HAMSTER APPARATUS v0.0 devlog

2026/11/05 HAMSTER APPARATUS cover

soooooooo i joined a game jam.

i don't really like game jams. historically, the overwork they force upon you has pushed me to burnout at record speed, and i'm too ambitious for my own good. i've dropped out of a lot of game jams because i couldn't make something i'd be proud of in the short timeframe. you can tell me not to over-scope as much as you'd like but my heart says the world has no need for more cookie-cutter games, and making something disappointing is more depressing than burning out.

if you're reading this post in the current month, you might notice it's the only thing on my website. that's because i was in the middle of redoing the whole thing when i dropped everything for this jam. in fact, i had just started writing my own static site generator in python instead of using any of the plentiful existing ones. why would i do that? it just seemed like it'd be valuable experience. i love programming and i'll take any excuse to do more of it. i had my work cut out for me, so when i came across the digital pet jam on itch.io and decided to try it, i felt like i was making a terribly irresponsible decision.

but consider: i got medicated in may. my threshold for overwork is higher than it used to be (i previously could barely manage a couple hours of work a week), and my head is the clearest it's ever been. no longer do i have to torture ideas out of myself, because suddenly they just flow out of me with no effort! and also, a pet sim? that's a genre i have a deep well of feelings for. working on something like that isn't even gonna feel like a chore.

and it didn't! this is the first time committing to a project hasn't felt difficult at all. after just over a month, i released a prototype called HAMSTER APPARATUS. it's downloadable for free, and has around 20 minutes worth of stuff in it.

screenshot of HAMSTER APPARATUS: the two characters are doing a jigsaw puzzle together

i've had this idea in my head since i was a kid: a simulation game taking place in a single apartment where you observe the relationship between two characters living together. the keywords "observe" and "relationship" had to be the bedrock of the entire project. see, simulations are my favourite genre of video game, but i hold nothing but hatred for most simulation games. i despise games where players are given absolute control over the world. i despise games where the NPCs have shallow inner worlds and no ability to influence anything. i want characters in simulation games to have more than a thin veneer of autonomy painted over their empty shells. i don't want any more player-oriented stories. thus, i decided to make it so the player observes the world through security cameras. the characters don't know you exist and you can't make decisions for them. your relationship with them is parasocial.

for the jam period, i wanted my focus to be on the game's codebase as much as possible. i set a deadline for when i would stop adding new features and pivot to working on the visuals, but until then, i did nothing but write code for 3 weeks. the character models were default donut meshes that slid around a world of grey cubes.

two grey donut meshes in a grey cube room watching a TV made of grey cubes

cute, right? they were so young.

the decision to make the game 3D was a pragmatic one. a game set inside a surveillance system would be much more interesting and dynamic if its objects and characters existed in 3D space, not to mention a lot easier to code and illustrate, at the cost of forcing me to get more familiar with blender. i'm a 2D artist and despite my years of experience making games in roblox of all engines, i never got very far in learning how to 3D model. this is a wonderful feature of game jams: they give you a low-stakes environment that forces you to try your hand at new skills, along with the opportunity to revisit and train existing skills.

in the first few days of development, the weirdest thing i noticed was how at home i felt. i've been learning game development skills basically non-stop since i was 17, and it feels like everything i've learned was leading up to making this game. the mechanics are like a mash-up of countless other abandoned projects and experiments i've worked on over the years. there actually weren't that many features in this game that i didn't already know how to write.

as an obvious example, a previous version of this site featured a virtual pet of its own. i had to write a fairly complex state machine/animation tree to accomodate all the interactions you can have with it, and since it was written in javascript, which is not designed to be a game development language, i had to be a lot more careful with how i coded it to make sure things wouldn't fall apart. thanks to that experience, i was able to get the characters switching tasks seamlessly pretty fast. see? learning things the hard way has a lot of merits.

i wanted the most detailed thing about the game to be the relationship system. when i tried to figure out what separates the simulation games i hate from the ones i love, the answer was simple: the ones i love have an interesting approach to simulating social dynamics in particular, and the ones i hate do not. there is no point in having more than one character in a simulation if their relationships can't evolve unpredictably. i tried to write the code for each action with consideration for how the central relationship can affect and be affected by all of it, but that's very high concept. focusing all my energy on that before the characters had their own internal lives was putting the cart before the horse. in the end, my focus shifted to creating a world for them to at least co-exist in at first. their home has some of the bare essentials to sustain life in it. they can eat, they can sleep, and they can do things to occupy the time between those things too.

the plants in the living room take up an absurd amount of space. back when the game was all blocky and everything had bigger hitboxes, teaching the characters how to navigate around the plants without getting stuck was the first logistics nightmare of the whole process. at first glance, they don't seem to have any impact on the relationship between the two characters, but i knew in my heart that their home needed to be full of plants. i like how they become a visual indicator of the health of the household. as soon as the characters' ability to look after themselves starts slipping, the plants start dying. each dead plant has a severe penalty on their happiness, which basically forces a soft game over that really isn't fair, but at this point in the game's maturity it doesn't matter much.

the grey cube room with a cube kitchen and 5 tall green cube-shaped plants

the plants are where the difference between the placeholder models and final models is the most hilarious to me. look at these fucking eyesores. this is what the living room looked like for the first few weeks. you might also notice one of the characters is on top of the counter. i was in the middle of configuring their navigation paths and she'd spawned there and gotten stuck. in the next playtest, she spawned INSIDE THE FRIDGE and had no way of getting out.

i added a memory system to give the characters some kind of internal monologue. the sims 2 has a lovely memory system that lets you read through the highlights of a sim's entire life at any time. it's one of the very few chances sims get to express their thoughts in their own words, and it's a major part of why the sims 2 will always be my favourite sims game. with how things work in HAMSTER APPARATUS right now, almost every action creates a memory that the characters hold onto. memories each have a mixture of positive and negative sentiments that tally together and influence their decision-making.

one character's attachment to the other is displayed in an in-game menu. her thoughts about the other read 'i hate playing cards with her, i can't stand losing.'

only memories that have text attached to them will appear in the attachments menu, and in this build there are only a few events that have text associated with them, so the characters actually have a lot more memories than it appears. it's a very experimental system at the moment and is yet to be properly fleshed out. positive sentiments are spread out so unequally that beneficial actions like eating and sleeping always end up scoring much lower in the priority list than actions like watching TV. watching TV especially needs a nerf LMAO, if you let them run around for a couple minutes, the TV will quickly become their favourite object in the house. the only thing saving them from sitting at the TV forever is the enforced variety that makes recently done actions unavailable.

my friend susan playtested the game for me after the jam ended, and she discovered a couple hilarious and kind of horrifying consequences of the way the game's memory and decision-making systems currently work. in her session, olive became unable to keep up with her needs very quickly and ended up trapped in her room with all her stats hovering around zero for the rest of the game. for a while, she cycled between the 3 actions that involve resting on the bed, getting off her bed at the end of each action before immediately hopping back on. then, she kinda just gave up on life and laid down on the floor.

characters are able to lay down on the floor when they have no available actions. i did not anticipate them staying on the floor for 3 real life hours.

we thought something broke and she was stuck forever, but later that night i got a text that she'd gotten up! so apparently if their stats are all at zero they might just decide to stay on the ground for hours. the next day, susan texted me saying "by the time i went to bed, the game had been running so long that olive had no memories left."

...i set an arbitrary limit for how many memories a character can have before the oldest ones are deleted. after hours of being unable to leave her room, olive eventually forgot about everything outside of it. i think i accidentally added dementia to the game. a death mechanic was on the list of ideas i didn't have enough time for, but it seems it crept its way in anyway.

alas, this is a one-man circus, so i had to be content with leaving the mechanics in this early prototypical state because a graphical game without graphics is a game that is unplayable. i gave myself two weeks to finish all the game's sprites, textures, sounds and UI. reading that back makes it sink in that that is not a lot of time at all...

this is where i should probably mention that i'm still recovering from a serious health scare that hit at the start of the year, and i'm very susceptible to adverse symptoms of overwork. yeah the ADHD meds help a lot but even now i'm quite sickly in my day-to-day life and i made some sacrifices that would probably horrify some people for the sake of finishing this thing on time. i think this jam's length was right in the manageable sweet spot for me: short enough to not drag on and kill my motivation, but long enough that i was safe from death by overwork. i still don't know if it was wise to go this hard for a deadline that would have affected nothing had i missed it, but since having the ability to finish things is still such a new feeling to me, i pushed for it. i'm writing this because i don't want to forget the sacrifice. i don't want to stay in the habit of undervaluing my health like that, especially when my art form of choice is one where people see failing to overwork as an embarrassment. i want to live long enough to avoid contributing to that. i should probably let this be the last game jam i participate in.

but the visuals! i've been pondering a lot about the place that the art side of game development has in my life. i've been drawing for much longer than i've been writing code, yet drawing has never felt "easy" to me like coding does, and it probably never will. drawing is stressful because i have to put in twice as much effort for the chance of making something i find acceptable, but no matter how difficult the struggle gets, i know i would never find any fulfillment in making anything unless it had my drawings in it. my complicated relationship with drawing must still be one of love, otherwise there'd be nothing stopping me from populating this game with hamsters generated by AI.

the inspiration for the look of this game was the music video for domino by ■37. i had this in mind from the moment i wrote the first lines of code, so its vibes have bled into every part of the atmosphere. i love how its cuteness is a thin veil for the more overpowering feeling of emptiness and foreboding, and yet they still exist in tandem. ■37 is one of my favourite vocaloid producers because of how well they achieve that dichotomy.

the girl from domino shown side by side with one of the characters from HAMSTER APPARATUS. they are both wearing round glasses and a formal black-and-white dress with a tie.

i based olive's design on the girl from that MV. this is the first time in years i've drawn a character who wears glasses. of course, the sheer amount of green in every corner of the game is also a trait it inherits from domino. i'm just now realising the shot of the girl staring into the camera from below is clearly where i got the idea for the game's cover, too.

my usual art style is already intensely chibi and yet i managed to reach new levels of chibification for the character sprites, which let me get away with not hand-animating a walk cycle for both characters. before i settled on the bounce animation, i thought of making them fly, with their arms spread out like they're being possessed. that would have made simple actions look insane. i don't think hamsters do that in real life.

i wasn't foolish enough to attempt to model everything myself with the little 3D modelling experience i have, so while i challenged myself to make a few models, most of the models are royalty free assets that i heavily edited or just completely redid the textures for. i learned how to use blender for UV editing and shading, which has made the whole business feel more approachable for future experiments. what did i make myself? the plate. the nameplates on the doors. the stools. and i was very proud.

i enjoyed editing stock photos to use as textures. most of the models in the game are still just default cubes squished together with some homemade textures. i think it enhances the low-fidelity look. old games didn't have the hardware budget to give background objects more than the bare minimum polygons, so textures did the heavy lifting instead. i was surprised with how few textures and models i needed before the environment started looking nice. even though the decor is strictly utilitarian, with nothing there that the characters don't interact with directly, it still has a style. that would not be as easy to accomplish in a 2D game.

my favourite texture i made was the jigsaw puzzle. i had no fucking clue what kind of image should be on the jigsaw puzzle, so i went on stock photo websites and searched whatever unusual keywords i could think of, and put them together in a collage. i didn't know what the result would look like, i just knew it had to be pleasant and strange. that was my strategy for finding background music, too.

a collage featuring monstera leaves, radio towers, and a giant eyeball

i really didn't expect the UI design to be as fun and easy as it was, seeing as i used to have a page on this website dedicated to complaining incessantly about godot's UI framework. i wouldn't have enjoyed it as much as i did if not for the offset transform properties they added in a very recent godot update.

godot's UI framework is very rigid: if you want UI nodes to appear in a horizontal list, they must be a child of a container with the dedicated horizontal list class. if you want a 2 pixel margin around that horizontal list, the horizontal list container must be a child of a margin container. god forbid you want a vertical list inside a horizontal list inside a vertical list with some margins sprinkled in (as i often do). everything has to be nested under 5 different containers for it to look right.

before the offset transform property was added, every UI element's position was set by the container it was a child of, and there was absolutely no way to move anything. you had to do a convoluted workaround where you duplicate everything and write code to position the duplicates manually if you wanted to animate anything in your menus. it was so fucking annoying. now, using godot 4.7, experimenting with adding little animations to each menu was so easy it felt liberating. it's never been so simple to make menus that are fun to use. the little cherry on top additions like the character's attributes list being skewed and the ability to smack around their profile photo with the mouse were both things i added on a whim just because it would be easy.

the check-in menu is my favourite part of the game, visually. the little donut models bouncing around in the background are in fact the original character models. they're probably like the tadpole form of this mutant hamster-girl species. giving them movement was the first time i had to interact closely with godot's 3D physics engine. the way they collide with each other like balloons reminds me of the bouncing DVD screensaver. it was hard to get work done because i was tempted to stare at it for hours.

i put the title screen and character creator together in like 2 hours at the tail-end of an all-nighter just before the submission deadline, by the way. again, i'm very impressed with how simple it is to play with some settings and make an interesting looking environment fast in godot's 3D engine. the parallax sky, the fog settings and lighting were all so easy to configure into something that looked nice and dreamlike. and also, i think you should know that the name of the title screen music is "a deep-sea fish thinks about the sky". i knew it would be the one from the moment i saw that name.

so that's the experiment done! this is a concept i've been passionate about for a long time and finally getting to put a prototype together with all the simulation mechanics i love has been a blast. now i have to decide what's next.

i'm intent on pushing this idea further, but i'm not going to abandon my main project, so HAMSTER APPARATUS v0.1 will be on the backburner until its time comes. this being the first project i've finished in years has taught me a lot about the way i like to work. for ages, i haven't trusted myself to make things that are good without a long, detailed plan, but i got through the entirety of HAMSTER APPARATUS's development with my only priorities being experimentation and keeping the momentum going. it seems i'm at a skill level now where i can trust myself to make something i'm proud of without figuratively putting myself on the chopping block.

i don't work on my art for the sake of making a product, the product is a less interesting side effect of the wider artistic process. working on this project was some of the most fulfilling work i've ever done, and now i'm daydreaming about what cool things i can do now that i have this experience. it was an absolute success.

as an experiment, i'm going to try open-sourcing the game's code. it's all available to look through on my github. study it, modify it, whatever. there's a lot going on below the surface that doesn't get communicated to the player in this build. go nuts, just don't send me a condescending email about how you think i should have programmed this or that mechanic, or i reserve the right to make fun of you publicly. thanks!