Map maker having fun?

[najaB]

I was just looking a the 'Trains in Street View' thread and came across this post from last year.

They definitely look more like 390 driving cars than 221's, so is this the map makers having fun by lining up two images just right, or is this genuine? If it is genuine, where's the loco that should be pulling them wherever it is they are going?

 

[simonmpoulton]

I would hazard a guess that the image consists of a few pictures taken at slightly different times resulting in only the front and rear carriage being visible.

 

[najaB]

I would hazard a guess that the image consists of a few pictures taken at slightly different times resulting in only the front and rear carriage being visible.

That was my first suspicion, but the alignment is too perfect for it to be just by chance.

 

[sprinterguy]

We've had a thread specifically dedicated to this sort of phenomena. There are other examples around Britian, such as a 2+11 First Great Western HST somewhere. It is a result of how consecutive overhead images have been stitched together, as simonmpoulton says. It certainly isn't a genuine occurrence.

 

[DaleCooper]

That was my first suspicion, but the alignment is too perfect for it to be just by chance.

Does the length difference tell us anything?

 

[najaB]

Does the length difference tell us anything?

I'll put it down to the map maker having fun as the alignment is too perfect to be chance - right down to the shadow lining up perfectly. When the computer is left to do its own thing, the result is more like this.

 

[DaleCooper]

I'll put it down to the map maker having fun as the alignment is too perfect to be chance - right down to the shadow lining up perfectly. When the computer is left to do its own thing, the result is more like this.

I find it difficult to believe that someone at Google has the job of looking at every image and it's unlikely that they spotted these by chance among the many millions of images captured.

I think it's just fortuitous.

 

[najaB]

I find it difficult to believe that someone at Google has the job of looking at every image and it's unlikely that they spotted these by chance among the many millions of images captured.

I think it's just fortuitous.

Having worked with Maps and Street View for ages, I find it harder to believe that the automated image stitching algorithm worked that well!

 

[greatkingrat]

I'll put it down to the map maker having fun as the alignment is too perfect to be chance - right down to the shadow lining up perfectly. When the computer is left to do its own thing, the result is more like this.

Don't forget that there are thousands of trains pictured in Google Maps. 99% won't line up exactly, but it is not surprising that some do.

 

[RailAleFan]

I'll put it down to the map maker having fun as the alignment is too perfect to be chance - right down to the shadow lining up perfectly. When the computer is left to do its own thing, the result is more like this.

I wouldn't be too surprised about the almost perfect shadow alignment - there will be seconds if not just fractions of between the shots, and I would imagine that aerial photography for this purpose is made along E-W / N-S tracks and this stretch of track is very close to E-W aligned.

I presume the train is travelling W > E so if the image capturing plane was travelling E-W I don't think it's unlikely for this kind of "phenomenon" to occur in the resulting composite imagery.

 

[DaleCooper]

Having worked with Maps and Street View for ages, I find it harder to believe that the automated image stitching algorithm worked that well!

The algorithm seems pretty good at lining up roads, rivers and railways so why shouldn't it cope with an essentially linear object as long as the inital images are suitable.

 

[najaB]

...I would imagine that aerial photography for this purpose is made along E-W / N-S tracks and this stretch of track is very close to E-W aligned..

Except this is imagery from a polar orbiting sun-synchronous satellite, the aerial images look more like this.

I've no reason to doubt that it is two images been combined, but I'm confident that it's been tweaked manually - why would an algorithm join the two images exactly on a vestibule?

 

[DaleCooper]

Except this is imagery from a polar orbiting sun-synchronous satellite, the aerial images look more like this.

I've no reason to doubt that it is two images been combined, but I'm confident that it's been tweaked manually - why would an algorithm join the two images exactly on a vestibule?

Who knows where the join is? From the length difference I'd say it's somewhere within one of the coaches.

 

[najaB]

Who knows where the join is? From the length difference I'd say it's somewhere within one of the coaches.

I see where you're coming from. Looking at the sleepers on the adjacent track shows a possible join point - what I'd taken as the shadow of the OLHE is actually probably the displacement of the image join.

 

[RailAleFan]

Except this is imagery from a polar orbiting sun-synchronous satellite, the aerial images look more like this.

Are you sure about that? The plates credit Bluesky who are one of the aerial photography sources for "Satellite view" - and they will be making use of various levels of resolution of aerial photography as higher resolution and oblique angle capturing such as that in your link becomes available.

 

[asylumxl]

The images will be aerial survey photographs taken from a plane flying at a specific altitude, then automatically stitched together using feature matching. Moving objects such as cars/trains will be ignored as they will be one of hundreds of features identified by the software.

The most likely explanation is that the plane was moving at a similar or slightly faster/slower speed and by chance the join happens to look as if it was intentionally joined.

 

[andyb2706]

Not read all the posts so this might already have been said. Seeing there is no locomotives knocking around it can't be genuine. It is definitely a Pendo but where is it drawing its power from? the driving vehicles don't have the pantographs on them.

 

[MCR247]

Not read all the posts so this might already have been said. Seeing there is no locomotives knocking around it can't be genuine. It is definitely a Pendo but where is it drawing its power from? the driving vehicles don't have the pantographs on them.

Maybe you should have read the posts then :roll:

 

[edwin_m]

The images will be aerial survey photographs taken from a plane flying at a specific altitude, then automatically stitched together using feature matching. Moving objects such as cars/trains will be ignored as they will be one of hundreds of features identified by the software.

The most likely explanation is that the plane was moving at a similar or slightly faster/slower speed and by chance the join happens to look as if it was intentionally joined.

Or even that the plane was flying back and forth taking photos of a strip of the land each time, and when it came back to do the next strip there just happened to be another train in nearly the same place.

 

[DaleCooper]

Or even that the plane was flying back and forth taking photos of a strip of the land each time, and when it came back to do the next strip there just happened to be another train in nearly the same place.

That's the most bizarre explanation so far.

 

[najaB]

That's the most bizarre explanation so far.

Especially given that there's at least an hour between Pendolinos - that would be an amazing coincidence.

 

[edwin_m]

That's the most bizarre explanation so far.

Why? How else would they assemble a large area photo other than by stitching together individual strips?

There are two Pendolinos in some hours as they run some of the Birminghams too.

The angle of shadow on the OLE poles doesn't change much so the pictures must have been taken at the same time.

 

[ac6000cw]

If the train and plane are travelling in opposite directions then it would be very easy to get the effect in the picture (when the images are stitched together) e.g a relative closing speed of about 200 mph is approx 90m per second, so if the photos were taken 2 seconds apart that is about 7 to 8 carriage lengths that would 'disappear' at the join.

What is the typical flying speed of a light plane ?

 

[DaleCooper]

Why? How else would they assemble a large area photo other than by stitching together individual strips?

There are two Pendolinos in some hours as they run some of the Birminghams too.

The angle of shadow on the OLE poles doesn't change much so the pictures must have been taken at the same time.

I doubt that the plane flies back and forth photographing the same area again and again surely that would be a waste of resources..

Of course photos are stitched together, that has been taken for granted all through the thread.

As the shadows are the same then clearly it hasn't captured two different trains.

 

[MK Tom]

One odd effect you get in Google Earth is the bending effect seen here: https://www.google.co.uk/maps/@52.2289762,-0.9123794,181m/data=!3m1!1e3

No idea why that happens. It happens again at the road bridge a little further north.

Note how the Northampton line is shut so often it's even shut in the GE pictures.

 

[blackfive460]

I doubt that the plane flies back and forth photographing the same area again and again surely that would be a waste of resources..

Usually, in vertical aerial photography that's exactly what happens.

The aircraft flies a fixed, straight and level course then returns in the opposite direction covering the next adjacent strip, usually with a 60% overlap both in each individual photographs and in the strips covered.

The large overlap is essential for two reasons; firstly, the photographs are often required as stereo pairs but, when turned into a flat map the individual photographs need to be 'rectified' to remove distortions due to variations in distance from the camera; the edges of the photographs are further away from the camera and land height differences cause a change of scale. The more overlap you have, the easier it becomes to join your photographs but, over the large areas that Google have at large scale there are still going to be minor distortions.

My guess with this one is that the plane was indeed flying almost parallel with the line and the front and rear of it just happened to be on adjacent photos nicely positioned to sit just where whatever software the mappers use decided to make a join.
Unlikely perhaps, but that sort of coincidence is bound to happen somewhere if you take enough large scale images.

 

[snowball]

Not rail related, but Google reveals this example of subsidence on the M4 in South Wales which has been kept out of the news media:

http://goo.gl/maps/Rvpir