GTA 6 Speculation & Discussion [Part 7] - The one with the announcement

[safde]

4 hours ago, Zapper said:

How much longer, though? It has already been about 4 years since RDR2 came out. Even if RDR3 is their next game after VI; it still wouldn't come out before 2029-'30, so a 12 year gap between sequels, pretty much in line with GTAV and VI.

If it's 2029-2030, then it might be a cross-gen game, which I don't think rockstar is gonna do again like with GTA 5. I think they'd rather risk to do a smaller crossgen game. 

Edited September 29, 2022 by safde

[Jason]

3 hours ago, CaliMeatWagon said:

That would take a lot to load and run. If they do multiple cities it's going to be maps that we load to. 

 

Pretty sure this conversation has been had before and the answer is no, it wouldn't require loads.

 

Especially with the SSD world we have now they can stream assets in/out fast enough so you load these assets in and out as you travel around a giant map. That was sorta one of the entire points of the SSD's as standards in the new consoles, lol.

 

The only real stumbling hurdle with giant maps nowadays is 1) asset creation and 2) storage space limitations.

[Kolateak]

I wonder if they'll do vehicle classes better for VI

 

I mean sh*t is absurd in GTA V, we got trucks in "Vans", vans in "Muscle", some cars being practically useless for races from how below the rest they are, or leaving all others in the dust from being way better than everything else

[RedDagger]

4 hours ago, Frabbi 01 said:

I think this decade will be the last decade of long development times. As I mentioned before, AI is going to play a major role in development beyond 2030.

 

Have an AI generate all kinds of stuff, then humans can go in and do all the details. This can go for artwork, modeling, mapmaking, voice acting, script writing, and I'm sure almost everything else.

 

Games will not only get larger, we will also see them developed faster. We will see concepts that no one can imagine, as well as full remakes of old games generated at a rapid pace, since an AI could just look at the source and get to work. Then the devs can touch it up and release.

 

The future is going to be weird as hell.

The general trend for AAA games is that when you create tech that allows you to do more for less, instead of producing the same games faster you instead produce larger and larger games. It's like improving hardware, or software being made more efficient; game development expands to fill whatever space was created. Large devs like Rockstar will have long development times until development time simply stops being a factor - which would require AI to generate not just production-ready code, but production-ready code that can tie together all those AI-generated assets. Which will happen tbh, but not this decade

[CaliMeatWagon]

11 minutes ago, Kolateak said:

I wonder if they'll do vehicle classes better for VI

 

I mean sh*t is absurd in GTA V, we got trucks in "Vans", vans in "Muscle", some cars being practically useless for races from how below the rest they are, or leaving all others in the dust from being way better than everything else

 

I would love a system like they have in Forza Horizon. They still have the style class (i.e. muscle, sports), but they have power ratings as well. And while that system isn't perfect, it allows for cross class racing and for more balanced racing. 

[Jason]

1 minute ago, RedDagger said:

The general trend for AAA games is that when you create tech that allows you to do more for less, instead of producing the same games faster you instead produce larger and larger games. It's like improving hardware, or software being made more efficient; game development expands to fill whatever space was created. Large devs like Rockstar will have long development times until development time simply stops being a factor - which would require AI to generate not just production-ready code, but production-ready code that can tie together all those AI-generated assets. Which will happen tbh, but not this decade

 

^^^^^^

 

Better tech that speeds up development times in terms of asset creation will not result in shorter development times for games overall, it'll just lead to bigger, grander and more detailed games.

 

Our expectations as gamers as the reason for this, we expect games to look, feel and play better every single year and games that don't do this aren't received as well as games that do, which will push developers to use the tech to make those bigger, grander games, which ultimately means any time saving benefits of new technology will be nullified.

 

One day, as RedDagger says, we may see this change but it if you ask me that probably won't even be in our life time, at least not for AI code. AI assets might be as little as 10-20 years out but even if that is a correct estimate there's a whole minefield of issues with AI assets that need to be figured out before that becomes a real thing. AI assets be it art, sound, voice (which is starting to be used more) are very delicate subjects in large part because it has the potential to basically render a huge part of the industry obsolete which would decimate jobs.

[CaliMeatWagon]

21 minutes ago, Jason said:

 

Pretty sure this conversation has been had before and the answer is no, it wouldn't require loads.

 

Especially with the SSD world we have now they can stream assets in/out fast enough so you load these assets in and out as you travel around a giant map. That was sorta one of the entire points of the SSD's as standards in the new consoles, lol.

 

The only real stumbling hurdle with giant maps nowadays is 1) asset creation and 2) storage space limitations.

 

The problem with the one big map idea is they would have to explain how all of a sudden a new landmass appeared off to the side. Unless they put enough empty water in between them that the islands are outside of render distance/obscured by weather.

If they went the route of the maps being connected, they would have to design the maps to be that way from the start. So invisible walls we can't pass on land until the next instalment. 

Plus, doing each individual map as its own load will help with the issue of file sizes as each map can be added, or removed, or switched between hard drives. 

 

Side note, does anybody know what the file size is of just the map, not the entire game? 

[Jason]

2 minutes ago, CaliMeatWagon said:

The problem with the one big map idea is they would have to explain how all of a sudden a new landmass appeared off to the side. Unless they put enough empty water in between them that the islands are outside of render distance/obscured by weather.

If they went the route of the maps being connected, they would have to design the maps to be that way from the start. So invisible walls we can't pass on land until the next instalment. 

Plus, doing each individual map as its own load will help with the issue of file sizes as each map can be added, or removed, or switched between hard drives. 

 

Honestly explaining new landmass is one of those things that doesn't really need to be explained, many other games have done similar things. Players know why it's came out of no where, so addressing it in-lore is honestly not needed.

 

The real issue with adding landmass is indeed having to design the map with the idea of it being modular from the start. It doesn't have to be landlocked, they could design a border of the map (likely the northern part, given it's Florida) with the intent of merging it with a new landmass in future.

 

It is not impossible, but it's not straight forward. Instanced maps would make it a doddle but I'm inclined to think we'll see the landmass grow because if anyone has the capability of doing it, it's R*.

[CaliMeatWagon]

1 minute ago, Jason said:

AI assets be it art, sound, voice (which is starting to be used more) are very delicate subjects in large part because it has the potential to basically render a huge part of the industry obsolete which would decimate jobs.

 

Not to mention that it "learns" by sampling other people's work, allowing you to replicate their style. Is that infringement? Is it stealing? Is it morally acceptable? These are all questions that are going to have to be asked as the tech moves forward. 

 

As for putting people out of work, it will, those who don't adapt. But for those who do it will just be another tool for creativity. Much like how computer animation software caused a lot of hand drawn animators to lose work. 

[KoolHoven]

4 hours ago, Frabbi 01 said:

I think this decade will be the last decade of long development times. As I mentioned before, AI is going to play a major role in development beyond 2030.

 

Have an AI generate all kinds of stuff, then humans can go in and do all the details. This can go for artwork, modeling, mapmaking, voice acting, script writing, and I'm sure almost everything else.

 

Games will not only get larger, we will also see them developed faster. We will see concepts that no one can imagine, as well as full remakes of old games generated at a rapid pace, since an AI could just look at the source and get to work. Then the devs can touch it up and release.

 

The future is going to be weird as hell.

We're far away from that goal, like really far. For one, AI generation needs to get a lot better before it can be used on an industrial scale. We've seen how it does with art, which while impressive to maybe the average person for cool pictures, when it really comes to it its quite far off being good. The problem is sentience. AI can recreate what you ask it to, from reference, however AI cant really understand what it is its making. It doesnt have the logic understanding of why a thing exists or what its used for. So you need to add at least 5, maybe 10 years or longer before its even close to useful for that kind of thing on the scale youre talking about. Secondly, everytime new technologies have come out, game development time has increased. Why? Because every time technology has gotten better and game size has gotten bigger. You need to remember has software tools to help developers make games gets better, so does the technology to run these games, and the expectations from gamers to exceed the previous game. Game development time is not going to get shorter for at least another couple of decades at best

[CaliMeatWagon]

1 minute ago, Jason said:

but I'm inclined to think we'll see the landmass grow because if anyone has the capability of doing it, it's R*.

 

I'm of the opposite thought. RDR2's map wasn't much bigger, if it was bigger at all, than GTA V's. From my understanding they are very close in size. 

And there is a balance with map creation. The smaller the map, the more handcrafted and detailed you can make it. The larger the map, the harder that becomes and the more you have to rely on procedural generation to fill in the gaps. 

You also have the aspect of player time. I've played some massive games, so I'm not saying this as someone adverser to travel distances (Hutton Orbital takes over an hour and a half to fly to once in system), but I think the current map size is a good one. Especially considering the pace of gameplay. It's just large enough that you groan and grumble a bit if you have to travel to Poleto (and for people to recommend avoiding those businesses), but not far enough that it really is a long journey. 

And let's be honest here, we are both arguing from wishes, lol.

[Smudjed]

6 minutes ago, KoolHoven said:

We're far away from that goal, like really far. For one, AI generation needs to get a lot better before it can be used on an industrial scale. We've seen how it does with art, which while impressive to maybe the average person for cool pictures, when it really comes to it its quite far off being good. The problem is sentience. AI can recreate what you ask it to, from reference, however AI cant really understand what it is its making. It doesnt have the logic understanding of why a thing exists or what its used for. So you need to add at least 5, maybe 10 years or longer before its even close to useful for that kind of thing on the scale youre talking about. Secondly, everytime new technologies have come out, game development time has increased. Why? Because every time technology has gotten better and game size has gotten bigger. You need to remember has software tools to help developers make games gets better, so does the technology to run these games, and the expectations from gamers to exceed the previous game. Game development time is not going to get shorter for at least another couple of decades at best

Somewhat related to this...

BUT anybody remember the AI tool that was demo'd a while back... the discription was 'the new machine learning tool helps make computer-generated images more realistic by analyzing each frame of the game animation and comparing that to real-life images before applying enhancements based on them'.

I remember there being a fair bit of fuss over it... 

 

IGN: Experimental AI Tool Makes GTA 5 Look Stunningly Photorealistic

[CaliMeatWagon]

5 minutes ago, KoolHoven said:

We're far away from that goal, like really far. For one, AI generation needs to get a lot better before it can be used on an industrial scale. We've seen how it does with art, which while impressive to maybe the average person for cool pictures, when it really comes to it its quite far off being good. The problem is sentience. AI can recreate what you ask it to, from reference, however AI cant really understand what it is its making. It doesnt have the logic understanding of why a thing exists or what its used for. So you need to add at least 5, maybe 10 years or longer before its even close to useful for that kind of thing on the scale youre talking about. Secondly, everytime new technologies have come out, game development time has increased. Why? Because every time technology has gotten better and game size has gotten bigger. You need to remember has software tools to help developers make games gets better, so does the technology to run these games, and the expectations from gamers to exceed the previous game. Game development time is not going to get shorter for at least another couple of decades at best

Corridor Crew just did an episode on using AI for art creation. The technology is moving a bit faster than you would think. 

 

And if you are not already, I would recommend following Dr. Károly Zsolnai-Fehér with 2 Minute Papers. He covers a lot of the advancements that are happening AI and related topics. And he does so based on the research papers, so this is before the public even gets. For instance, most of the people out there are using a neutered version of DHALI-1, but DHALI-2 already exists. 

https://www.youtube.com/c/KárolyZsolnai

 

[Jason]

1 minute ago, CaliMeatWagon said:

I'm of the opposite thought. RDR2's map wasn't much bigger, if it was bigger at all, than GTA V's. From my understanding they are very close in size. 

And there is a balance with map creation. The smaller the map, the more handcrafted and detailed you can make it. The larger the map, the harder that becomes and the more you have to rely on procedural generation to fill in the gaps. 

You also have the aspect of player time. I've played some massive games, so I'm not saying this as someone adverser to travel distances (Hutton Orbital takes over an hour and a half to fly to once in system), but I think the current map size is a good one. Especially considering the pace of gameplay. It's just large enough that you groan and grumble a bit if you have to travel to Poleto (and for people to recommend avoiding those businesses), but not far enough that it really is a long journey. 

 

Travel distance isn't really an issue cause there's many fast travel options they could use in VI (airport, taxis, trains, etc), and I mean when the alternative is loading between maps I'd wager the vast majority of people would prefer one big map w/ fast travel vs split up maps with loads. Size also isn't really a problem, we aren't talking launch map size but long term map size after many updates, new places to explore is high on any open world games list of wanted features, and I'd say the ability to say load up GTAO2 and just straight up go drive to the new areas would be pretty cool vs loading in and being dumped at the new airport or w/e.

 

The main conversation though is really how feasible it is to expand a landmass for R* and the answer to that is that, obviously, we don't know. I'd speculate they'd want to go that route cause it's ambitious and just a little bit more special than splitting up all the maps, but yea, we'll have to see. From a technical perspective simply having and running a giant map is a non-issue in 2022 (or should be), creating a giant map should be a non-issue given the whole idea is they'll expand the map over time, so that really just leaves the technicalities of merging landmasses together.

[Guest]

This has probably been mentioned loads of times in this thread but there's nearly 800 pages.

 

Is there any possiblity we could be seeing a map/city expansion as a DLC in SP? AFAIK it's just some parts of Florida right now?

 

Afterall this project is named "Americas" and there was a leak mentioning there would be a place in South America.

Edited September 29, 2022 by ghasts

[Jason]

6 minutes ago, ghasts said:

This has probably been mentioned loads of times in this thread but there's nearly 800 pages.

 

Is there any possiblity we could be seeing a map/city expansion as a DLC in SP? AFAIK it's just some parts of Florida right now?

 

Afterall this project is named "Americas" and there was a leak mentioning there would be a place in South America.

 

Basically your question is "will the game get SP DLC?" and if we're to look at the previous R* games the answer to that question is "probably not".

 

So... keep yer fingers and toes crossed, pretty much.

[cant remember]

2 minutes ago, ghasts said:

This has probably been mentioned loads of times in this thread but there's nearly 800 pages.

 

Is there any possiblity we could be seeing a map/city expansion as a DLC in SP? AFAIK it's just some parts of Florida right now?

 

Afterall this project is named "Americas" and there was a leak mentioning there would be a place in South America.

 

I guess it's possible but nobody knows for sure.

[KoolHoven]

15 minutes ago, CaliMeatWagon said:

And if you are not already, I would recommend following Dr. Károly Zsolnai-Fehér with 2 Minute Papers. He covers a lot of the advancements that are happening AI and related topics. And he does so based on the research papers, so this is before the public even gets. For instance, most of the people out there are using a neutered version of DHALI-1, but DHALI-2 already exists. 

https://www.youtube.com/c/KárolyZsolnai

 

I mean tbh I dont follow it because art shouldnt be computer generated anyway imo, Assistence sure, but full control feels like a slippery slope imo

[MrBreak16]

Just don't let people like Grove Street Games use AI to make games 

 

[Jason]

5 minutes ago, KoolHoven said:

I mean tbh I dont follow it because art shouldnt be computer generated anyway imo, Assistence sure, but full control feels like a slippery slope imo

 

It's a very slipper slope and one that we're already starting to slide down. AI art is getting really good and we're already seeing it starting to see it used in commercial stuff at the more local level. It's inevitable developers will be looking into to using it on a larger scale but it's going to be a real delicate subject.

[Spider-Vice]

3 minutes ago, MrBreak16 said:

Just don't let people like Grove Street Games use AI to make games 

 

This is a really puzzling one because by the time GSG produced the Trilogy there were already very good AI upscaling techniques available. Not to mention it's clear they were fixed manually but still with typos? Lol...

[Len Lfc]

9 minutes ago, MrBreak16 said:

Just don't let people like Grove Street Games use AI to make games 

 

No they just have Augmented Reality Guitars avaiable for sale.

[cant remember]

"Grove Street Games" sounds like the name a group of enthousiastic GTA modders would give themselves.

[Jason]

AI / proc gen is one of those things where people don't want it being used cause there's some shocking examples of it being used poorly but at the sme time don't realise it's probably used far more extensively in the industry than they realise.

 

Machine learning / AI stuff is being used for some of the best animation systems in the industry, for example - motion matching.

 

Procedural generation in terms of map creation is also used by... pretty much everyone, including R* I believe, but the level of which it's used differs. Some games use it extensively (like NMS) and others use it to build basic maps which they can mould to their liking. A common use for it is for things like foliage, so the placement of plants, shrubs, etc.

 

It's just a tool, basically, used right it helps, used as a crutch it doesn't.

 

Just now, cant remember said:

"Grove Street Games" sounds like the name a group of enthousiastic GTA modders would give themselves.

 

The name actually has zero to do with R* or GTA, weirdly enough.

[(Vip)Met]

Grove Street Games managed to take a very advanced technology and make it look like it was from the 80s

[Zello]

Can’t wait until they rebrand again to Van Der Linde games

Edited September 29, 2022 by Zello

[Gentlebreeze]

I remember when they change the name because it was known as something before. Crazy a year or so later they released the GTA Trilogies DE

[Len Lfc]

Just now, Zello said:

Can’t wait until they rebrand again to Van Der Linde games

Vercetti Estate Games?

Roman Bellic Enterprises Games?

Trevor Philips Industries Games?

[Kolateak]

14 minutes ago, MrBreak16 said:

Just don't let people like Grove Street Games use AI to make games 

 

 

It's just a teaser for the AR Guitar weapon in GTA VI 

 

[(Vip)Met]

1 minute ago, Zello said:

Can’t wait until they rebrand again to Van Der Linde games

I still can't believe of all the names they could have chosen they went for Grove Street Games... I still think they wanted to be part of Rockstar but they rejected them and so GSG went for the closest possible name