[EN] Project Hura Gun: Your Questions Answered

Over the past months I’ve published three videos about my Commodore 64 Ultimate project. The first was a proof of concept of the engine, the second showed the engine and its editor in more detail, and the third was the first trailer of the game itself, Project Hura Gun. The response has been far bigger than I expected, so it’s time to sit down and answer the questions you’ve been asking.

Listen to this article (about 18 minutes). The audio version was generated automatically with AI text-to-speech.

First, a few numbers and a big thank you

Across the three videos you left 484 comments: 345 top-level comments and 139 replies in discussion threads. Roughly three quarters of the top-level comments were positive, about one in ten were questions or suggestions, and around one in ten were critical.

Thank you all, and I really mean all of you, including those who didn’t like what they saw. The critical comments were some of the most useful ones. They made me think more broadly about the project, and they show what C64 players and C64 Ultimate owners actually care about. The „this isn’t a real C64” debate, the requests for scrolling, the remarks about the player character’s animation: all of it helps me understand the audience, and some of it has already changed how I think about the game.

1. Commodore 64 Ultimate Game Engine – Proof of Concept
2. Commodore 64 Ultimate Game Engine – Preview 2
3. Project Hura Gun – first trailer

The questions below are grouped by topic rather than quoted one by one. Many of them came up again and again under all three videos.

Contents


Hardware

No. The game targets the Commodore 64 Ultimate and Gideon’s Ultimate 64 (Elite) boards. It isn’t meant for a stock C64.

This wasn’t a decision made lightly. The core idea of the whole project is fully animated backgrounds built from bitmap graphics, and that simply isn’t possible on an unexpanded C64. The engine relies on what the Ultimate hardware adds to the familiar machine: the faster CPU mode, a large REU memory expansion, the Ultimate Audio sample channels and the programmable palette.

For the same reason it won’t run on a classic C64 with an Ultimate-II+ cartridge either. The cartridge provides the REU and audio, but the game needs the fast CPU.

I fully understand the people who would prefer a game that runs on original hardware, and there are many great developers doing exactly that. This project explores a different question: what a C64-style game can look like when it uses what the Ultimate machines offer, while keeping the look and feel of the C64.

The extra speed matters, but it’s only part of the story. The display is still the standard VIC-II with its standard graphics modes, the palette is still 16 colors at a time, and the sprites are still 24×21 multicolor C64 sprites. What changes is how much data can be moved per frame and where it comes from. In this engine, almost everything is streamed from REU memory.

So where do the 64 MHz go? Every frame the engine:

  • copies a full-screen bitmap from the REU,
  • draws the tileset on top of it, which is a big deal: it’s something a stock C64 simply can’t do within a single frame,
  • runs the sprite multiplexer (currently 16 sprites),
  • runs the logic of all the actors.

There’s no double buffering. All of this fits in a single frame with plenty of CPU time to spare, even at 48 MHz.

In theory, yes, but it’s not the fastest approach. The engine loads its data from the Ultimate filesystem into the REU first, and then streams graphics and audio from the REU during play.

No. The game relies on features specific to the Ultimate hardware, and it doesn’t run in an emulator. To play it you need a real C64 Ultimate or Ultimate 64 board.


Graphics

What you’ve seen so far is neither. The graphics mode is the standard C64 multicolor bitmap. Every background frame is a full 320×200 multicolor bitmap, which is 10,000 bytes once you count bitmap, screen RAM and color RAM. The engine copies these frames from the REU into C64 memory.

This is where the project started. The very first proof of concept was nothing more than copying bitmaps from the REU, plus a simple sprite multiplexer. My current library can copy any number of animation frames from REU into C64 memory, so a background can run at 50 frames per second, limited only by the amount of REU memory.

Color cycling is also already supported by the engine’s API. I just haven’t used it in any level yet, so you haven’t seen it in the videos. Combined with the per-room palette described below, it opens up some nice possibilities without spending REU memory on extra frames.

One commenter pointed out that animating every background all the time would be too much in a real game, and I agree. In the videos the animation is everywhere because I’m showing what the engine can do. In the game, full animation will be used where it matters: boss fights, danger and special moments.

This was the most common question under the trailer, and a fair one. Right now Project Hura Gun is a flip-screen game, mainly because I wanted the backgrounds to be fully animated, and full-screen animated bitmaps and scrolling don’t go well together.

I haven’t closed the door, though. I’m thinking about horizontal and vertical scrolling and looking at possible ways to implement it. For now, the game is screen-based.

Some of you noticed a „scrolling” level in the first video. That wasn’t scrolling at all: it was just an animation. I made it as an experiment, but it shows how many interesting background effects can be achieved with animation alone.

Several of you spotted extra shades of red and purple. On screen there are still 16 colors at a time, but the latest Ultimate firmware lets you change the RGB components of each of the 16 palette entries in real time.

The engine uses this every time you enter a new room or screen, so each location can have its own palette. The base is close to the classic C64 colors, and some entries are adjusted per screen. All of this is set up and edited in Godot, together with the rest of the map.

Because the palette entries are only re-tinted, the graphics keep the C64 character: same modes, same wide multicolor pixels. They’re just colored differently from room to room.

They’re palette effects, not glitches. Because the palette can be changed in real time, a flash is simply a quick change of the colors on screen. There will be more effects like that in the game.

As long as the backgrounds are multicolor, the sprites will stay multicolor too. It keeps the visual style consistent, and it saves sprites: hires detail on top of multicolor graphics usually means stacking extra sprites, so we would be spending twice as many of them.

For me it’s either hires or multicolor, not a mix of the two. That’s a design decision, and as the game’s designer it’s one I’ve made on purpose.

Yes. That’s the sprite multiplexer, and it’s on my list of things to improve.


The engine and the editor

The runtime that runs on the C64 Ultimate is written in C, compiled with Oscar64, a C compiler for the Commodore 64. Using Oscar64 was one of the core assumptions from day one. It produces fast 6502 code, and C lets me iterate much more quickly than writing everything in assembly.

Yes, well spotted. Early on I realized I would need a proper tool with a graphical interface, a GUI editor, to build levels. I didn’t want to write one from scratch. I know Godot reasonably well, so I decided to write scripts that convert Godot assets and scenes into Commodore formats instead.

That’s how the engine began. Today I can place elements in Godot, build maps, add objects and describe their behavior. A set of export scripts then converts everything into C64 data: bitmaps, sprites, tilesets, palettes, music and map data packed into REU images, plus generated C sources for the runtime.

So Godot is the editor and the exporter. The game itself does not run in Godot. It runs as a native program on the C64 Ultimate.

Not entirely. A lot can be done visually: maps, rooms, enemies and their movement patterns, blocks, triggers and music assignment are all set up in the editor. Game-specific logic, such as what a particular switch does or how a boss behaves, is written as small C handlers that you attach by name in Godot. It’s closer to „editor plus scripting” than to a no-code tool.

The editor side is Godot, so it works on macOS (that’s where I develop), Windows and Linux. The final game runs on Ultimate hardware.

I’d say the engine is about 80% done. What’s ready already lets me build screens pretty much any way I want. Most of my time now goes into the graphics, which are drawn by hand and take far longer to get right than the code.

Not for now. Turning the engine into a separate, standalone product would be a project in its own right, and right now the priority is the game. Project Hura Gun has to come first.

A lot of things in the engine are still hard-coded around this one game. Before anyone else could build on it, I would need to think carefully about how to deliver it as a reliable engine for third-party developers, with stable tools, documentation and support. That’s not something I want to do halfway.


Sound and music

The soundtrack of Project Hura Gun is composed by Magnar Harestad, a musician with deep roots in the demoscene. He comes from the scene, and that background shapes both his sound and the way he approaches music for this kind of hardware. You can find more of his work here:

No, and many of you guessed correctly. The music is PCM audio played through the Ultimate Audio sample channels, streamed from the REU. Sound effects use separate channels, so music and effects play side by side.

In the first two videos, where I was only showing the engine, the music was AI-generated. I needed something to play so I could test the audio part of the engine, and I said back then that I would work with real musicians for the game.

That has now happened, and it’s the news I’m happiest to share: the excellent musician Magnar Harestad has agreed to create the soundtrack for Project Hura Gun. What you heard in the trailer is his work. So the game’s music is made by a real musician, not by AI.

Yes, I can make those tracks available for download. Just keep in mind that they are AI-generated test music, not the game’s soundtrack.


Technical questions and details

Yes. There’s no floating point anywhere in the runtime. Actor movement uses lookup tables (for example sine waves and jump curves) together with fixed-point accumulators, so smooth sub-pixel motion costs almost nothing at runtime.

The first video showed an early state of the engine. Sound effects are in now. They play on their own Ultimate Audio channels, separate from the music, and you could hear them in the later videos.

That was the recording, not the engine. While capturing that video I was streaming the full picture and sound from the C64 to my PC at the same time, which caused the occasional stutter.

The game is aimed at the C64 Ultimate and Gideon’s Ultimate 64 boards, including the Ultimate 64 Elite II. The original Ultimate 64 Elite tops out at 48 MHz instead of 64 MHz, and the game will run on it too.

Yes. I’ve seen the technique on a few YouTube channels and in CSDb demos, and I tried it myself, but for a game it was rather cumbersome, both to use in code and to create assets for.

The bigger issue is memory and time. With this technique the image isn’t stored in video memory: the CPU has to „draw” the entire background again every single frame, one color write at a time. That takes a lot of memory for the drawing code and data, and it might not keep up every frame once the game logic, sprites and sound also need their share. The standard multicolor bitmap turned out to be far more practical.

Not with this engine. The editor and the export pipeline run on a regular computer (I work on a Mac). The C64 Ultimate is where the finished game runs.


The game: Project Hura Gun

I’ve read every comment about the main character: the walk cycle, the legs, the stance, the jumping and falling. You’re right, and these are things I’ll keep working on. The character in the trailer isn’t final.

With a joystick or the keyboard. If you’d rather hear just the game, without the music, press M to pause and resume the music. N switches the sound effects on and off.

The goal is to have Project Hura Gun finished and playable in 2027. That’s where my focus is.

Yes, I’m planning Project Hura Gun as a commercial product. Beyond that I haven’t given any thought yet to pricing, stores or distribution. When I have something concrete, I’ll share it.

People in the comments estimated around 30,000 C64 Ultimate machines, plus the Ultimate 64 boards Gideon has been selling for years. I’d be happy if 30,000 people saw the game. To be honest, I’ll be happy if even a handful of people like it.

I’m making the game I always wanted to have on the C64. Now I can have it on the C64 Ultimate, with new colors, great music and a lot of action on screen. That’s reason enough for me.

It’s too early to talk about a physical edition. I want to concentrate on getting the game done first. If a physical version ever happens, it won’t be a cartridge. It would be a device that connects over USB, which suits the way the game loads its data on Ultimate hardware.

I thought for a long time about what to make first. One option was something in the spirit of Turrican or a metroidvania. The other was a game connected to Miecze Valdgira (The Swords of Valdgir), a C64 game I co-created back in 1995. A sequel, or a game that refers back to the Swords, is still on my mind.

But first: Project Hura Gun.


The use of AI

AI came up in the comments more than once, so here’s an honest picture of where it is used and where it isn’t.

Not in the sense of „type a prompt, get a background”. Every background starts in Photoshop, and more recently in the new free tool PhotoCraft. At this stage I work conceptually: I compose each background from many different elements (compositing). Some of those elements are occasionally generated with AI, but never a whole background.

For generating fragments of graphics I use InvokeAI, running locally on my own PC with an NVIDIA RTX 4070 Ti under Omarchy Linux. It’s used for individual pieces that go into a composition, not for complete images.

Once the general „sketch” of a background is ready in PhotoCraft, I move it into Aseprite and work on it by hand until it fits the native C64 multicolor mode. This is the most time-consuming part of the whole project, and it’s done pixel by pixel.

To help with that I use my own Aseprite plugin, partly written with the help of Claude Code. It’s a simple tool that checks in real time how many colors are used in each 8×8 or 4×8 character cell, so I can see immediately where the graphics break the C64 limits.

They’re also made in Aseprite.

Partly, but the design is mine. About a year ago, in 2025, I wrote the basic Godot code, the scripts that export assets to C64 formats, myself. Since then I’ve continued developing them with Codex.

The whole game runs on code written in C for the Oscar64 compiler. The architecture of the engine is my own, and I built the proof of concept from the first video on my own. Since the beginning of 2026 new code has been written with the help of Codex and Claude Code, while the architecture has stayed essentially the same.

As mentioned above: the AI-generated music was only used for testing in the first two videos. The game’s soundtrack is composed by Magnar Harestad.

This article was written in collaboration with Codex.


Questions I’ll leave open for now

Some of your questions concern things I haven’t decided yet, so I’d rather not guess:

  • Will there be tutorials on how it all works?
  • Will the game also use the SID chips (or extra SIDs) in addition to PCM audio?
  • Will the game’s soundtrack be released separately?
  • Publishers, and whether there will be a beta test.
  • When will you show more gameplay?
  • Diagonal shooting and other gameplay changes.

When I have answers, you’ll hear them first, here and on the channel.


Thank you again for watching, commenting, arguing and cheering. It’s great to see so many people excited about what the C64 Ultimate can do, and just as great to hear from those who care about the original machine’s spirit. Both keep this project honest.

Stay tuned.

Zobacz także