I’m currently trying to program a side-scrolling shooter using the same sprite engine (along with its sprite editor) that I developed for my game “Sternenschwarm”.
In fact, there are two programs running “side-by-side” here: the shooter itself and the background scrolling. Since it uses hardware scrolling, I can’t slow it down without it starting to stutter.
Notes:
TIP: Don’t forget to switch controls to keyboard
So far, I have only implemented 4 of the 20 levels.
I’ve disabled sound again for now. It’s still terrible.
Each level has its own unique design.
High scores are saved, as long as the disk is writable.
The game features “power-ups”
The floppy disk is already almost full, so I’ll probably have to rethink the overall game. I’m also having a really hard time striking the right balance between difficulty and fun this time around. I don’t really have any exciting ideas at the moment, other than more themed worlds.
Hi there. You are popping a new game every other day and there is a certain ‘AI aura’ starting to develop. So, to be frank, how much AI has been used here?
I had ChatGPT provide and explain the code for the hardware scrolling to me. I specifically tailored it to my needs (see the continuous redrawing of the graphics on the right side. That’s 2 new pixel columns per frame (= 1 byte, Mode 0). See also my published sprite editor for this, which supports the required in-memory data layout as an export format).
The note sequence for the background music comes 100% from AI (disabled in the mockup, though, because it’s still dreadful to listen to).
For the player, I use the brilliant ArkosTracker3. It already exists, so I don’t need to write it myself.
The other tools (sound effects, sprite engine, interrupt handler) were written entirely by me. Btw: the interrupt handler and stack management of all objects are written in C, taken from a microcontroller project. It’s always the same everywhere. That’s the beauty of it.
My sprite editor (HTML file, published elsewhere) was created with AI assistance. However, it’s used exclusively for creating graphics. Special feature: it can convert arbitrary images into CPC sprites. It supports a data layout that enables fast sprite drawing.
So how much AI is actually in the mockup? All in all, quite a lot. AI makes a great teacher and builds motivation. That’s how I use it.
Take a look at the little details in the current mockup, like: the plane’s flame moves depending on its own movement. That’s handcrafted. That’s the soul of a program. Does an AI include loving little details? I don’t know…
Also, the graphics get worse from level to level. I lost motivation. An AI would have finished it with a prompt?!
Two sentences about myself:
I’m 59 years old and have a degree in computer science (Diplom-Informatiker).
My motivation is keeping an old hobby alive. I’m now at an age where professional demands are winding down.
Back when CPC magazines were still sold at newsstands, I already had several publications in them about the CPC: games, copier and disk tools… Ever since then, my heart has remained tied to this computer. (Oh, the shock—that’s over 40 years ago!)
There were and still are publications of mine on other forums (pre-COVID; see Asteroids in an earlier version), but unfortunately, those forums have since vanished.
This leads me to the idea: Would anyone be interested in a series of articles here on how to develop software on a CPC using a cross-assembler? An AI can explain that too, though… Hmm… maybe not, then?! There are already YouTube courses as well…
It’s true:
Without AI, this mockup wouldn’t exist.
With more AI, it would look nicer and probably be finished.
There are exactly three or four topics left that interest me. All of my previous mockups and games head in this exact direction:
A game idea inspired by Elite, but with an integrated economic simulation where every decision the player makes has a direct impact on the overall system. It takes place entirely in our solar system (planets and moons).
A game idea modeled after The Bard’s Tale (Part 1). However, featuring a graphics engine that visualizes rotations (like in my first mockup). The puzzles should be so difficult that you need to do real-world research to beat the game.
A game that features continuous diagonal scrolling. Similar to the classic game Zaxxon.
(possibly also a more complex version of the Amiga game Carrier Command. An evolution of my Elite economic simulation. But for now, this one is just in the planning stages on a sheet of paper. I also haven’t decided on the target platform yet. Possibly just for the Raspberry Pi (natively), general Linux, browser game, CPC 6128, depending on what interests me more at that point.)
After that, I’ll turn my attention to other hobbies
See, it’s not that hard. Be brave. Now, with regards to the game.
Have you played it? I literally finished it just by pressing fire, without shifting the ship even one pixel from its starting position. Fun fact, if you have your browser in windowed mode and you start the game with the embedded emulator, you press fire repeatedly until you get all the pickups, then keep the fire button pressed and with your mouse click outside the wiki browser window. You can do whatever task you want on your PC while watching the game completing itself.
So, it does look to me as one case of (and this isnt aimed at this game in particular, but it’s a broader observation) ‘design of the game was left to the AI and the prompter doesn’t have a clue about what’s going on’; or even doesnt care.
The other thing that blows me away (again, broader observation) is the how many people will jump and cheer how great one release is, when it really is so abysmally crap.
(edit: just to clarify, both last sentences are meant to be about AI releases in general, not aimed at this release in particular)
And to be honest, his releases start to change my opinion about the usage of AI in retrogames. They are clearly inspired by classics - but also they are somehow a fresh inspiration with crisp and colourful gfx and not just a blatant 1:1 copy from an arcade original.
Sure, they lack that final polishing that changes the PoC into a fun and great game, but looking at the game, I see the potential for it already. I would LOVE to play that game after the level design was manually adjusted so it becomes an actual challenge.
But yeah, not with that level design as it is. But anyway … I love the scrolling, I love the gFX, I love the general feel of it - and I am surprised to say that about a game where AI played such a crucial role.
It appears that the non-disclosure of the use of AI is a sensitive matter, so I guess this should be taken into account. Then, people can only comment on what is being available to them. I’m not sure what is the point of posting a ‘prototype’ if one isnt aiming at getting feedback.
I published this prototype to get feedback. In whatever form. So far, I consider the discussion to be constructive. I read every piece of feedback. Thank you very much for your thoughts and input.
Well, I make shooters (the graphical part), so I know shooters. And I am very picky when it comes to shooters. What you’ll get from me is the same feedback I’d give my own self.
