How Medium Rare Uses Pixel Art & Collages to Evoke Dread of Industrialized Cannibalism

Medium Rare's Aesthetics

I'm still not entirely sure what other people see in the aesthetic. A lot of older CD-ROM games did something similar. We're essentially taking 3D assets and 2D photographs, crushing them down to a chunky resolution via downscaling in Photoshop or Aseprite, and indexing everything into specific color palettes.

What makes ours a little different is the collage approach. I think some people call it photomoshing. Our artist, Giovanni Fim, is a Photoshop wizard. He takes these crazy ideas I have, finds stock photos, and mashes them together until we have an image or a character that looks like a cohesive unit. Then we downscale and index it to whatever palette is used for that area in game (each area, forest, village, factory has a specific palette). After that, he'll go back in, paint new highlights or details on top of the image, then reindex the image, and repeat until it looks right.

We actually arrived at this style because of a limitation. Originally, I wanted a fully hand-drawn dungeon crawler, but Giovanni didn't want to animate all those sketches. We needed something that could compete with all of the interesting visuals on Steam without requiring that amount of animation.

So after exploring indexing and downscaling and realizing that Fim was a Photoshop legend and could pull stock images into really interesting compositions that he could then paint over and color index, we realized we had a very compelling style. It looked strange, but in a way that we felt would make people curious. We wanted people to see the game and immediately wonder what they were looking at. We wanted to get interviews like this!

Dark Matters in Stylization

I like picking shocking premises because I enjoy exploring darker themes. I'm an English major, and I think we're living through a particularly anxious time. Exploring horror is one way I deal with that anxiety. But I like making non-traditional horror games – games that make you feel dread instead of making you jump.

Pixel art gives us a way to explore these really visceral ideas without making everything feel completely awful. There's enough suspension of disbelief that you can do terrible things, but because it looks like a video game, it still feels fun instead of messed up.

My first game, Execute, is about using a cursed Windows 98 computer to kill everyone on Earth, but because everything looks like pixel art, it desensitizes you. We're doing something similar with Medium Rare. The art makes these horrible actions feel entertaining. You're having fun doing things that should disgust you. And ideally, by the time you reach the end of the game, you start feeling a little disgusted with yourself. That’s a goal of our studio: to evoke dread instead of horror. 

Industrialized Cannibalism in Visuals

I think the aesthetic does a lot of the work here. Industrialization naturally makes things feel homogenized and one-note, but I wanted the world to feel like it was psychologically falling apart. It's like this psychedelic dread that permeates the world through our protagonist's eyes. His name is Leo Night, but we call him Leo N. for short. Nobody's actually on psychedelic drugs, but the fear and anxiety are dripping off every encounter. The world is falling apart in front of you, and your character is psychologically falling apart along with it.

That's why we picked such normal environments. You move through a forest, a village, and a factory. Your character uses a Glock pistol rather than magical weapons. I was really inspired by EarthBound and how it places this strange, intense adventure inside something that resembles ordinary reality.

The environments themselves aren't necessarily that unusual. What's unusual is experiencing them through the eyes of someone who's psychologically damaged.

I think that's how reality feels to a lot of people right now. We're living in an anxious time, and sometimes the world feels like it should look like Medium Rare. The game is essentially trying to visualize that psychological damage. Through this lens, nothing can be visually boring, as the ordinary will be turned on its head in each area.

Art Pipeline

We actually have four different asset pipelines.

First, the player animations are Creative Commons 3D assets. I modify those, swap weapons between animations, and use PixelOver to downscale everything, index the colors, and adjust the lighting. That keeps all our 3D assets visually consistent.

Second, our backgrounds are actually full-motion stock video. We spent a long time finding a footage library that doesn't accept AI-generated videos because we wanted to avoid using generative AI in the game.

I take footage of someone walking through an environment, use FFmpeg to extract frames every 5th frame (sometimes more or less depending on the clip), then make those frames into a sprite sheet in Aseprite (I downscale them first using a bilinear filter to keep them greasy). After that, we index them into the color palette. Each of our flows ends with this downscale then color indexing. We call it crushing, but I think people normally call color indexing posterization. The backgrounds also have various shaders applied to them. I have one that just plays semi-transparent fog-looking noise in random patterns across them. Another duplicates the image 4-6 times and uses different wave shapes to warp the image in a psychedelic way. The shaders you see while walking around are subtle. We also have a DOF shader running to help make the harsh pixels look softer and more realistic. 

Third, we have the enemies and event illustrations. There is a video about this on Giovanni Fim’s YouTube channel. He builds these in Photoshop using stock photographs that he surgically stitches together. In the shared example, we wanted to make a man hanging from a snare for an event. Fim found about a dozen images of various woods locations and people that he cuts and merges together to create the scene. He then paints over the top of that image to make it look more realistic. He adds blood, dirt, different highlights or shadows to fix lighting problems from merging images, etc. After this, he will crush and check to see if it looks good. If not, he will go back to the original and crush again until it looks right.

For enemies specifically, the process is the same, but he takes those finished images into Aseprite and creates the idle and attack animations and touches up any finer details so the enemies look more video gamey.

Finally, we have the items. Those are generally stock PNGs crush. Sometimes we clean them up in Aseprite, and sometimes we don't. If we need an item in the game, occasionally we'll just find something that works, crush it, and throw it in.

It's a pretty unusual pipeline, but it's allowed us to create a tremendous amount of art very quickly that looks very high-level and isn't slop!

Tools & Workflows

We're building Medium Rare in Godot 4.6. I've used Godot for my previous game and probably half a dozen unreleased projects, so I'm very comfortable working in it.

The game is mostly UI. Almost everything is either a button or a flat image somewhere on the screen, so the programming has actually been relatively straightforward. Combat has been the complicated part.

Beyond Godot, we use Photoshop, Aseprite, PixelOver, and FFmpeg.

We haven't really needed to build custom tools. That's one of the things I love about this art style. Almost anybody could replicate the basic process if they can find interesting images.

I'm just extremely lucky to have an artist with a great eye for color and highlights. Giovanni is a sicko who will sit there for 30 minutes adjusting highlights, reindexing the image, and doing it again until it looks interesting.

There's nothing particularly complicated about the technology. It's really about having a good eye.

Readability

Readability has actually been one of our bigger challenges. Originally, the entire game was built around a grid in the center of the screen. Eventually, we shrunk that into the little GPS in the bottom left corner so we could show more of the world. Players initially didn't like that because everything became harder to read.

I think making incremental games before this was fortunate. Those games taught me how much information you can put on the screen before overwhelming somebody.

My co-dev, Peter, and I are also both teachers. We understand that information overload ruins the onboarding experience, and in games, onboarding a player is everything (besides the game being actually fun once they understand it).

Because the art is so loud, we've tried to make everything else extremely minimalist.

We use four main colors to communicate information. Yellow means something needs your attention. Green means something positive or something you'll gain. Red means something negative or something you'll lose. And blue indicates an enemy is exposed or vulnerable.

We've also stripped away a lot of unnecessary UI. Originally, enemies had health bars, percentages, and other information surrounding them. Now their names and outlines change color as they take damage, similar to Dragon Quest.

The combat grid also helps tremendously. Peter originally suggested adding range to our combat system, and I immediately thought of Mega Man Battle Network. A small 3×3 grid allows us to create interesting movement patterns and positional mechanics without making the game visually overwhelming.

So even though the game is densely packed with rules and information, we've tried to strip the presentation down as much as possible. The louder the art gets, the quieter the UI needs to be.

Inspirations

Resident Evil is obviously a huge influence. I've always loved how those games deal with absolutely horrible situations but still have this sick sense of humor. We leaned a little further into dread with Medium Rare; to me, it feels a bit closer to the sense of dread in Dead Space.

Another major influence was Tender Is the Flesh, a novel about industrialized cannibalism. I read it shortly before we started development and immediately thought we needed to do something with that premise.

And then there's EarthBound. I've always loved its strange, psychedelic atmosphere and the way it takes itself seriously by being so finely crafted, yet maintains a very video gamey and playful experience.

Medium Rare is basically a giant blender of all those influences. It's horror, satire, dark comedy, and a bunch of other things thrown together. 

I'm also lucky to have a great team that helps make sense of my ideas. I'm a chaotic mess of ideas. I'll throw out something completely ridiculous, and Giovanni will figure out how to make it visually interesting. Then Peter helps refine the mechanics and the tone so that the ideas actually communicate satire, discomfort, or horror instead of just becoming random noise. I think that collaboration is really what makes the tone work.

Animation

Animation has been tough because all three of us have full-time jobs outside of game development. We actually chose this art style because it was one of the fastest ways we could make something visually interesting.

Originally, we weren't going to animate anything beyond the 3D assets we found. But we couldn't find enough 3D enemies with a cohesive style, so Giovanni started making them in Photoshop. At first, he didn't want to animate those either, but once we realized we could downscale them into pixel art and animate them in Aseprite, he changed his mind.

We've been very selective about what we animate. Playing JRPGs taught me that idle animations often do the most work establishing a character's personality, so that's where we spend a lot of our time.

I've also noticed that games like Dragon Quest, EarthBound, and Pokémon get a tremendous amount of mileage out of simple tween or shader effects. So we use tweens and shaders wherever possible. When enemies die, they squash, turn red, dissolve into static, and shoot out these little blood particles.

I handle sound design and tech art, and combining that with Giovanni's disgusting death sprites creates a lot of impact without requiring bespoke death animations for every enemy.

Honestly, some of the most impactful moments in the game aren't animations at all. They're single frames showing an enemy's corpse. You shoot something, it explodes into this horrible pile of gore, and even though you're seeing the same sprite every time, it doesn't really get old. We don't have the time to animate everything perfectly, so we've found that consistency is the next best thing.

Balance & Boundaries

As I mentioned before, pixel art allows us to get away with quite a bit. There isn't really a hard line, although sexual violence is definitely something we won't cross.

Even though you're eating people in this game, we almost never actually show it happening. There's one instance where you see the character eating part of a foot, but it's more a gag than an actual visceral cannibalistic scene. You never see the main character cutting somebody up or eating someone alive. We've had people request things like that, but I think some of it is better left to the theater of the mind. 

Originally, we had actual body parts as resources, but eventually we abstracted everything into cuts of meat. When you kill somebody, you get a steak. It's literally a stock photograph of a steak that we've crushed until it looked crispy.

Human meat probably wouldn't look like steak, but that abstraction helps you avoid feeling completely disgusted by what you're doing. So we're constantly trying to show as much as possible without showing too much. There are definitely some enemies in the game that might push that boundary. You'll have to see. But generally, I think suggesting something horrible can be more effective than actually showing it.

Challenges

The hardest part has definitely been readability, and honestly, I think the game still struggles with it sometimes. Everything is so psychedelic, strange, and crushed that it can become a little nauseating. We're constantly fighting this battle of how much information belongs on screen, how intense the visuals should be, and how much we need to strip away to keep things understandable.

A good example is movement between areas. Originally, we experimented with turning animations using tweens to simulate the player looking around. But we found that those animations were actually more disorienting than simply cutting between areas with a slight warping transition. So we removed them. 

That's been the challenge throughout development. We're constantly asking ourselves how much we want to overstimulate the player versus how much we want to keep them engaged. We want the world to feel uncomfortable and overwhelming, but we don't want actually playing the game to feel that way.

Payne Robinson, Game Developer

Interview conducted by David Jagneaux