Hatsumei Mage – Proper Post-Mortem

We all seen that kinda movie where main character goes from being lucky and happy to being loser with nothing left by the time movie timeline hits 70% just to fill the rest 30% with non-stop process of reclaiming what was lost.

It was only 16 hours before the submission date when something sparked inside my brain. It was a third idea for a game. It had to be good or implementable at least, or otherwise that’d make LD#28 my biggest LD failure yet.

Then there was 16-hour long crunch, powered by nicotine, caffeine and instant noodles. And a surprising way 3 year-old Blender lessons paid off.

We can’t draw that much

The beginning was splendid. A programmer/modeler, a programmer/game designer, two artists, one highly talented musician – I believed we could take on the world with that kind of team!

We decided to make a game where you control a stupid boy, with his brother’s head in his backpack. You describe things to this boy, since you’re a smart head that needs its body collected before you can go on a date. With these descriptions active, your brother semi-automatically reacts on the world around him. You tell him to eat everything red – and he eats apples and landmines alike.

The idea got fleshed out and turned into a kind of LucasArts classic adventure with semi-logical solutions to obstacles and a completely linear path. A trivial thing to implement; a non-trivial thing to draw.

I really didn’t see how bad this idea was until our first day was fully spent on discussions, and half of the next day on attempts of drawing a simple skeleton of a main character. We needed sharks falling, Jesus running from Minecraft dude, a whole lot of scenes that didn’t make sense and had to look good.

It got even worse when our second artist told me she was really busy. I called our game designer and we spoke about things we could implement. He though simple games were the way to go. I thought his older idea he suggested for LD#27 was the way to go. I ignored his advice and scrapped the first prototype.

I was ready to make the best entry ever, the one with sandbox elements and physics – just the kind of a game I love.

Sure, put the joints all over the place

For LD#27 our game designer suggested to make a game about constructing a rocket. You have loads of junk and you put those together in 10 seconds.

Rockets. Physics. Sandbox. Sounds awesome, right? Well, let’s make it better, dammit!

Let’s replace rockets with cars.

Sure, that’s really simple. It’s not like cars are very different from rockets. Except that placing wheels correctly is one of the most important tasks for a player, while placing thrusters on rocket can be done in any chaotic way you want.

Wheels themselves are also terrible. PhysX wheel collider tends to rotate randomly sometimes, and since raycast is the way to determine the collisions, what falls underground stays underground. Shamanism with the center of mass was also kind of annoying, since in one case it yielded perfect results, and in a bit different case car would flip over any time it find appropriate.

And then there’s a junk pile waiting to become whole with the rest of the car. Why bother with making a base carcass for a car when you can make one outta ponies and GameCube controllers? Turns out, there was a reason to bother.

Transferring car between a racing and a construction scene required some ugly hacks that changed hierarchies, switched colliders and did a lot of stuff most countries, except for Japan maybe, would call «child porn».

Also, since placing wheels was a task worthy of a dedicated game, I tried to restrict it to some base configuration. But that was so uncool and lame and player-limiting so I decided to discard this idea. No compromises here.

The last, but not the least, are joints. They are awesome. They are springy. They can break after a certain force is applied. And they lose their stability after a chain of three cubes is formed.

I was devastated. Day two was over and my own attempt at Katamari with cars failed. By that time I was the only working person left out of five and I was confident that no game will be done.

You only get one day – at least most of it

Depression wouldn’t leave me alone that easily, just as my desire to fix the broken car prototype. Eventually I decided to forget about LD and watch some “Twin Peaks” instead.

It was 16 hours before the submission date when I felt an urge to take a dump. My thoughts were all about how I can’t make a game and how this LD turned into a disaster, but as soon as I sat on the toilet, I felt enlightened. An idea suddenly sparked inside my brain and it was good. It was just the kind of game that should’ve been done right from the start.

I ran out and approached our artist. Eyes burning, I told her we’re going to make a game. It was the moment when “LD#28: The Movie” timeline hit 70%.

The game was about the sorcerer that used a single word to finish his “You only get one” spell. You could leave enemies defenseless or unable to fight. There were tactics and simplicity alike in this idea.

We bought some energy drinks and entered crunch. She was making models in Blender as fast as she could, while I was making one of the messiest prototypes.

The second artist suddenly returned. She drew us some skins and screens and this was just the kind of help we needed. Character design was a clear copy of Bayonetta world map characters. Level design was about putting more nice decorations along the pre-computed path.

It was a strange thing that I was the one to teach our first artist modeling. Even stranger, it happened 3 years ago as a part of a project she helped me with. 3 years later, she’s suddenly just the help I need to finish the game.

As we progressed, the place we were at became more crowded. People brought booze and hookah, and started partying. I couldn’t hold myself from smoking some hookah, which perfectly combined with my lack of sleep and high caffeine percentage in my blood. I stopped understanding half of things I coded.

The party was over after a couple of hours, but then someone knocked the door. “Who’s there?” – I asked. “Police, open the door” – a voice replied. In an instant my only thought was “God, I hope I hid my weed well”.

Turns out, we were too loud. They took 20 minutes of our time and left, and I thought we could continue peacefully. Well, no way. As soon as I sat down with my laptop, a guy ran past me. It was our musician, drunk as hell and doing his worst to prevent himself from vomiting on everything around him.

It took us some time to send him to sleep and clean up the mess he left. We could finally breathe easily, until we realized we had 3 hours left.

No balancing was done, just some character prefabs with several variations of equipment. A “hand” word was implemented, as suggested by another drunk artist at the party. Animations were done in 15 minutes. Only a bit of testing was done.

And then timer hit zero. The build was uploaded. The artist fell asleep on the floor as she watched me submitting the entry. It was done.

It was done and I barely could believe we did it.

But why “Hatsumei Mage”? Well, take a look at the master and his creation here:

Make Hatsumei happy – play “Hatsume Mage” here!