A game developer's story with Vanillaware's Kentaro Ohnishi

We all know the big players, the Kojimas, the Miyamotos, the Suda51s; but the magic of video games is about more than just the front men. It's about the details, the microcosm of daily working life, and the less recognisable figures that somehow contributed to the industry at large.

Somewhere, down a narrow alleyway, up several flights of stairs, in an unpretentious office where men and women sleep on keyboards and under mountains of sketch filled paper, accost soda machines for not fulfilling their vending duties and shoot hoops with overflowing waste paper baskets, is a collective of individuals with a story worth telling.

You may not know Kentaro Ohnishi, and he's okay with that. Today he's a programmer at Vanillaware, and one of director George Kamitani's trusted associates - but it hasn't always been this way. He has his own story to tell about where he began, his inspirations, and how playing Starcraft on lunch breaks can end up being a foot in the door.

With Odin Sphere Leifthrasir heading to PlayStation 3, PlayStation 4 and PS Vita this month - Vanillaware's HD remake of their well-received PS2 fantasy action adventure - Mr Ohnishi managed to escape the office to talk to us about how he started at A and how he's still getting to B.

Thanks for taking the time to talk with us today. Can you give us a brief overview of your current professional profile?

I would like to start by saying that this is the first time I have had an interview with foreign press. Thank you so much for taking an interest in my work.

Regarding my career history, after graduation I started working as an intern at the now well-established developer Racjin. My first job was porting the shooting game R-Type to the PlayStation. After that, I met up with members of Atlus Osaka (Osaka also being Racjin's base), and it was there where I met the now President of Vanillaware, Mr. Kamitani.

At first he was just was just a lunch break Starcraft buddy, but later - somewhat thanks to this gaming relationship - he called me up when he went to work on a project for Sony and we have been developing games together ever since. After we finished making Fantasy Earth in Tokyo we returned to Osaka and started Vanillaware. I worked on the player and stage programming for Odin Sphere, Muramasa: The Demon Blade and Dragon's Crown. And now, even though I am the Director of Odin Sphere Leifthrasir, I still handle the player programming.

Because Mr. Kamitani writes the scenario, does the art, and also has his hands full with the overall direction, multitasking is often the norm in a small company. As such I also work as a planner in addition to programming.

Which titles inspired you to get into the gaming industry?

Being part of the so-called Famicom Generation, as a child I played Super Mario, The Legend of Zelda, Dragon Quest and Super Metroid. The classics released during that era were burned into my memory and, as the hardware continued to develop, I gradually came to think that I wanted to be a game creator. I can't really say that there was any one game that led me to this career, but one event that really made me realise I wanted to work in the industry was the Street Fighter 2 boom. I was really impressed by the depth of that game's systems.

I understand you studied programming in your spare time when you were still at school. What were some of your homegrown projects and what hardware did you use to practice programming?

Sadly there isn't anything left of my home-grown projects now, but I remember making a horizontal shooter where if you didn't hold the button a shield would appear. If you fired as normal, your ship would shoot out in front of it, and, if you held the button down, you could charge a cannon. Another game was a simple quarter-angle view dungeon crawler.

The hardware I used back then wasn't even Windows-based! It was NEC's PC-9821. It was the first time I delved into PCs, but at the time the machine cost more than 300,000 yen, so I couldn't afford it by myself and had a hard time convincing my parents that it was necessary. Even though the resolution was 600x400, which was considered high at the time, it was different from DOS/V in that it couldn't display pixels individually and could only display 16 colours. It wasn't really a machine geared for games and was very standard for its time, but there were many books about the hardware and I felt that the configuration was straightforward. Rather than actually gaming on the machine I remember spending more time creating a system that displayed sprites, a tool for making sprites and a map editor.

You mentioned you got your first opportunity when you joined Osaka based Racjin Co. Ltd, a company responsible for Bleach licenses and a few Bomberman spin-offs. What was it like working there as a young programmer?

The reason that I began working at Racjin was because they had won a contract to port R-Type and I had learned programming on the same PC and CPU - not because I was a skilled programmer. At the time I didn't have knowledge of 3D computer graphics, which was becoming mainstream, so I hastily bought a book on 3D programming and studied on the train commute to work. I wasn't someone with great ability, but rather someone with perseverance. Because I entered the project right in the middle, and was then constantly put on different projects during the process, in just three years I was able to work on six projects that offered unique experiences. It was very interesting.

Do you have any stories about your time at Racjin?

Yes, there was a time we were working on a game called JET de Go, which was a commercial airplane simulator. I was young and didn't understand the appeal of a simple flying game where there were no missiles or enemies. I remember asking if we could put in events where a passenger would get sick, so you had to make an emergency landing, and even suggested a hijacking event! But those ideas were declined. In fact, even if the player made a mistake, I was told the plane couldn't fall out of the sky or explode and I must simply have the game fade out with a "Game Over" screen. I remember thinking, "What part of this is fun?" If they would have let me, I probably would have put in an ending where you crash into an alien mothership.

Eventually I realised I needed to understand what it was flight simulation fans wanted. There were only three people on this project and we had to work all night and stay at the office for about three months. During this period I remember thinking, "Primitive people didn't bathe... I wonder if it was like this? Perhaps it's something you get used to in a couple of weeks."

One thing they did let me do in order to gain experience for the project was to use JAL's four billion yen flight simulator. That was pretty cool.

One of your early works is Trap Gunner, a 1998 Atlus published title for the PlayStation. It's a unique game that was well received by critics. Can you tell us a little about what you did on Trap Gunner and some backstory about the game's development?

It was a long time ago, but that's still a title I love (so much so that I would like to remake it with online PvP!). I joined this team after working on R-Type and the lead programmer told me to work on effects (at the time, we worked on programs one by one). I worked on explosions, light flashes, poison gas and other things like that. It was fun. Effects are simply visual performances that enhance the game's appearance, so I didn't have to worry about how they would affect the gameplay and could do whatever I wanted. The only thing to worry about was the load placed on the processor. I added effects for things I wasn't even told to. Areas on the ground where bullet shells could roll around, billboards that would always appear towards the camera etc. This taught me the basics of 3D and it was also fun to make things that showed up in the visual areas of the game.

By the time I joined the project it was already in development for over a year and they were facing problems. Even though it was a top-down game, the stages were composed of many layers that were arranged vertically and were confusing. Also, although the game was one versus one, the stages were extremely large and sometimes the players couldn't find each other. There were only bombs and poison gas for traps and the player only had a handgun and it just wasn't all that fun. In spite of this, the planners and programmers did not compromise and spent many hours playing-testing the game. In the end it turned into a good game. I think that the experience of this process was very important for me. I worked on extra traps and programming the bullets. The force panels and homing missiles were really fun.

Do you have any amusing anecdotes from your experience making Trap Gunner?

One that springs to mind is when the programmer in charge of player AI and the planner would often play against each other, but the planner would always lose. It seemed like he would be playing very carefully, but then he would make a rash move and get caught in a trap. Every time that happened, the programmer would burst out laughing, so the planner's face would turn beet red and he would say, "One more time! One more time!" That was really funny. He ended up with an 88 loss-straight total, never winning a single game!

These days you're chiefly a programmer at Vanillaware. Can you explain the different challenges between programming 2D and 3D software?

With the words 2D and 3D it seems like it's just a difference in dimensions, but I believe the difference is greater than that. 3D games must be rendered using a camera and shaders. What is then displayed on the screen is simply the end result of that. The angle, zoom and even lighting changes are rendered based on the player's control inputs. The designer cannot tune the data that produces the rendering and the programmer can only work to improve the rendering quality. What finally determines the visuals is the game hardware itself. You can obtain the ability to express overwhelming freedom, space and photorealism, however artistically expressing solid shapes takes much time and effort and still ends up unnatural somehow.

With 2D graphics you can arrange the screen so that it appears just as the artist drew it. Naturally, there is an intense charm to this. As a matter of fact, this is 2D's primary advantage. So, when dealing with 2D, unless you directly use the charm of the image as it was drawn by the artist's hand, there's no point.

Beyond that, the programmer's job is...indirect, with few glamorous tasks. In our company the situation is that in order for planners to do more minute adjustments, programmers continually improve animation tools and separate a large amount of the animations though complex branching processes, foregoing the desire to make programs that have a simplified structure. However, in terms of expressive elements, for Dragon's Crown we worked really hard on a pixel shader that was something like a gamma corrector that created the sunlight patches and reflections on water (laughs).

For Odin Sphere Leifthrasir, we tried making something that created shadow on the floor and walls, but unlike in 3D, because there are no actual floors or walls (there are floors and walls that have been drawn, however) we didn't know how far to extend the shadows, so we ended up having to set parameters for the floors of each map. Also, unlike 3D where the camera is set to view great distances, the character's field of view is only as narrow as the width of the TV, so we always have problems with the camerawork. Basically, the area where the character is looking is displayed broadly, but if you're in the centre of the screen and an enemy attacks out of some invisible space, it isn't really fair, so this also affects how we program the AI of the enemies.

Vanillaware director George Kamitani's visions are becoming increasingly complex, as we saw with Dragon's Crown. What's the most challenging programming task you've faced at Vanillaware?

Actually, the most difficult thing programming-wise I dealt with was for Muramasa: The Demon Blade and thinking up an action system that allowed for cancels to be made freely. Basically, our character control system was designed so that the animations made by the designers can be beautifully connected, so we had to merge this system with one completely opposite. On the design side, we received instructions that said, "You can cancel the animation from here and at that point you can cut into this point of the next animation and join them this way" and then attempted to implement it. That is how the cancel system works. The programming source code at first glance would make you say, "Why is this so roundabout?" and actually at one time another programmer, without understanding the necessity of the system, said, "This is some dirty source code" and deleted it! (laughs).

The difficult part of Dragon's Crown was, more than anything else, getting it to work online. It was very difficult making the response time for high-speed cancels keep in-line with network's response time.

Odin Sphere Leifthrasir is an HD remake of 2007's Odin Sphere. Is the process of creating 2D graphics in HD quite difficult? Did the conversion require any redrawn assets, and what does the process of converting it into HD entail?

Actually, making this game HD wasn't all that difficult. This was because the original assets were painted using pen tablets and painting software and these assets were double the resolution of the PS2. These large images were shrunk to make them look better, so the main work this time was to have them be displayed as they originally were created. However, the main characters' faces and the like were still the PS2 ones, which were compressed, so Mr. Kamitani actually went to work to make sure that faces looked the same for this HD version. This isn't about about the remake specifically, but when the designers started working on the HD remake, there were sections that they felt were made inefficiently according to their current standards, so although the visuals aren't any different, there are some elements that were remade.

We know Odin Sphere Leifthrasir will feature new difficulty levels and boss rush modes, but have you tweaked any aspects of the gameplay?

This is something that really bothered us, that there were some customers of the original Odin Sphere who couldn't play to the end of the game and ended up selling it. We felt very badly about that. There were many people who just made it to the end of Gwendolyn's chapter and then quit. This was because they got sick of the unreasonable difficulty.

For example, in the PS2 version, increasing the attack power level and the HP level were different, and even though there are important HP raising elements like cooking or seed planting, those players who were most interested in the combat found these systems tiresome. Because of that, as they got closer and closer to the end, their Max HP ended up being much less than what we expected them to have and the game ended up poorly balanced so that they player could be killed in one hit. With Alchemy, its power wasn't related to the player's level, but rather was dependent on the current chapter in the tale, so it was easy to use it to become very powerful. The players who figured that out ended up breaking the balance in a different way. At the time, we didn't understand that "players should play however they want to."

This time the game is balanced around the concept that, no matter how the player decides to play, we want to have them to reach the end and savor the story. In the case that the player's level is much lower than what we thought, the amount of EXP received is increased and even though it will be more difficult than normal, it's balanced so that the player won't get stuck. The balance is easy so that those players who carefully use the cooking system and aren't good at action will be totally OK. As before, magic leaves are powerful, but this time the leaves use a 'grading' system, so as the player gets closer to the end low-grade magic leaf power will be so low that they will be practically useless. Now, if a player doesn't intentionally go to great effort to use Alchemy, they won't have the same kind of overwhelming power at their disposal. Also, we added action skills for players who don't like to rely on items or don't want to use Alchemy, so there are many ways to play. Once the game is completed, the number of enemies is increased and there is a "Second Playthrough" mode added in addition to the "Hell Mode" that was in the original.

You were both programmer and director for Odin Sphere. How much influence did you take from Vanillaware's early Saturn game, Princess Crown, in creating the game?

Originally, the Odin Sphere project started as Princess Crown 2. Therefore, there are many settings that make one think of Princess Crown. There's the setting of a little girl reading a book, that feeling of the story taking place in a fairy tale, your power decreasing when you swing your sword and you being unable to move, lots of openings in normal attacks so that item usage is important, planting seeds, cooking, and the thought process that it's fun to go out of your way to do things ritualistically, in a manner of speaking. This is characteristic of Mr. Kamitani's thinking. For the creation of Leifthrasir we thought about the "origins of Odin Sphere" and tried not to cut things while still making it fun to play.

On the programming side, the concept of large characters with smooth animation was carried on from Princess Crown, so the technique of animating we accidentally found at that time was greatly improved upon and strengthened.

Leifthrasir has also been incorporated with the newest systems from Dragon's Crown.

Was there anything in Odin Sphere that you wanted to include first time around but weren't able to?

Several. One was a randomly generated dungeon that you could play after beating the game. In fact, this is something that the stage programmer was already working on, but wasn't able include. Another one was a one versus one online minigame called Erion Wars. Each player would have their base at the opposite ends of the map and you would send out troops (the game's enemy sprites) and they would advance automatically, while enemies who reached the base could be attacked directly. It was action plus simulation. We considered this, but couldn't get Mr. Kamitani's approval (and didn't really have enough time).

Also, the main story of Odin Sphere is wonderful, but there are many scenes that aren't shown (King Odin's past etc.) so we had the idea to expand upon those and make them into cutscenes and create something like an archive separate from the main game where they could be viewed. Due to time limitations and story consistency, we weren't able to include this.

There was also an idea to add an "Ingway" chapter [Ingway being an Odin Sphere character] but he doesn't have a Psypher, so it wouldn't match the game system and it would have been difficult to angle from a story point of view, so it wasn't possible.

Can you give us an inside idea of what it's like to work at Vanillaware and with Mr. Kamitani? With such a small team you must need real camaraderie to bring such impressive projects to life!

For better or worse, Vanillaware is a loose company where Mr. Kamitani creates the art and scenario, but is the type who doesn't give instructions (he simply gives a general overview), so each staff member must think about the game themselves and works towards the project's completion. There are no instructions like with model kits that say, "Insert Part A into Hole B".

For example the concept for Dragon's Crown's skills and what the skills were, as well as the Diablo-like treasure system, were thought up by me, while the enemy character's movesets were made by the animation designer and AI programmer. Judging whether the game was fun or not, and everything from the movement to the balance was done by me through trial and error. Mr. Kamitani is the type of person who is happy to use ideas thought up by the lowest person in the hierarchy if there are no foreseeable problems. He will never say not to do something without his permission. However, people who don't like working without clear orders or don't like doing work outside their job description aren't a good fit for this company.

Also, one idea that Kamitani-san constantly brings up is that of "company branding." Nowadays, there are many free diversions like smartphone apps, video sites and social media. If you want to kill time, there are many ways to do it, and within that sphere, console games are relatively expensive. If you don't enjoy the game you spent good money on, you won't buy another one. In the worst-case scenario, you may end up disliking games and never touch one again. We don't like that outcome and we want the person to say, "I like this company's games so I'm going to buy this."

All of us believe that we can make our own futures.

Finally, you were programmer on Odin Sphere, Muramasa: The Demon Blade and Dragon's Crown: Which of those projects is closest to your heart, and why?

Hmm, as a programmer, probably Muramasa.

For the world setting and game system I would say Dragon's Crown, but after some of the negative comments regarding the PS2 version of Odin Sphere, Mr. Kamitani and I wracked our brains and Muramasa was the title where we totally changed our thinking regarding programming structures and games. There was so much trial and error with Muramasa, I have a special attachment to it.