GT1 - recompilation project

  • Thread starter Thread starter Ambivalent_
  • 14 comments
  • 1,520 views
Messages
59
Croatia
Croatia
Messages
Ambivalent_
There is a website that tracks recompilation projects for games on various platforms;


There are about a dozen PS1 games that are nearly finished recompiling;


It would be great if GT1 were recompiled like this because it would enable us to do any kind of upgrade and modification; new cars, new tracks, new tournaments, even graphical updates, both the visual boost with stuff like DLSS5 and path tracing and developing the visual damage system - all while preserving the original GT1 vibes and aesthetics.

If there is any interest from this community to try something like this, please let me know.

Supposedly, it's not so difficult tech-wise, AI can do most of the things for recompiling older games:

recompilee.webp
 
Last edited:
It would be great if GT1 were recompiled like this because it would enable us to do any kind of upgrade and modification; new cars, new tracks, new tournaments, even graphical updates, both the visual boost with stuff like DLSS5 and path tracing and developing the visual damage system - all while preserving the original GT1 vibes and aesthetics.

Recompilation does not enable any of that. Recompilation is essentially just emulating the existing code unchanged. A decompilation on the other hand recreates matching source code which can be modified to do absolutely anything. A recompilation can be done with existing tools and minimal effort, whereas a decompilation would be a labour of years for a dedicated team of reverse-engineers.
 
Recompilation does not enable any of that. Recompilation is essentially just emulating the existing code unchanged. A decompilation on the other hand recreates matching source code which can be modified to do absolutely anything. A recompilation can be done with existing tools and minimal effort, whereas a decompilation would be a labour of years for a dedicated team of reverse-engineers.
I mixed up decompilation and recompilation.

However, I think your view that "decompilation doesn't enable any of that" is too rigid.

Have you seen decompiled Zelda 64? The decompiled version has many great features such as new control options, dual analog controls, gyro aiming, autosaving... There are even mods changing gameplay and replacing the character models!
 
Last edited:
I mixed up decompilation and recompilation.

However, I think your view that "decompilation doesn't enable any of that" is too rigid.

Have you seen decompiled Zelda 64? The decompiled version has many great features such as new control options, dual analog controls, gyro aiming, autosaving... There are even mods changing gameplay and replacing the character models!
Looks like you mixed it up again there. The comment was that recompilation doesn't enable that. ;)
 
Recompilation does not enable any of that. Recompilation is essentially just emulating the existing code unchanged. A decompilation on the other hand recreates matching source code which can be modified to do absolutely anything. A recompilation can be done with existing tools and minimal effort, whereas a decompilation would be a labour of years for a dedicated team of reverse-engineers.
Is there currently a decompilation of GT1, let alone GT2, or any other GT game at present?
 
Is there currently a decompilation of GT1, let alone GT2, or any other GT game at present?

There was an attempt at GT2 a couple of years ago but it died off after a few months due to a lack of people with enough skills and time to contribute to it. There are no debug symbols for GT2 so it was essentially pure effort involved in figuring out what each instruction in the code did, one by one.
 
This is how we get de/recompiled GT1 🙏
 

Attachments

  • IMG_20260904_162556.webp
    IMG_20260904_162556.webp
    84.6 KB · Views: 5
This is how we get de/recompiled GT1 🙏

Don't believe the hype, it's a press release from a company trying to sell the product. LLM marketing people have a long history of false claims. As Rivit above points out, if it's all left to a machine to guess at then nobody will have an understanding of how to make it work when it inevitably gets it wrong, as these systems are not intelligent and do not work consistently. If they were genuinely as brilliant as claimed, the tech companies wouldn't have to be working so hard to advertise them and subsidise their use.
 
Last edited:
Don't believe the hype, it's a press release from a company trying to sell the product. LLM marketing people have a long history of false claims. As Rivit above points out, if it's all left to a machine to guess at then nobody will have an understanding of how to make it work when it inevitably gets it wrong, as these systems are not intelligent and do not work consistently. If they were genuinely as brilliant as claimed, the tech companies wouldn't have to be working so hard to advertise them and subsidise their use.
There's nothing for me to "believe", I see first hand the progress in AI capabilities (no matter what "team hype" or "team skeptic" tells me).

If the next 4 years bring as much AI progress as the last 4 years, there will be no human programmer who could match AI.

That's why I think that community development is the future, all of us could contribute and develop amazing things for our favorite IPs.
 
Don't believe the hype, it's a press release from a company trying to sell the product. LLM marketing people have a long history of false claims. As Rivit above points out, if it's all left to a machine to guess at then nobody will have an understanding of how to make it work when it inevitably gets it wrong, as these systems are not intelligent and do not work consistently. If they were genuinely as brilliant as claimed, the tech companies wouldn't have to be working so hard to advertise them and subsidise their use.
It's obvious you are just making up excuses in recent years because you are too lazy :lol:

There's nothing for me to "believe", I see first hand the progress in AI capabilities (no matter what "team hype" or "team skeptic" tells me).

If the next 4 years bring as much AI progress as the last 4 years, there will be no human programmer who could match AI.

That's why I think that community development is the future, all of us could contribute and develop amazing things for our favorite IPs.
Agreed! GT1 and GT2 and all the other old stuff will be great again :D:gtpflag:
 
I hope if it happens that recompiling would allow to add content without having to replace other content.
 
It's obvious you are just making up excuses in recent years because you are too lazy :lol:
If you believe your "great warrior" is too lazy, I'd like to see you get Codex, Claude Code or whatever to vomit a fully functional PC port of GT1 yourself. Preferably before the end of 2026. Let's see if the task is as easy as that tweet in OP's message implies.
 
There's no reason for arguing and rude comments. Guys like @pez2k did a fantastic job and with AI tools he can be more productive. Until AI is fully autonomus for this type of work (maybe in the next 1-2 years), it's a great tool for guys like Pez.
 
Last edited:
There's no reason for arguing and rude comments. Guys like @pez2k did a fantastic job and with AI tools he can be more productive. Until AI is fully autonomus for this type of work (maybe in the next 1-2 years), it's a great tool for guys like Pez.

No thanks, it's unreliable and untrustworthy, as well as a massive waste of resources and damaging to the environment. I'd rather just write code myself.
 
Back