The GT Sport Epic Whining and Crying Thread

[mister dog]

At least we are mostly being civil.... well except for Hitler.

I feel his pain though, Forza 1 only launched in 2005.

 

[Johnnypenso]

You crazy bro... it was laser scanned, I know it in my gut!

BTW Turn 10 had to rent out Nurburgring for two days to do it.

Pretty sure I read that all the recently laser scanned version of the Nords come from a single laser scanning source provided by the Nurburgring themselves. T10 probably rented the track to take photos to flesh out the model of the track in game.

 

[Ridox2JZGTE]

Both sources claim 5mm at 50 meters.

From this, we can conclude that if it was in fact laser scanned with a CP3200 that it would be accurate up to 5mm +/- depending on distance of 50 meters +/-.

So I'm going to call shenanigans on this idea.

In fact, since the maximum range of the device is 120 meters then I would even say that it might even be impossible for it to be as inaccurate as 15mm from it's max distance.

The Nurburgring GP track is 10-13 meters wide.

http://www.nuerburgring.de/en/fans-info/race-tracks/grand-prix-track.html

Which is about as wide or wider then any other part of Nurburgring.

So in this scenario, Polyphony would have had a truck with CP3200s scanning a track with a width of about 10-13 meters and a device capable of 5mm at 50 meters so let's say about 1.5 mm at 10-13 meters as a guess.

How could that possibly result in 15 mm?

Not just track width, that would be a waste of the time/energy. They would have scanned everything that can be captured surrounding the scanner, including the trees, hills, barriers, everything at maximum distance while still maintaining reasonable accuracy data point cloud/reflectivity of the objects . Not sure if the scanner can stay accurate above 50m, but scanning at maximum distance would have saved a lot of time, instead scanning slowly every 13m. This is just me thinking by simple reasoning If you look at the specs, the survey point cloud also quite low by today standard, 562.544 compared to 24.350.000 ( today tech ) and only 140 vertical field of view. It's a balance thing, and the specs are not always what the result would be, they are just claimed numbers by manufacturer that can be achieved ( more like theory ). Who knows, it could only do 7-10mm at 50m depending on the reflectivity of the objects/surface. Maybe PD decide on 15mm accuracy setup is enough, scan between 50-70m perhaps, and with only half a million data point within that radius, I can only speculate.

Oh, they may not scan from onboard the truck while rolling, as far as I know, the old device has to stand stationary and with low vertical FOV, need to be as close as possible to the surface, which likely used a pod/stand on the ground.

 

[FordGTGuy]

Pretty sure I read that all the recently laser scanned version of the Nords come from a single laser scanning source provided by the Nurburgring themselves. T10 probably rented the track to take photos to flesh out the model of the track in game.

From what I can find, it looks like they scanned it themselves:

Turn 10 were allowed to book the ‘Ring out completely for two days to scan it to their satisfaction. With an all-new, next-gen compatible laser scanning set up to capture every last, unforgiving, terrifying inch. “Laser scanning’s been around for a while,” continues Dan, “and we’d looked at laser scanning at last generation, but found it wasn’t detailed enough for what we really wanted to accomplish. So we invested in new laser scanning rigs and a new process for Forza 5, and most of the tracks were laser scanned.”

Not just track width, that would be a waste of the time/energy. They would have scanned everything that can be captured surrounding the scanner, including the trees, hills, barriers, everything at maximum distance while still maintaining reasonable accuracy data point cloud/reflectivity of the objects . Not sure if the scanner can stay accurate above 50m, but scanning at maximum distance would have saved a lot of time, instead scanning slowly every 13m. This is just me thinking by simple reasoning If you look at the specs, the survey point cloud also quite low by today standard, 562.544 compared to 24.350.000 ( today tech ) and only 140 vertical field of view. It's a balance thing, and the specs are not always what the result would be, they are just claimed numbers by manufacturer that can be achieved ( more like theory ). Who knows, it could only do 7-10mm at 50m depending on the reflectivity of the objects/surface. Maybe PD decide on 15mm accuracy setup is enough, scan between 50-70m perhaps, and with only half a million data point within that radius, I can only speculate.

Oh, they may not scan from onboard the truck while rolling, as far as I know, the old device has to stand stationary and with low vertical FOV, need to be as close as possible to the surface, which likely used a pod/stand on the ground.

If it required to be stationary then we know for a fact they didn't do it.

You know how much time it would take to scan a 13 mile track 13 meters at a time?

Let's say they scanned every 13 meters, like you suggest, every 10 minutes.

If they scanned 8 hours a day, it would take 11.12 days in a row to scan the track....

do we honestly believe this?

Also you keep bringing up 50 meter accuracy, we are talking about the accuracy of the track surface and at most the track distance from the scanner is going to be 6-7 meters at max when scanned.

15 mm, even with reflective surfaces... ain't gonna happen.

If this were true then the modern scanners wouldn't be capable of mm accuracy on a moving truck.

 

[Johnnypenso]

From what I can find, it looks like they scanned it themselves:





If it required to be stationary then we know for a fact they didn't do it.

You know how much time it would take to scan a 13 mile track 13 meters at a time?

Let's say they scanned every 13 meters, like you suggest, every 10 minutes.

If they scanned 8 hours a day, it would take 11.12 days in a row to scan the track....

do we honestly believe this?

At 15 mins per scan every 13 metres it would take over 400 hours. At 5 mins 130+ hours etc.

 

[TenEightyOne]

If they scanned 8 hours a day, it would take 11.12 days in a row to scan the track....

do we honestly believe this?

The maths seems right, your presumption about the number of scanners they employed might be off.

 

[FordGTGuy]

The maths seems right, your presumption about the number of scanners they employed might be off.

Hell I'm presuming any scanners at all because I don't believe it was scanned at all.

Normally, you scan with a single rig that makes a single model.

You don't stitch together models when using laser scanners.

 

[Samus]

Let's just say it never happened until someone can prove otherwise and move on. 👍

 

[TenEightyOne]

Normally, you scan with a single rig that makes a single model.

You don't stitch together models when using laser scanners.

How odd, I've done exactly that on many occasions.

Providing the central reference for your points clouds match up then you're good to go. Not all installations/locations are available in the entirety for the length of time it would take with a single scanner.

Let's just say it never happened until someone can prove otherwise and move on. 👍

Given that this is Turn10... quite. Greenwalt lost my trust several iterations ago.

 

[Huks]

Which definition of accuracy are we using here? The term accuracy in the context of LIDAR refers specifically to the resolution of the vertical axis (or height) while it seems like you guys are using it as a synonym for resolution in general.

 

[Samus]

How odd, I've done exactly that on many occasions.

Providing the central reference for your points clouds match up then you're good to go. Not all installations/locations are available in the entirety for the length of time it would take with a single scanner.



Given that this is Turn10... quite. Greenwalt lost my trust several iterations ago.

Turn 10? I was referring to the idea it was laser scanned for GT4 in 2004. It was obviously scanned by T10.

 

[TenEightyOne]

Turn 10? I was referring to the idea it was laser scanned for GT4 in 2004. It was obviously scanned by T10.

I'd probably read the references to Forza/Dan backwards in this thread. The point about multiple scans remains though, apologies

 

[FordGTGuy]

How odd, I've done exactly that on many occasions.

Providing the central reference for your points clouds match up then you're good to go. Not all installations/locations are available in the entirety for the length of time it would take with a single scanner.

Given that this is Turn10... quite. Greenwalt lost my trust several iterations ago.

Maybe for scanning interstates but I've always seen them talk about importing a single model into their software after scanning.

I'd probably read the references to Forza/Dan backwards in this thread. The point about multiple scans remains though, apologies

I never said it was impossible, I just said that it's not something I've seen done for scanning tracks but it's not something they seem to do nowadays when scanning tracks as they scan it with a single rig.

 

[TenEightyOne]

Maybe for scanning interstates but I've always seen them talk about importing a single model into their software after scanning.

Yes, you end up with a single model. Getting there in a timely fashion requires flexibility.

 

[FordGTGuy]

Yes, you end up with a single model. Getting there in a timely fashion requires flexibility.

That's what I'm referring to though, GPS stitches together because they collect data from multiple rigs for different parts of the track. Laser scanning nowadays is done on a single rig driving around the track that captures the entire track within scanning range(including buildings and trees), texture data and lighting at the same time.

 

[TenEightyOne]

GPS stitches together because they collect data from multiple rigs for different parts of the track. Laser scanning nowadays is done on a single rig driving around the track that captures the entire track within scanning range(including buildings and trees), texture data and lighting at the same time.

At the risk of labouring the point... multi-scans are centre-located by a fixed point, usually provided by GPS, you're talking about the same thing. Using a single scanner is the cheaper option but you need the facility you're scanning to be available for the entire scan time. Where that's limited (as in the Nordschleife example) using multiple scanners to populate your final model is far more efficient.

 

[FordGTGuy]

At the risk of labouring the point... multi-scans are centre-located by a fixed point, usually provided by GPS, you're talking about the same thing. Using a single scanner is the cheaper option but you need the facility you're scanning to be available for the entire scan time. Where that's limited (as in the Nordschleife example) using multiple scanners to populate your final model is far more efficient.

Turn 10 rented out the entire Nurburgring track for two days in order to laser scan it......

I don't see how it's more efficient to use two rigs and stitch the model together when you can just drive one rig and get it all at once.

 

[Ridox2JZGTE]

From what I can find, it looks like they scanned it themselves:





If it required to be stationary then we know for a fact they didn't do it.

You know how much time it would take to scan a 13 mile track 13 meters at a time?

Let's say they scanned every 13 meters, like you suggest, every 10 minutes.

If they scanned 8 hours a day, it would take 11.12 days in a row to scan the track....

do we honestly believe this?

Also you keep bringing up 50 meter accuracy, we are talking about the accuracy of the track surface and at most the track distance from the scanner is going to be 6-7 meters at max when scanned.

15 mm, even with reflective surfaces... ain't gonna happen.

If this were true then the modern scanners wouldn't be capable of mm accuracy on a moving truck.

Did I suggest they scanned every 13m ? I only guessed between 50-70m scan distance for 15mm accuracy. You need to take into account the low data point cloud, 500k point over radius of 50-70m would not give detailed data points which are spread. Lower density, less accurate.

As I said, I only could speculate, but the accuracy is not only device dependent which I said many times that the accuracy claimed is just theory max that can be achieved, the operator also decide how accurate the data being collected by balancing between project requirements, scan time/distance/cost ( number of scanners ), and the condition during the scanning that also influence the end result ( bright sun or too dark is not good )

Static and mobile scanning also usually use GPS to get precise positioning of scanner after every few scans or each scans, but cloud to cloud point or use object target as reference also possible.

A 10+km road scanning could take 2 months with mobile scanning and 1500+ scans ( this is commercial level use ), using today's tech. I don't know how Nurb can be scanned in a few days , maybe with less accuracy/vehicle going faster while scanning or multiple mobile scanners.

This is what I read from here : http://www.laserscanningforum.com/forum/viewtopic.php?f=176&t=10429

The point density depends on the speed of the vehicle. (lower vehicle speed, higher density). The point density do fall off as it get further away from the mobile scanner. Typically, we keep it under 30 meter for accuracy and density reason.

Mobile scanner can get you 5 cm (95%) in z and about 2-3 cm (95%) without any ground control (+/-30m of the vehicle path). Error goes up, as the points are further away from the mobile scanner. You will likely get better vertical accuracy (cut by 1/2) with a single target / vertical adjustment.

With ground control (depending on the spacing), you can get 1.5 cm (95%) in z and about ~1 cm (95%) using a 150m to 200 m ground target spacing assuming you have good GNSS converge, GNSS base station (<15Km baseline), and good GNSS/IMU initialization.

If you need higher accuracy, use static scanning.

The current tech for road project service offered by Landair can do 1 inch accuracy over 100m scan distance. 90's and early 2000, static scanners in early stage of tech, I imagined would be a lot different than what's capable today with mobile scanning lasers. Ironically, with today techs, static scanning is more accurate than mobile scanning ( the slower the more accurate ) - mobile scanners seems to be used by game devs.

I just want to give more perspective of laser scanning difference of old days ( early stage of tech ) compared to today. I do not claim PD did it, maybe yes, maybe not, no evidence to suggest anything.

 

[FordGTGuy]

You know, you guys keep moving this along but you haven't been able to come up with any evidence at all of GT4 having a laser scanned Nurburgring.

GPS was capable of 15mm accuracy at the time of production of GT4 and we know they used GPS to scan tracks for GT4.

As of now I'm no longer going to keep pushing along this circular argument for no reason.

 

[Luminis]

Let's just say it never happened until someone can prove otherwise and move on. 👍

This isn't how this works. GT was the first to do everything and has achieved everything until proven otherwise.

In dubio pro Gran Turismo. Words to live by.

 

[TokoTurismo]

Heard the videos were taken down at the request of PD/Sony......FAIL!!

 

[Luminis]

Heard the videos were taken down at the request of PD/Sony......FAIL!!

Reaction wasn't positive enough, I take it?

 

[TokoTurismo]

Reaction wasn't positive enough, I take it?

Guessing so but that shouldn't be the one reason why they should take the videos down. Most developers take similar heats like that as well so why can't PD?...

 

[Spitfire77]

Guessing so but that shouldn't be the one reason why they should take the videos down. Most developers take similar heats like that as well so why can't PD?...

Which videos were taken down?

 

[mister dog]

Which videos were taken down?

Check the news page, basically all high quality direct feed videos.

 

[AudiMan2011]

Which videos were taken down?

The recent direct feed gameplay videos like the 458 GT3 on Nurb, ND Miata on Brands and the Bug VGT on Tokyo Expressway.

 

[TokoTurismo]

It also requires you to log-in on Youtude in order to watch the videos as well, which IMO is pretty stupid on PD/Sony's part. They're trying, they're trying way too hard now...

 

[Luminis]

Most developers take similar heats like that as well so why can't PD?

Beats me.

I don't even think the reaction was that bad - this thread is obviously filled with negativity and I haven't had a look at the YouTube comment section (I like my sanity), so I can't really judge just how much heat they've been taking. Eh, just another decision to add to the list of decisions I can't understand.

 

[FordGTGuy]

More likely that they want more control over the way media is released for the game. Did these videos originally have expressed permission from Polyphony?

 

[HuskyGT]

Game looks good. But that's about it; the looks. Other than that, GTS feels underwhelming, pointless, another waste of time focused on "innovation" which is pretty much just a brand publicity stunt or business partnership with external entities that do nothing to strengthen to the game's core. Personally I don't care about FIA nor "real" racing. I care about the cars, which is what GT is supposed to be. GT1, GT2, GT3 and GT4... that's the idea that needs to be brought back. Fortunately, Forza is here to do that for now.

Typical PD bringing bells and whistles to yet another bland piece of software full of empty promises and lack of improvement. Almost six more months for a game that's probably not even a fraction of what Forza or Pcars is. Or, even Assetto Corsa (physics).

And if you expect a real "GT7," that is not going to happen in another 4 to 5 years. PD just can't deliver. They spend way too much time with their heads up their rears. It took them about that time to bring up a game with recycled assets and flaws unforgivable for even the past gen of consoles. what makes you thing they will develop two games in a decent time period? GTS will be the only GT we will see in years to come. And it will remain pretty much untouched (no DLC, no expansions)for what remains of its life.