Rally Stages Moddeling Costs

[FMaiso99 48]

I was wondering how much it costs to Codemasters to make the 20km of each Rally we have at DiRT Rally

[thecraicbear 0]

If it's not too rude of us to ask, I wouldn't mind knowing of a rough figure, purely out of curiosity and to gain some perspective on that aspect of game design. 

[BrySkye 14]

It's going to vary a lot depending on the location and how many team members are sent there.
Being based in the UK, Pikes Peak will have been the most expensive in the game by some margin not only because it needed two (at some point three) versions, but also because the costs of flying some of the team over to the US is drastically more expensive than the European locations.
That's going to be the main reason why the game is, with the exception of Pikes Peak, 100% European.
It's more practical to chuck some of the team on an easyJet flight to Greece for a few weeks.
Assuming a European location, the research trips alone probably cost a few grand, depending on how much accommodation varies (for Pikes Peak, the flights along will have cost a fair few grand!)

In terms of work hours to model all the assets, texture them, build the track...
Unless that's already been stated somewhere, I don't think we could even really speculate how long that is and what it translates to in terms of wages.

[FMaiso99 48]

BrySkye said:

It's going to vary a lot depending on the location and how many team members are sent there.
Being based in the UK, Pikes Peak will have been the most expensive in the game by some margin not only because it needed two (at some point three) versions, but also because the costs of flying some of the team over to the US is drastically more expensive than the European locations.
That's going to be the main reason why the game is, with the exception of Pikes Peak, 100% European.
It's more practical to chuck some of the team on an easyJet flight to Greece for a few weeks.
Assuming a European location, the research trips alone probably cost a few grand, depending on how much accommodation varies (for Pikes Peak, the flights along will have cost a fair few grand!)

In terms of work hours to model all the assets, texture them, build the track...
Unless that's already been stated somewhere, I don't think we could even really speculate how long that is and what it translates to in terms of wages.

Yes, but a pure approximately one, i know it is different a km from Monte Carlo stages than a km in Finland, but like €200 per km for example

[KevM 44]

                                                                                                                                                                     

[BrySkye 14]

Fedemaisonnave said:

Yes, but a pure approximately one, i know it is different a km from Monte Carlo stages than a km in Finland, but like €200 per km for example

Like I said, it's impossible to make even an approximate figure for the cost without having an approximate figure for the total work hours from all the people who worked on that stage and the average wages per hour.
It's possible there might be a mention of how many hours go into making a stage in an interview somewhere, but I don't think we're going to get information on peoples wages.
So the only way we can really get an answer to this is if someone like @KickUp gives us a very approximate average, but generally those aren't the kind of figures which get freely disclosed. Same with anything budget related.

Pulling a figure out of my arse, I doubt it would be any less than £75,000 per location, minimum.
So something closer to €4000 per km.
Modern game development be expensive.

[Evilsmurf 90]

I know it takes 500 man hours to build the stage, but that's just the modelling. 

[Bongomaster 3]

I would reckon they really need to take another look at how they create stages. Sorry, but 6 rallies with just over 20 unique kilometres each is not enough for a really game. Is it really necessary to travel a team to each rally location? If so, use laser scanning to reduce modeling time. How about sandbox levels (even of this can be made available to modders) where you have templates and terrain deformation tools? All stuff that needs to be considered for dirt rally 2. We need to be able to forget what lies around the corner. 

[BrySkye 14]

Bongomaster said:

I would reckon they really need to take another look at how they create stages. Sorry, but 6 rallies with just over 20 unique kilometres each is not enough for a really game. Is it really necessary to travel a team to each rally location? If so, use laser scanning to reduce modeling time. How about sandbox levels (even of this can be made available to modders) where you have templates and terrain deformation tools? All stuff that needs to be considered for dirt rally 2. We need to be able to forget what lies around the corner. 

That's the old quantity vs quality argument.
Can't really have both, there's a compromise to be had there. If you want attention to detail, yeah, people are going to have to actually go to the locations.
Laser scanning for roads can be compared to SRTM data for terrain in that people assume it's an easy silver bullet type solution.
It doesn't mean that a huge amount of work no longer has to be done in order to turn that data into something usable in a game.
It's not strictly cheap either.
Plus it's one thing to laser scan a nice 6km circuit which can go days without a single car on it and a public road in the middle of nowhere which is what makes up most rally stages.
Damned if you do, damned if you don't. Generally, people want real rally stages, not randomly generated made up ones.

[dwkGravey 0]

€200 per km! Just laser scan it!

Man, I'd love to see a reality show where the average forum poster is put in charge of a game studio for a week. :lol:

[mesa 637]

BrySkye said:

Plus it's one thing to laser scan a nice 6km circuit which can go days without a single car on it and a public road in the middle of nowhere which is what makes up most rally stages.
Damned if you do, damned if you don't. Generally, people want real rally stages, not randomly generated made up ones.

Assetto Corsa and  iRacing devs already did the laser scanning of the whole Nordschleife which is 22 KM long.
In other hand there are plenty of Gmaps also available,  with a good  terrain or track editor which can work with Gmaps the develop of a stage could be much more cheaper.

[enamel 0]

I like to imagine Paul furiously writing down all these suggestions, like his team isn't the best in the world at creating rally stages "Oh man, why didn't we think of that!" :D

[JZStudios 636]

Bongomaster said:

I would reckon they really need to take another look at how they create stages. Sorry, but 6 rallies with just over 20 unique kilometres each is not enough for a really game. Is it really necessary to travel a team to each rally location? If so, use laser scanning to reduce modeling time. How about sandbox levels (even of this can be made available to modders) where you have templates and terrain deformation tools? All stuff that needs to be considered for dirt rally 2. We need to be able to forget what lies around the corner. 

Laser scanning isn't really cheaper, easier or quicker. there's a lot of retopology and stitching involved unless they use a low resolution, in which case, laser scanning is actually still not practical. And they would still have to send a team of photographers and computer guys to scan the track and photograph the surroundings. They aren't going to laser scan a 10 mile view radius or all of the track side objects. A good laser scanner will give millions of vertices on something as small as a rock which needs to be retopologized. Sandbox levels also completely eradicate what you just suggested for using a laser scanner, and even still it takes just as much time to get something to look good, and chances are that's pretty close to how they're doing it.

[versedi 39]

I'm curious if it could be possible to use someone from community to do the research when he's living at the place already. How much specialized tech/equipment is already involved in the process of research.

[Madhun1967 54]

satellite maps and onboard footage i think also could be good references.For me that tracks wouldnt have to be perfect replicas of actual stages,just close enough to get the location and the  feel

[ramPage16 0]

JZStudios said:

Laser scanning isn't really cheaper, easier or quicker. there's a lot of retopology and stitching involved unless they use a low resolution, in which case, laser scanning is actually still not practical. And they would still have to send a team of photographers and computer guys to scan the track and photograph the surroundings. They aren't going to laser scan a 10 mile view radius or all of the track side objects. A good laser scanner will give millions of vertices on something as small as a rock which needs to be retopologized. Sandbox levels also completely eradicate what you just suggested for using a laser scanner, and even still it takes just as much time to get something to look good, and chances are that's pretty close to how they're doing it.

Would it actually be less efficient than creating every stage by hand? No doubt it's more expensive and complicated, but shouldn't it still be faster in the context of making a number of long stages? I wonder if using a mobile laser scanner gets around the stitching issue. I saw in an article that iRacing used mobile scanning for Nordschleife.

How much is retopology an issue? I ask out of ignorance. Anything on the edge of the road like rocks, cliff faces, bushes, etc would be priority. Everything else like trees, houses, landmarks and geography aren't needing to be very accurate. The only laser scan data that's actually being used is the road surface and anything in its immediate proximity. The rest of the data would be used as a reference of what to make the environment look like.

If a Dirt Rally 2 is to be made and they're wanting to include long stages, I hope they seriously look into laser scanning and weigh the pros and cons. I think it could possibly benefit them in the long run.

[dwkGravey 0]

ramPage16 said:

If a Dirt Rally 2 is to be made and they're wanting to include long stages, I hope they seriously look into laser scanning and weigh the pros and cons.

I'm pretty confident they already have. @KickUp has talked about why they still use cameras and tape measures instead of laser scanning.

I'm sure a lot of racing game studios and/or their PR like to talk up their laser scanning, since it's new and it has the word "laser" in it; but there's a big element of spin there, as they all try to keep up with the Joneses. Like @BrySkye said, just because it's new doesn't mean it's a magic bullet, automatically creating better tracks, faster and cheaper, for every type of racing game.

[ramPage16 0]

dwkGravey said:

I'm pretty confident they already have. @KickUp has talked about why they still use cameras and tape measures instead of laser scanning.

I'm sure a lot of racing game studios and/or their PR like to talk up their laser scanning, since it's new and it has the word "laser" in it; but there's a big element of spin there, as they all try to keep up with the Joneses. Like @BrySkye said, just because it's new doesn't mean it's a magic bullet, automatically creating better tracks, faster and cheaper, for every type of racing game.

For sure. Like I said, weight the pros and cons. If it's too expensive and takes away too much from the rest of the game then clearly it's not worth it. However, if it's possible then it would be a really exciting feature.

[JZStudios 636]

ramPage16 said:

JZStudios said:

Laser scanning isn't really cheaper, easier or quicker. there's a lot of retopology and stitching involved unless they use a low resolution, in which case, laser scanning is actually still not practical. And they would still have to send a team of photographers and computer guys to scan the track and photograph the surroundings. They aren't going to laser scan a 10 mile view radius or all of the track side objects. A good laser scanner will give millions of vertices on something as small as a rock which needs to be retopologized. Sandbox levels also completely eradicate what you just suggested for using a laser scanner, and even still it takes just as much time to get something to look good, and chances are that's pretty close to how they're doing it.

Would it actually be less efficient than creating every stage by hand? No doubt it's more expensive and complicated, but shouldn't it still be faster in the context of making a number of long stages? I wonder if using a mobile laser scanner gets around the stitching issue. I saw in an article that iRacing used mobile scanning for Nordschleife.

How much is retopology an issue? I ask out of ignorance. Anything on the edge of the road like rocks, cliff faces, bushes, etc would be priority. Everything else like trees, houses, landmarks and geography aren't needing to be very accurate. The only laser scan data that's actually being used is the road surface and anything in its immediate proximity. The rest of the data would be used as a reference of what to make the environment look like.

If a Dirt Rally 2 is to be made and they're wanting to include long stages, I hope they seriously look into laser scanning and weigh the pros and cons. I think it could possibly benefit them in the long run.

If you look at iRacing's little video about scanning the track, they said it takes 10 people 6 months to produce a track. Laser scanners are all portable, and they make millions of vertices in a point cloud. I don't know how much track vertices iRacing uses, but it definitely needs to be seriously reduced, and they keep all the major points. There are some auto stitching softwares out there, but they do tend to still need some manual input. Roadside objects, yes are a bit less important, but you could still scan them so they are actually in the accurate position, and again need to be retopologized. I guess you could argue the productivity point, but if they're going out for a close resemblance, it's less work. That being said, laser scanning is still cool and would be appreciated, but people overlook all the work that goes into it. Proper laser scanning also requires that they get both sides of track and any object they scan to make sure they get any hidden faces correct.

This is technically photogrammetry, but the process is actually very similar, and it's a cool video anyways.
https://youtu.be/eXMJr-aB4qw

[Bongomaster 3]

Woah, ok guys you don't like the laser scanning idea, I get it. I'm actually in the industry of laser scanning so I like to think that I actually do know what I'm talking about here. I'm just trying to say that there needs to be a more cost-effective stage-development solution, because 23km of roads per rally isn't enough.

I can confidently say that a Digital Terrain Model from scanner data is a much more effective solution than cameras and tape measures  :D. Then you just need to worry about props, textures, etc, but still you've saved a bunch of time on the actual road surface which is what counts in a rally game.

[JZStudios 636]

Bongomaster said:

Woah, ok guys you don't like the laser scanning idea, I get it. I'm actually in the industry of laser scanning so I like to think that I actually do know what I'm talking about here. I'm just trying to say that there needs to be a more cost-effective stage-development solution, because 23km of roads per rally isn't enough.

I can confidently say that a Digital Terrain Model from scanner data is a much more effective solution than cameras and tape measures  :D. Then you just need to worry about props, textures, etc, but still you've saved a bunch of time on the actual road surface which is what counts in a rally game.

No one said we don't like it. I was just saying it's not as quick and easy as everyone thinks it is.

[BatteryPowered 0]

Scanning is extremely long and tedious process. I work in engineering industry and we're doing some high resolution scanning of various installations.
First you would level scanner (which is pretty heavy depending on model) on tripod, followed by deploying several markers around scanner for reference points. After which scanner would slowly do a 360 scan of area that is approximately 10 square meters. Next you move the scanner, level it again, deploy markers and repeat the process. Usually each scanner needs a team of 2-3 people. And then you have to spend lots of time cleaning up, stitching and interpreting the collected data before it can be useful for game development. 

I think scanning is only applicable for closed loop tracks, where there's a finite distance and each bump matters. In rally game you don't have to go to such level of details and it would definitely not save you any time developing levels.