Polyphony Digital Is Experimenting With Procedurally Generated Cars

I know computers are suppossed to help us better our design ideas... but these images are nothing but proof that some folks are too lazy to freehand something asthetically or subjectively beautiful.

Staring at a blank screen, the "talent challenged artist" artist tosses hands into the air in disgust... 'screw this!... l'll let the machine do the work for me'.

No thanks. I'd rather freehand mine own designs than let some machine do it for me. Not a pretty image to be found on this page but for Steve's Saleen S7.

It's more of a necessity in PD's case,game development takes alot of time these days,they have to cut corners somewhere,but i'm sure the car models will have PD's attention of detail even with procedural generation.
 
It's more of a necessity in PD's case,game development takes alot of time these days,they have to cut corners somewhere,but i'm sure the car models will have PD's attention of detail even with procedural generation.
As if laser scanning isn't enough? There goes another talented artist out the door, hamstung by a CAD Design student loan. But it's the way of the world... sadly.
 
As if laser scanning isn't enough? There goes another talented artist out the door, hamstung by a CAD Design student loan. But it's the way of the world... sadly.

It's either that or PD stops making Gran Turismo games,they held as long as they could but i think its reasonable to say that PD's love for making everything "handmade" hurt GT Sport's day 1 car list.
 
It's either that or PD stops making Gran Turismo games,they held as long as they could but i think its reasonable to say that PD's love for making everything "handmade" hurt GT Sport's day 1 car list.
No more GranTurismo?... please say it ain't so, Autitore! I'm personally waiting on the next full GT release.
I'm no 'know-it-all' but as far as I know, many other developers are scanning most, if not all, environments including cars for a realistic experience.
For many years now, I have been thinking of a sim racer that would allow players to design their own cars within certain parameters and guidelines in accordance to a specific class, effectively foregoing the need for licensing... GrandTheftAuto style.
I cain't think of a worse solution than PGing to solve a financial issue. Besides... you'd think this pay-to-play model, Sony, PD and Kaz have embraced, would generate the kind of cash they are looking for. But I just think that they are just plain greedy... like the lot of them.
 
Honestly, this is nonsense. It doesn't replace artists at all. Just the same as all the other layers of automation haven't replaced artists yet.

What it does is free up artists' time to focus on the areas where their artistry makes the greatest difference. No one is actually forced to use it if a certain type of artistic expression is desired.

The guy who created that "art" is a programmer. "Programmer art" is basically a games development in-joke.
 
In my times trying to game program, I have been accustomed to the concept of procedural modeling and such. Thing is, I don't know what procedural cars would entail. I've tried to learn procedural content, but I have yet to fully understand or implement such features. So I am wondering what can actually be accomplished with PD experimenting with procedural modeling in this case.
 
The adaptive tessellation was already a procedural model of sorts, at least the warping between one mesh configuration and another was. The way that mesh was turned into a progressive one in the first place is a procedural process, nobody did that by hand.

Trees are a great example, they've been procedural for many years now in some cases for obvious reasons. Any foliage, in fact - especially placing all those grass sprites!!

Any other scenery items, such as a stone wall or a house or a bridge etc. would benefit from some proceduralisation (e.g. exact placement, colour and condition of constituent stones in said stone wall). These can still be hand tweaked after the effect.

Any kind of randomisation of content effectively involves proceduralisation. Games like the Signal from Tolva use procedurally generated loadouts for enemies. It's more procedural and not just purely random, because some effort is made to create a cohesive look if nothing else.

Materials shading is effectively procedural; e.g. PD didn't paint the carbon fibre onto the cars, they just lined up the coordinates and scale for the material shader to do the rest.

On that note, many materials now are designed as combinations of layers of different procedures - PBR is a procedure that alleviates some of the painstaking work of colour matching, shading and other artistic mixing / blending of colours and materials on screen, and opens up new tools for different kinds of artistic expression, depending on how you "tune" it.

Photogrammetry is one big mathematical procedure that takes e.g. aerial photographs as its inputs and gives something like the coursemaker environments as its output.


Obviously, as far as content goes, we've got things like the course generator, but we could possibly also get a livery generator, championship / career generator, racing series (regulations) generator etc. etc. There's a lot that could be done, but it will take a significant amount of time to tune properly, but once in place and working will be highly valuable.


Still: be very cautious about equating "procedural" with "random", as it doesn't have to be. Take the demoscene, much of it is procedural (in some cases from an embarrassingly small filesize), but it's the same every time (demo) because it is at least partially generated by fixed mathematical processes at runtime instead of being entirely stored pre-made. This applies to the visual content, the music, the animations etc. All of it.

In that sense, procedural generation also has (always had) substantial usefulness in performance optimisation, by trading different resources off against each other
And, as a final note, all of the platform graphics calls, for example, are procedures. One of the things that made the demoscene so striking is that they often made their own procedures for drawing to create visuals that just weren't available via the "standard" APIs (e.g. DirectX) and their own workflows that serve their purposes, rather than the same old editing tools everyone else was using (e.g Unity). This is getting harder to do these days, though.


People often say that procedural generation is the future of gaming, but the reality is that it is already its past and present.
 
I'm sure sometimes Kaz looks at a Omega Boost poster at his desk and thinks "oh yeah this had potential" then he goes back to look at which Nissan Skyline is coming next to GTSport

I wish they made an Omega Boost sequel, it's genuinely one of my favourite PS1 titles ever.
 
For reference, there are games like the independent/indie game "No Man's Sky" that features many different procedural elements. There is even a game I found on Steam called "Racecraft" which is an eSports racing game where everything from cars to tracks can be generated procedurally. It is all very interesting. I could understand the idea of possibly a procedural environment. However, procedural cars? I almost sort of thought of if Gran Turismo 5 or Gran Turismo 6 were re-done so that you could have some better implementation of certain visual upgrades or (especially) more interesting generated/custom courses.

I will am interested in seeing what Polyphony Digital could possibly come up with using procedural technology.
 
Man, that's pretty cool! I think that many games should give players ability to randomise car, especially in funny games like RL or GGE. Really waiting for this ability.
 
So this would probably be possible, huh? 💡:lol:

images
 
Back