite effective on its own. All in all, focusing on the debut title of the genre and evolving it in a different direction than its successors helped us revive the genre in a way that feels unique and fresh today.
3) Creating a new IP in an old genre
Abé: Reviving a “dead” genre is often linked to reimagining an existing IP. This was, for example, the case with XCOM: Enemy Unknown which brought turn based tactics back on the map in AAA scale. Not having the means, the opportunity nor the rights to do so with the stealth real time tactics genre, we had to invent our own IP. This bears the risks that fans may not like the new title in the same genre because it lacks important things from their beloved game like the setting, the characters, the artstyle, maybe even just that big IP name that gets you all warm and fuzzy inside. We had to deliver something new for all of these.
Regarding the setting, we took the risk and made a drastic change with staging the game in Edo Japan. We were confident in this decision as we felt it was crucial that setting and gameplay align in the best way possible so the players never feel alien in what they are doing. The image of a ninja sneaking around and backstabbing a whole army can be considered to be an even more convincing stealth character than a WW2 soldier; who is more likely pictured shooting. Because the stealth ninja stereotype is so strong, we were convinced from the beginning that this was the perfect match for the genre.
With our chosen art style of Shadow Tactics we embraced the new setting by referencing Japanese ink drawings. This helped us to create a unique look that is instantly recognizable in each and every screenshot you may come across. This is something we are always aiming for when developing a new art style which is easier to achieve with more stylized than photorealistic graphics. Whether doing so for this genre was the right decision will be further discussed under What went wrong: Art Style versus Genre.
So with everything aligning to the new setting and ninjas, why would the game be called Shadow Tactics with only a light reference to the setting in its subtitle Blades of the Shogun – we don’t even have “Ninja” in the title. Truth is we and Daedalic wanted to think big and create an IP that allows for more than one setting. This is something very common nowadays in larger productions e.g. Assassin's Creed, Total War etc. For a small studio this is a bit more risky because it often means your game title is more generic. But we love the idea of exploring different settings, and we now have an IP that lets us do that. So if we ever get the chance to make “Shadow Tactics: Legions of Rome”, that’s one less thing to worry about.
4) Bringing the genre to consoles
Abé: From the beginning it has been an important part of our plan to bring the genre to consoles. This had been tried only once before with Commandos 2; more or less successfully with porting mouse and keyboard controls rather directly. We were quite confident in delivering a better overall experience on modern consoles but from a design perspective we still saw our biggest challenge in getting gamepad controls right. The other big challenge had been performance on consoles.
Right from the beginning, we knew we had to develop the controls for PC and consoles in parallel to have an equally good game on both platforms. At the same time it has always been clear that it wasn’t an option to let our prime platform, PC, suffer from possible limitations due to the gamepad controls. In the process of design and implementation, we became more and more aware that checking features against the gamepad controls helped us figure out the essentials of our mechanics and streamline them.
As an example: after killing an enemy you most likely have to pick up the corpse to hide it somewhere else. In our first design, killing and picking up the body required two separate user inputs. But when prototyping this feature for gamepad we figured that those actions can be done with only one input; a long press for a kill and the immediate pickup. Of course we then implemented this feature for the mouse and keyboard controls as well, where you can hold down the ctrl key (pick up and interaction key) to do the same.
Regarding the technical aspects of the console version, our biggest fear was the framerate and the quick save/load times. Since Shadow Tactics was our first inhouse console project, we had next to no experience with consoles. On top of that, Unity 3D, the game’s engine, did not have the best reputation on console at the time. But since we had been working with Unity for the last seven years, losing any expertise by switching to an engine that had a better reputation on consoles was not an option for us. For some time we developed PC and console versions in parallel and constantly checked whether at least an acceptable performance could be achieved on console. However, at some point we decided on a non-simultaneous release because we needed all resources for the PC version.
After finishing the PC gold master, we immediately jumped back on the console version and again started testing builds – with horrible results. Some levels only achieved constant 15 fps in areas where you would play for roughly 30 minutes; as was the case with the start area of “Matsuyama City”. The “quick” save/load times were just as bad, with saving taking up to 10 seconds and loading ranging from 50-60 seconds. Our task was clear: we needed to aim at more than twice the framerate, make saving around 5 times and loading around 20 times faster. All without any graphical downgrades. A typical “kill me now” moment in a game developer's life.
(Matsuyama City Level: One of the worst case scenes on consoles)
We improved the save/load times by pre-generating the serialization code into our code files and caching everything that was possible at runtime. Loading was still not fast enough though because recreating the GameObjects and adding their Components on the Unity side needed around 15 seconds. In the end, we got around this by dynamically creating template GameObjects which already had the needed Components on them and cloned them. The final results even surprised me a bit; saving now takes 1-2 seconds and loading 3 seconds in the worst case.
For optimizing the overall frame-rate, we did two major things.Firstly we custom batched our models into larger chunks that roughly fit a screen in our camera perspective to reduce draw calls and culling operations. Secondly, we flattened the scene hierarchy, which is very important for moving objects, in order to avoid multiple matrix calculations. After doing tons of other optimizations, we ended up with a more than steady 30 fps on PS4 and XBox One. Also, Unity’s William Armstrong and John Elliot of the Spotlight Team supported us a lot with the console version and custom Unity builds, making sure that we hit our internally set goals. Thank you!
Rewriting and optimizing our part of the code in a way that the engine needed to do as little as possible, we managed to improve the overall performance. An important lesson we learned is that you are the only one who knows how your game works and what you need. A commercial engine like Unity does a lot more because it is not designed for a single game or genre so it is your job to leverage the load of the engine. Our talk at Unite Europe 2017 with our Technical Director Frieder Mielke and Gameplay Programmer Philipp Wittershagen goes into much more detail here.
In the end we succeeded and developed a very well performing console title with Unity on Xbox One and PS4 without sacrificing graphical quality. Therefore we are super happy and proud of the consoles version which has the same high ratings as the PC version.
5) Marketing the game, reaching hardcore fans
Roth: Unlike with our previous game The Last Tinker, we knew exactly how our target audience for Shadow Tactics would look like. The game was designed to be hard from the start and had to appeal to the original fans of the genre. Marketing Shadow Tactics still proved to be quite difficult, as we were facing three major challenges throughout the production:
- Reaching hardcore fans: People who love the old games of this genre are much older now, more difficult to reach and might not even play games anymore.
- Reaching a new audience: Reviving a dead and non-existent genre means that younger players don’t know the genre at all. Also, Commandos and Desperados sold exceptionally well in Europe, but many US gamers (one of our most important markets) never heard about the games.
- Selling the visuals: Given the inflexible camera perspective and real-time aspects of the genre, it’s highly difficult to create any kind of cinematic feeling in trailers or gameplay footage. Games like XCOM: Enemy Unknown and Mario + Rabbids Kingdom Battle made some great steps into the right direction to overcome the limitation of their genre (action-shots, extensive cutscenes), but after thinking about similar concepts in our real-time based gameplay, we realized that those wouldn’t work out or might even be distracting. We were also a much smaller team on a very limited budget, so visual fidelity wasn’t on the right level for anything besides the usual game perspective.
Given these challenges, we struggled a lot to solve them. This is what we think worked out:
When attending events like PAX and gamescom, we often received feedback from genre-newcomers that our game was an “isometric Metal Gear Solid” or a “hardcore Assassin’s Creed with multiple characters at once”. While that was very flattering, we knew we would never be able to compete with these games nor reach their audiences. On the other hand, we also quickly learned that nearly all gamers who gave Shadow Tactics a try – even if hesitant at first – quickly felt in love with it. And while this commonly happens at trade shows, we felt the huge difference compared with positive reactions to our previous games.
This convinced our Communications Director, Dennis Huszak, to internally push for a demo: the game would have to sell itself. We had to believe in the quality we delivered and let people experience it first-hand. No trailer would ever be able to have the same impact, so after deciding for going that route, we made the free demo release our main call-to-action of all messaging.
“Play the demo!” was our main call to action, even when the game was released.
We released the demo on November 21st, 2016. That quickly increased the pre-release-wishlists from 15,000 to 40,000 within two weeks, when we launched the game on December 6th. For reference: The first 15,000 wishlists had been cumulated within nearly seven months. A day after launch we had 150,000 wishlists, a year later roughly 700,000. We didn’t have any pre-order campaign, so we think that all wishlists before launch were much more valuable; essentially a “I want to buy this on release”-flag instead of “Waiting for a sale”. We’ve sold more than 400,000 units on desktop platforms and the demo was downloaded more than 160,000 times.
Suddenly, our communities became much more vibrant. Our Reddit AMA was a huge success for us and even our Facebook Live Stream reached roughly 31,000 users according to FB statistics, so was worth the effort. Releasing the demo was the best way for us to tackle all three issues, and the community loved us for doing that in 2016. And even more: Each time someone would come up and ask about how the game played, someone in the community would respond by telling them to simply play the demo we had put out there. We also publicly embraced the fact that we needed everyone’s support to spread the word for our unknown game. This way, thanks to our very first fans, we’ve especially reached many of the original Commandos players.
Still, for the record: Deciding to release a demo wasn’t an easy thing. Dennis really had to convince us of his idea, because I was afraid that gamers would already have enough after one level and move on to the next big AAA game available for $10. With general games industry wisdom regarding demos telling us that a demo would hurt sales, it actually was a huge leap of faith for us.
What Went Wrong
1) Going all in, achieving near bankruptcy
Roth: As stated above, the launch of Shadow Tactics was a dream come true: we made a successful game with better ratings than ever expected.
Also, two follow-up projects were canceled shortly before the game’s release, we were nearly bankrupt and half the team, including some of our earliest team members, had already received their termination to ensure that we could finish the console version with the money left and a minimal work force.
While the “all in”-success-story probably sounded romantic in some way, it was an absolutely awful experience. And we had predicted it: by adding four more months to the project, the game became much better, but all our savings were gone. When we realized that our backup plans for new projects didn’t work out as planned, we knew we were screwed. Because we exclusively focused on Shadow Tactics, we didn’t invest in any new pitches and had nearly nothing to offer to new partners. On top of that, they understandably waited for the release and the first weeks of SteamSpy data before they wanted to sign a project.
That left us in a horrible spot to be in: the whole team walking onto the stage in front of the whole German games industry, winning awards, being happy and at the same time knowing that it might be the last month of the studio.
Part luck, part delivering a decent game, we had been able to talk to publishers after the release and none of them took advantage of our situation or tried to rip us off with a bad deal. Negotiations that normally can take up months went through quickly and fair. In the end we surprisingly had to choose between projects, all of which we’d wholeheartedly loved to develop. I vividly remember being on the train on December 23rd, the very day before Christmas, and writing the final email to the team, letting them know that the contract had finally been signed and we were safe for the time being.
Mimimi after … getting rid of all traumatic paperwork.
We then got back together in early January, we ritually destroyed all paperwork which somehow had to do with the potential demise of our studio. Which sounds like fun, but speaking for myself: the whole situation still weighed heavily on me for the next six months. Even though everything was back to normal, I felt restless and had trouble sleeping, which took me weeks to even understand.
Mimimi was always a kind of safe haven, never having to let anyone go due to insufficient funds, always having at least a budget worth three months on our account. This was the first time we had been really unstable, and we knew it months before. Adjusting the workload without increasing the budget from a safe source was a risky and ultimately bad decision, because the toll this whole mess took on us was much higher than anticipated. I consider it a personal failure in not securing the potential follow-up projects with specific deal terms and signatures. This is something I normally always do, but for personal reasons and by trusting people I didn’t do at the time.
2) Lacking the resources to develop new content
Roth: As mentioned above, we had to cancel all planned DLC levels during production to be able to completely focus on the main game, which was the right decision for the initial release. But being such a small team means that when we sign on a new project, we don’t have any resources left to work on our current release. Things like releasing patches or adding new languages are already a huge burden. Although many fans are asking us for additional content and we’d love to deliver it, we simply can’t.
There are two major reasons why:
- It’s incredibly expensive to do so. The game features 13 different languages and completely spoken English and Japanese dialogue. We would have to translate the DLC to all of them and invest heavily to get back the original cast for the voice over recordings. Bug testing and running full QA would be another huge investment (this genre is a beast to test, due to the sandboxy structure of all the levels). Although Shadow Tactics is generating revenue, it can be questioned if we could ever recoup the investments.
- Our players and the Mimimi team would expect the DLC to be on the same level of quality as the main game. That means that our most experienced team members would have to be heavily involved instead of, let’s say, just finding an outsourcing partner or hiring new personnel. But everyone is already on the new project, which has even higher expectations to fulfill, and we are neither fans of crunch nor did we find any available cloning devices. For the sake of simply making sure that our new game won’t disappoint, we have to stay focused on one big task.
So while the financial aspect plays a role, the dealbreaker here is really the available time we have. We would never consider releasing a subpar DLC just to make money. We probably should have done the DLC right after release, but hindsight is easier than foresight, and at that moment it simply didn’t happen.
3) Breaking console conventions
Abé: Bringing a classic PC genre with mouse and keyboard controls to consoles is a real challenge that thankfully some great games like Diablo 3 and XCOM: Enemy Unknown managed to overcome. While designing the console control scheme, we tried to support the habits of console players as much as possible because we did not want to overwhelm them with a new genre and new controls. This was not really possible though because some systems like freely moving, rotating and zooming a camera are rare on consoles and there is no controller standard for them. In the end we came up with the following layout on PS4.
Shadow Tactics PS4 controller layout.
There are some “natural” fits like square for killing and cross for jumping; some other common mappings like the shoulder buttons L1 for the skill-wheel and L2 for aiming. And then there is the right trigger R2 for activating the special camera mode which is not common at all because it is mostly mapped for shooting, acceleration or some other important gameplay input.
So why did we choose the trigger? Because we needed a button that allowed stick movement while pressed and therefore only shoulder buttons were a real option and all other buttons were already taken by some more “natural” fits. Understandably this breaks the expectations of a console player and maybe it is even more problematic as the rest of the layout tries to stick to the “console conventions”. On the one hand we are still not sure if it was a good idea or not to map a complex and new feature like our camera controls to a button that players are used to doing so many other t