Jump to content

DiRTy Gossip

Recommended Posts

Hmm, might be onto something possible here, you wouldn't actually need to do *that* much so far as calculations were concerned, it would be more a 'record' of where you went if you were first, and then simulated cars in front for the stage to be be created live each time (from simulated records, not in real time)  

Basically, have a base stage as we have now, choose how many cars will have run the stage in front of you, apply pre-calculated 'ruts' etc. to the stage, which would obviously have variances as part of their formation, and then 'compile/build' the version of the track you see on stage load.  It would cause longer load times, but might be pretty nice to experience the variation :)  I'm onboard with this version.

This is exactly what I was trying to say, thanks 

Share this post


Link to post
Share on other sites
Hmm, might be onto something possible here, you wouldn't actually need to do *that* much so far as calculations were concerned, it would be more a 'record' of where you went if you were first, and then simulated cars in front for the stage to be be created live each time (from simulated records, not in real time)  

Basically, have a base stage as we have now, choose how many cars will have run the stage in front of you, apply pre-calculated 'ruts' etc. to the stage, which would obviously have variances as part of their formation, and then 'compile/build' the version of the track you see on stage load.  It would cause longer load times, but might be pretty nice to experience the variation :)  I'm onboard with this version.
It would mean that we'd have to read the road, like so many drivers comment on having to do when the stage conditions change so much either due to road order or weather 
Right, and as i commented earlier in this discussion, we don't have the feedback/tech/etc for that level yet, but we can handle the corner not being perfect, we can visually display ruts etc to the level this would run.  I don't think having live deformation would work, nor do i think we can go to the extremes of that deformation, but a definite level of change in the stage surface could be achieved, but that's also a design and implementation all of it's own (programatically setting constraints on how much and in what way it gets changed)  Might be better to have a certain number of canned surfaces, more creation work in the game build, but it'd work given enough variations.

Share this post


Link to post
Share on other sites
BrySkye said:
gfRally said:
now about terrain deformation, I am talking about pre-simulating the change in terrain, not real time.  It would be great for real time sure, but it doesn't have to be real time for the effect to work on gameplay.  I'm not sure how to explain it clearly, but based on your position on the stage the game would simulate (mathematically) the other cars run and lines, even an off.  This would change the look of the road from a previous run.  If you are first on the road, then it will be clean, but you may not fine the best line or grip or what ever advantage/disadvantage that allows.
It's still real time. It has to be in order to be dynamic.
You might be able to to just use an algorithm to generate the track conditions, but the mesh of the road and it's surroundings has to be more complicated in order to be manipulated. The 'new' mesh is still being rendered in real time, even if you're not watching the changes being made, and would feature significantly more polygons than what is in the game right now.

Like I've said, this isn't new tech by any means, but it's not a commonly adopted one.
So ask yourself why it didn't become an industry standard after Sega Rally did it nearly a decade ago. 
(plus then you have the issue of a level playing field given the nature of how Codies want the community aspect to be implemented - leaderboards & leagues)

Generating, or rather modifying mesh must not to be real time, until we consider RX. Mesh might be modified during loading. probably it will cause loading taking longer time, but  it shouldn't be show stopper.
RX is another case, but with current game rigs it's possible to modify surface conditions in semi-real time (see rf2, iR). Modifying mesh is nothing else than doing some calculation on underlaying data. Doesn't matter it's temperature, amount of rubber, grip or mesh shape.

So I assure you - technically it's possible
Question is what approach would be chosen as the first. To me, it would be enough to just modify road surface a bit just to make it vary between rides. It would be starting point which opens new possibilities. Devs have to start from simple goals creating a prototype and going deeper. after that

Share this post


Link to post
Share on other sites
What I was trying to get across is that rendering those grooves and ruts will dramatically increase the polygon count across the whole stage.

The stages will be created to a polygon 'budget'. A max number of polygons to be in any stage as a combination of hitting certain stability and performance targets.
If you were to take the current stages as they are and add the affects of cars going over the track to deform the terrain going over the whole length of the stage, the polygon count of rendering those extra details will shoot up exponentially.

Whether or not it happens in real time or is calculated as part of the loading isn't necessarily the biggest hurdle.
To use the Sega Rally method, you would take the current stage and add another layer of deformable polygons over the top of it.
So things have to be taken away in order to free up the polygon budget enough to create this new deformable layer which can be manipulated, even if it's generated during loading.
That would most likely be stage length.

Remember the stages are already the length they are because of engine and memory limitations.
Throwing a ton more detail in there means something has to be taken away.

Share this post


Link to post
Share on other sites
I wonder if it's possible to fake the 'physics' over the ruts, and have the ruts using that rendering technique i can't remember the name of for creating a non flat looking surface out of a flat surface.

Share this post


Link to post
Share on other sites
I wonder if it's possible to fake the 'physics' over the ruts, and have the ruts using that rendering technique i can't remember the name of for creating a non flat looking surface out of a flat surface.
That'd be bump mapping, most likely. 
Bump mapping is less intensive than raw polygons, but it's not very flexible and wouldn't help in this case.
It's done through textures, which means it can't be rendered in a dynamic way. You can't really generate a truly random element through bump mapping.
To say it again, reasons why it's still not a common feature. :P It's super intensive to do.

Share this post


Link to post
Share on other sites
Yes, it is bump mapping i'm thinking of, my curiosity was as to whether you could fake the physics to line up with the areas they laid down suitably visual bump mapping.

However, in your description, you reminded me why not; it would need a lot of rendering work done for all of it, on top of the fact that it would need a ton of faked geometry for the physics to react to in that place.  Way too much to go wrong even after the ton of work is done.

Not sure we care *too* much about the randomness, i don't think it's practical to go that route, it's also pointless on a start-finish stage (as opposed to a circuit)  You'd have x number of stage conditions rendered depending on how many simulated cars have gone before, and if you wanted to get fancier, have a variation or two for each of those levels.

Share this post


Link to post
Share on other sites
Could they perhaps make a handfull of different "stages" of  deformation. They would be a solid (not deformable) surface, that is shaped as if it has been deformed by previous cars, replicating a deformed surface. 
For instance they could do a clean surface as we have now,  a surface after 3/4 cars, after 7/8, after 15, after 25, then a "last in the line" thats fully worn. I know its not a truly variable surface, but it could make it a bit more interesting with 6 or so different surface qualities.
For a rally, is there much need to have the surface changing in real time as you go over it, if there's only AI following you? So they are basically predetermined settings much like "rubbered in" variants on circuit racing games. With the different being a physically difference surface is being swapped not just grip values.

Edit: This could also allow for different surface models being used for weather conditions, such as more/deeper puddles in a heavy rain effected stage.

Share this post


Link to post
Share on other sites
austinb said:
Could they perhaps make a handfull of different "stages" of  deformation. They would be a solid (not deformable) surface, that is shaped as if it has been deformed by previous cars, replicating a deformed surface. 
For instance they could do a clean surface as we have now,  a surface after 3/4 cars, after 7/8, after 15, after 25, then a "last in the line" thats fully worn. I know its not a truly variable surface, but it could make it a bit more interesting with 6 or so different surface qualities.
For a rally, is there much need to have the surface changing in real time as you go over it, if there's only AI following you? So they are basically predetermined settings much like "rubbered in" variants on circuit racing games. With the different being a physically difference surface is being swapped not just grip values.

Edit: This could also allow for different surface models being used for weather conditions, such as more/deeper puddles in a heavy rain effected stage.
It's still more polygons. A rut needs to be formed out of polygons and requires more of them than something flat.
Think of it as the result being a big part of the problem, as opposed to the method of how you get there.
In super laymens terms, think of drawing a single, flat, straight line.
_  
Then draw a rut out of straight lines. You need to draw more than one line to do it.
v

We do already have some measure of deformation to  the game. We can usually see it in some parts of Greece, like the string of hairpins.
Though I think it's just a purely visual effect.

Let's also take a breather to think of the 'deformable' snowbanks in Sweden, they aren't really.
There is a simulated feeling, and that's good enough for the driving, but the snow bank isn't physically changing.
If you were to drive head on into one of those snow banks, then reverse out, there won't be any kind of imprint on the snow from the car.

Share this post


Link to post
Share on other sites
austinb said:
Could they perhaps make a handfull of different "stages" of  deformation. They would be a solid (not deformable) surface, that is shaped as if it has been deformed by previous cars, replicating a deformed surface. 
For instance they could do a clean surface as we have now,  a surface after 3/4 cars, after 7/8, after 15, after 25, then a "last in the line" thats fully worn. I know its not a truly variable surface, but it could make it a bit more interesting with 6 or so different surface qualities.
For a rally, is there much need to have the surface changing in real time as you go over it, if there's only AI following you? So they are basically predetermined settings much like "rubbered in" variants on circuit racing games. With the different being a physically difference surface is being swapped not just grip values.

Edit: This could also allow for different surface models being used for weather conditions, such as more/deeper puddles in a heavy rain effected stage.
This is exactly what I'd love to have. No need for it to be dynamically generated, a handful of "worn in" variants do the trick just as well—design for effect.

I imagine it would still be labour-intensive, since you would also need to create the weather and time-of-day variants of each of the dirty variants... but at least the end result wouldn't be memory-intensive.

Share this post


Link to post
Share on other sites
dwkGravey said:
austinb said:
Could they perhaps make a handfull of different "stages" of  deformation. They would be a solid (not deformable) surface, that is shaped as if it has been deformed by previous cars, replicating a deformed surface. 
For instance they could do a clean surface as we have now,  a surface after 3/4 cars, after 7/8, after 15, after 25, then a "last in the line" thats fully worn. I know its not a truly variable surface, but it could make it a bit more interesting with 6 or so different surface qualities.
For a rally, is there much need to have the surface changing in real time as you go over it, if there's only AI following you? So they are basically predetermined settings much like "rubbered in" variants on circuit racing games. With the different being a physically difference surface is being swapped not just grip values.

Edit: This could also allow for different surface models being used for weather conditions, such as more/deeper puddles in a heavy rain effected stage.
This is exactly what I'd love to have. No need for it to be dynamically generated, a handful of "worn in" variants do the trick just as well—design for effect.

I imagine it would still be labour-intensive, since you would also need to create the weather and time-of-day variants of each of the dirty variants... but at least the end result wouldn't be memory-intensive.
Has anyone been reading the last page at all?!  This is what we've been saying for two pages.

Anyway, @BrySkye is right on it anyway, it will need more polygons regardless, it's a large perf hit and needs to be part of the engine when designing it - remember part of the shorter length of our current stages is due to the performance constraints to include all the detail we have.

Share this post


Link to post
Share on other sites
BrySkye said:
Think of it as the result being a big part of the problem, as opposed to the method of how you get there.
In super laymens terms, think of drawing a single, flat, straight line.
_  
Then draw a rut out of straight lines. You need to draw more than one line to do it.
v

We do already have some measure of deformation to  the game. We can usually see it in some parts of Greece, like the string of hairpins.
Though I think it's just a purely visual effect.

Let's also take a breather to think of the 'deformable' snowbanks in Sweden, they aren't really.
There is a simulated feeling, and that's good enough for the driving, but the snow bank isn't physically changing.
If you were to drive head on into one of those snow banks, then reverse out, there won't be any kind of imprint on the snow from the car.
Sure, but if that was the goal from the outset, then they would budget for the most deformed version of each stage. I imagine baked lighting and textures would also go some way in communicating the look. Tarmac stages would be almost all texture work of skid marks and dirt on the roads.

Edit:
F2CMaDMaXX said:
Has anyone been reading the last page at all?!  This is what we've been saying for two pages.

Anyway, @BrySkye is right on it anyway, it will need more polygons regardless, it's a large perf hit and needs to be part of the engine when designing it
Ha, ok I'll give you an "agree" for that. :)

Share this post


Link to post
Share on other sites
This is exactly what I'd love to have. No need for it to be dynamically generated, a handful of "worn in" variants do the trick just as well—design for effect.
Too repeatable, requires too much effort to prepare similar but different versions of roads.
Anyway, @BrySkye is right on it anyway, it will need more polygons regardless, it's a large perf hit and needs to be part of the engine when designing it - remember part of the shorter length of our current stages is due to the performance constraints to include all the detail we have.

I think you guys spend too much time on analysing if it can be done or not. Funny is, that some of us are  trying to bias discussion without needed knowledge about programming, actual techs, engine abilities, various methods available to achieve goals. No - you don't need more dense mesh for most cases. You can do that for example with mix of underlying physical layer and bump-mapping over the top of textures. Yes: bump mapping can be generated dynamically in the same way as any other texture (some one stated it can't). We can see that in D:R when car is "deforming" loose surface (ie going sideways or when hard breaking)
On the other side, gfx cards are more and more powerful. 10 years ago 1 square meter polygon was quite common. That's why iRacing uses underlying physical layer to model micro bumps. But in recent sims you can get mesh 10x10cm or smaller making existence of physical layer questionable.
Maybe, when developers will be ready, gfx cards will be powerful enough to render 1 square cm mesh. We don't know that. Maybe CM will develop the feature in more optimized way than we think. As for example let's compare grass technology developed by CM and SMS. CM uses some techniques which resulting nice, blended, resources optimized grass, while SMS' grass (put into pCars) has a lot of flaws and requires a lot of computing power.

We should act like customers: we need feature A, we need feature B. That's all.
What is the point to justify features for development or (which is even worse) arguing against it at this stage? Doesn't matter what we say here, we will not decide about it. Decision belongs to developers and publishers. For years simulations had have no dynamic road technology. In the past some faked solutions was in use (live track in GTR, GTR2 etc). Now we have Real Road in rF2, New Surface Model in iRacing. It's time to add dynamic road to rally simulations. One day it will be delivered to us for sure.

Share this post


Link to post
Share on other sites
MaXyMsrpl said:
We should act like customers: we need feature A, we need feature B. That's all.
Sorry, but I prefer not to just be blissfully ignorant of the realities of game design.
I like to know and consider the compromises that are required to get a game out of the door. Where resources are allocated, how things are prioritised.
I think the process is important.

When people don't care about those aspects, they tend to descend into saying things like developers are just "lazy" when certain features aren't implemented, and that's something which irritates me.

So please don't tell me that I should just want things and not care about what it takes to make it happen.

Share this post


Link to post
Share on other sites
Developers are not lazy. They are employees working for company... and now important part: for which maximizing profitability is the major goal.
If absence of a feature is being justified by customers their self, you can be almost sure the feature will be never developed (save resources, get more profit). It's really as simply as that.

Another thing is what I said already: we have no enough data/knowledge to judge if it is doable or not. Also because the conditions under which some thing are being developed is constantly changing. What is not possible today, it might be tomorrow. But tomorrow have to be scheduled today.

Share this post


Link to post
Share on other sites
In DiRT facebook they posted a tbt of the CMR1, which in the box have Colin's Subaru... Maybe what they wanted this 5 years was the Subaru Impreza (dont want to say the exact name because i am maybe wrong and here there are a lot of subbie fans) but i think it is the 22b


:smiley: 
In DiRT facebook they posted a tbt of the CMR1, which in the box have Colin's Subaru... Maybe what they wanted this 5 years was the Subaru Impreza (dont want to say the exact name because i am maybe wrong and here there are a lot of subbie fans) but i think it is the 22b
I may be wrong here, but isn't that the 1995 GrpA that's currently in the game?
It's not. Easy way to tell between Group A and early Impreza World Rally Cars is that the Group A was a four door and the early World Rally Cars... 

Darhour said:


I hope this would be a gossip material; it was posted on Facebook. By the way, 1997 Impreza GC8D or arguably whatever lol

It was a tbt, not sure if a gossip material or not because in this thread it has been lately debated that there are no mentions of cars addons.
...were all two doors. The car in the photo is an Impreza S4 WRC. 

Ah, that's better. I'm sure it's been well over 5 pages without a Subaru model discussion. It just didn't feel like the DiRTy Gossip thread anymore...

Share this post


Link to post
Share on other sites
^^^ P2000 ^^^   :D

I think it's safe to conclude that 'surface deformation', whether dynamic, or pre-canned, is not what Paul hinted at in this instance, but a pipe-dream for the future possibly.

I still think a lot of fingers are pointing toward the R5 Maxi ?

Share this post


Link to post
Share on other sites
Has anyone been reading the last page at all?!  This is what we've been saying for two pages.

Anyway, @BrySkye is right on it anyway, it will need more polygons regardless, it's a large perf hit and needs to be part of the engine when designing it - remember part of the shorter length of our current stages is due to the performance constraints to include all the detail we have.

Yeah sorry I started my comment, got distracted, didnt refresh the thread, and 45 mins after starting I posted my comment, and in that you must have commented my same point haha.

Obviously having a "grooved/worn" surfaces will add more polygons than on the "clean" equivalent, but that increase will surely be very minor compared to a having deep layer of deformable polygons on top. That was my reason for the suggestion. 

Share this post


Link to post
Share on other sites
urgaffel said:
The polygon density on the road in Sega Rally Revo is crazy :open_mouth: 
Yeah, it was a 720p/30fps cap despite the somewhat simple nature of the general scenery and the cars (though this was at least in keeping with the Sega aesthetic).
But it was far from solid on the consoles.

The arcade version, Sega Rally 3 (which Jamie definitely did work on!) was 720p/60fps with Sega's Europa-R arcade board.
Basically a PC.
Intel Pentium D 945 (3.4 GHz, dual-core), 8GB RAM and a GeForce 8800.

Actually, the Europa-R board is the same one used for the arcade version of GRID!

Share this post


Link to post
Share on other sites
Paul said it had been 'approved' - I don't see surface deformation as something that has to be approved from elsewhere, more something they may have looked into and decided whether it is possible or not

Share this post


Link to post
Share on other sites
It's a more vague statement than you might think.
Could also be a matter of getting approvement (and budget) from management 

Share this post


Link to post
Share on other sites
Evilsmurf said:
"approved" does suggest a licence of some sort, with Renault being the most likely.  
Can't imagine that Paul has been dreaming for a Renault license for over 5 years, especially when DiRT 3 had a Renault in it. So no, it's by 99% not a piece of content like a track or a car, but a feature more of the ant-cheat sort that can work in all their games, regardless of their genre/content.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×