mid_gen's Launch a Game Studio Thread

refreshing my shader programming (AI a godsend here for stuff outside my wheelhouse)

AI is definitely a godsend for getting basic shader programming off the ground to build on, otherwise such a tedious process and you have to go over and hand tweak anyhow whether you start with AI or not to optimise for what you want.
 
AI is definitely a godsend for getting basic shader programming off the ground to build on, otherwise such a tedious process and you have to go over and hand tweak anyhow whether you start with AI or not to optimise for what you want.
I went a step further yesterday, I wrote my own MCP server, with a few tools, one grabs a screenshot of the editor so Claude can see what's on the screen, and then I added a python remote execution tool, so Claude can interact directly with the Unreal Editor Python API.

Now I can just ask it to create and edit blueprints for me, including the material editor. It's pretty good at it too....I have it tweaking my shield material right now.
 
One major issue today, turns out I built my workstation and server when Notepad++ was compromised....and I have the potentially vulnerable versions installed. So I currently have Claude reading the threat repots and scouring my machines. Seems this malware was quite targeting and they weren't after domestic users (looks like it was state-sponsored and after state targets).

My server apparently is clear.....manually removed all traces of NP++ from both machines.
 
yep im sure a lot of people are worried about the NP++ we were talking about it Monday internally.
I had Claude read the threat report and check for any traces of the indicators, and it didn’t find anything. The had it manually scour any trace of NP++ (don’t trust the installer).. think I will periodically check for a while, but think I got away with it.

I didn’t have auto updates on, which think saved my bacon.
 
Last edited:

Nearly forgot this week's vlog! Been ridiculously busy....I was meant to be dropping to 4 days a week for this contract from beginning of Feb but predictably it got ridiculously busy and I've just finished 5 days....and am now sat here on the sofa on a Friday night working on the game.....ah well.

I said I'd be doing just 3 days for them next week but I can imagine how that'll pan out.....baaaaah but it's hard to turn down the money just now. Need to get my house move done and then can stick it to the man and go full-time indie.
 
Will be interesting to see how much AI impacts AAA style game development - obvs GTA6 has been delayed so it's not exactly a magic bullet yet but there's been some incredible toy examples - like this guy knocking up a "flight simulator" very rapidly:


edit - looks like he had to iterate again and make it a bit more efficient:


Of course he's made a previous iteration before that too and so I'd not expect someone starting from scratch to be able to build the same in just 1 hour but it's still very impressive even if just a toy/demo rather than a fully fledged game.

I presume that for professional games it's currently still just used as a coding assistant and maybe for generating some artwork, tiles etc. (modding seems to have been sped up dramatically by it too - so it does seem like now is the time when a one person studio + the odd contractor can feasible create things that would've previously taken an entire team.)
 
Last edited:
I presume that for professional games it's currently still just used as a coding assistant and maybe for generating some artwork, tiles etc. (modding seems to have been sped up dramatically by it too - so it does seem like now is the time when a one person studio + the odd contractor can feasible create things that would've previously taken an entire team.)
Probably, but I very much like @mid_gen's approach of trying to use local creatives for artwork and stuff rather than using AI.
 
Probably, but I very much like @mid_gen's approach of trying to use local creatives for artwork and stuff rather than using AI.

Presumably there's a middle-ground - creatives for core characters, general style and vibe etc.. and then AI for some of the less prominent stuff, some of the grunt work etc.

Like, I dunno, say the interior artwork used in the buildings in some random village an adventurer only visits infrequently in some adventure game etc.
 
Last edited:
A flight sim like that, you don’t need to create any content, it’s literally just a very, very simple engine, and sucking the map content from some API, and a plane mesh to move around, so it’s not more impressive than any toy demo….other than thanks to the cloud and an API it can let you fly anywhere.

Content is where the differentiation is….even amongst 100% human teams the difference between genuinely talented artists and bang average ones is what makes the difference to successful titles. You can’t stand out in a crowded market with mediocre artwork.

AI can do the technical part really well, building the systems and engines that make up the game. I’ve built out the whole macro game loop in my prototype today with Opus 4.6 while i was doing other work. Will be human artists doing the content.
 
Oh for sure, it's only presented here as a toy/demo - the fly anywhere on a "real" map aspect is quite novel and fun though, he seems to have gotten thousands of signups and hundreds online simultaneously.

The other guy (Levels) did one like a year or so ago and made a low six figure sum within a few days from a mixture of paid users and sponsors of in-game ads. Again it was more of a toy and had frankly very crude graphics but as a demonstration of vibe coding a game it was quite interesting. I think building an audience is pretty crucial too, which you're doing here with youtube - Levels has a big audience of generally fairly well off tech people so even with his naff graphics/artwork he had many people willing to hand over money to play something fun and new.

There's certainly a lot to be said for the build in public approach, especially when it comes to either consumer apps or indy games. I guess also (depending on the genre) then building a community, discord or whatever could certainly help with anything that might require ongoing subscriptions too.
 
I had a quick play and it isn't very impressive, albeit it is just a demonstration, outside of the main areas - most parts of the world are just flat textured fairly low resolution basic terrain with a lot of slow loading in.
 
Oh for sure, it's only presented here as a toy/demo - the fly anywhere on a "real" map aspect is quite novel and fun though, he seems to have gotten thousands of signups and hundreds online simultaneously.

The other guy (Levels) did one like a year or so ago and made a low six figure sum within a few days from a mixture of paid users and sponsors of in-game ads. Again it was more of a toy and had frankly very crude graphics but as a demonstration of vibe coding a game it was quite interesting. I think building an audience is pretty crucial too, which you're doing here with youtube - Levels has a big audience of generally fairly well off tech people so even with his naff graphics/artwork he had many people willing to hand over money to play something fun and new.

There's certainly a lot to be said for the build in public approach, especially when it comes to either consumer apps or indy games. I guess also (depending on the genre) then building a community, discord or whatever could certainly help with anything that might require ongoing subscriptions too.
Vibe coding where you’re primarily just connecting existing services is very powerful. I use it all the time to just try out different tech stacks for this fintech….you can just create a new repo, or Claude code, “ hey Claude, create a new three.js project that will render terrain data from <insert mapping api> and allow the user to fly a plane around the landscape, streaming in the terrain and converting on the fly.”

Claude or Codex could knock that out in a few minutes. This guy has implemented a backend and stuck it on AWS most likely, again, super easy, Claude knows how to talk to the AWS CLI and can deploy it all for you to, servers on Farage, frontend on amplify….job done.

It’s very powerful. But it’s not the difficult part of making a decent game.
 
To quote one of my MBA professors, Chris Dalton:

The world is changing. Information is cheap, AI can generate plausible text in seconds, so memorised knowledge is no longer the scarce resource. The scarce resource now is sense-making, which is critical thinking on what to trust, what matters, what to do next, and how to do it with others.
 
To quote one of my MBA professors, Chris Dalton:
To be more specific, with games…..it’s easy to make tech, but it’s difficult to make players *feel* something, and AI is never going to be able to do that.

It’s where I’m at with my prototype, I’ve built a lot of functionality this week, framework of the main map loop, into the jump loop. But at present it’s still just a bunch of visuals that kinda do a thing. Actually making it into an engaging gameplay experience is the bit AI can’t help me with.
 
To be more specific, with games…..it’s easy to make tech, but it’s difficult to make players *feel* something, and AI is never going to be able to do that.

It’s where I’m at with my prototype, I’ve built a lot of functionality this week, framework of the main map loop, into the jump loop. But at present it’s still just a bunch of visuals that kinda do a thing. Actually making it into an engaging gameplay experience is the bit AI can’t help me with.

100% on board with you. I maybe didn't explain myself fully - the AI can do the "work", but without creativity, without ideas, without the human component, it can only take you so far. For doing legwork, it's great. For art, it's not.
 
Don't think I'll manage a vlog this week. I've done quite a lot of work on the prototype, but I've ended up 5 days a week still for the fintech client. I've just had a chat with em though and they want to ramp down for a bit from next month once I've got this current chunk of work delivered so I will definitely be getting a lot more time to focus on the game studio!

Also, got mortgage appointment tomorrow, going to be in the throes of a house move so the time will be nice to get that done and dusted ASAP.
 
Back
Top Bottom