In-game VS Satellite

[Shack95 2,658]

11 hours ago, in2tech said:

Is your job a Map Maker or Geologist?

Far from it😅! But yes, I’ve always had a thing for maps, aerial views, geography and the like. There’s just so much beauty on our planet.

11 hours ago, in2tech said:

ow do you find the locations that are not in the file name when you download them?

In some rare cases the location is in the file name (e.g. Istanbul). Sometimes there are other hints in the file name, e.g. this week’s Italy screenshot was called “ciao”, so I knew it was probably in Italy. Guessing from the vegetation and flat terrain it had to be northern Italy. Then there’s this straight canal next to the meandering river. You can see such things in Earth View without having to zoom in closely. This narrowed it down considerably and I just had to check out 3-4 places.

Sometimes it’s also pure luck, as it was the case with the China screenshot. Those terraces indicated Asia. I picked a hilly area in China at random and to my surprise it looked exactly like in the screenshot. 5 minutes later I found the spot.

And of course sometimes the location has already been found out and posted somewhere, e.g. Edwards AFB this week.

What makes it relatively easy is that in Earth View you have the exact same imagery, so when you “fly” across the spot you’re looking for you usually notice it at once.

[Shack95 2,658]

12 hours ago, Tuskin38 said:

Is that another water way to the right of the river? The Sim doesn't seem to be rendering it as such.

Yes, it's a canal.

[Shack95 2,658]

23 minutes ago, tweekz said:

Even if some of the houses look a bit "blocky" (need a little more detail, even if it's just random), with all the reconstruction of trees and bushes, fields, roof colors this will be the first sim that is truly useful for VFR training.

I agree. Comparing it to the photo it actually looks rather good for a non-photogrammetry place that hasn't been edited manually. Well, apart from that grass strip that should really be a canal maybe...

[GlideBy 324]

6 minutes ago, Shack95 said:

Yes, it's a canal.

If you zoom in you can see some distortions and subtle water effects, I think.  So it seems to know it is water.  But I don't see any reflections in the water.  The other water has some tree reflections even with the waves.

I have to say, the first couple times I looked, it didn't register to me as a canal filled with water.

[Shack95 2,658]

2 hours ago, domkle said:

Nice : I drive to the airport several times a year. From the West, the mountains are strikingly beautiful. They are unfortunately hidden in the pic and seem not so good in the Bing pic.

It‘s a lovely speck of land indeed! I really hope they get this right. As I‘ve said before, mesh quality is one of my biggest concerns regarding scenery (though that doesn't seem to be too bad in this particular pic). I know it‘s still in alpha and things could change but so far there doesn‘t seem to be much difference between Bing and MSFS in that regard. The only exception is Sugarloaf Mountain but since that‘s such an iconic rock it might as well be handcrafted. 

Edited March 21, 2020 by Shack95

[Dominique_K 10,138]

40 minutes ago, Shack95 said:

 As I‘ve said before, mesh quality is one of my biggest concerns regarding scenery (though that doesn't seem to be too bad in this particular pic) 

Me too

[in2tech 588]

@Shack95 First off I can't wait to fly this sim 🙂 In a few of the Alpha tester's screen shots the whole image almost has a Green tint to them? Is it the Azure, Bing data getting computed wrong, a Alpha testers screen settings, etc... I was gonna post the images, but you do a much better job and they are already posted.

Cavarzere, Italy this page and others

Bucharest, Romania page 27 this Thread, I think!

I am not complaining, just asking a question and getting in on the conversation here! It's still amazing to me you find all these locations, as well as other members here! Great work everyone!

And where did you get the REAL image of the Canal on page 33 ( in case this ends up on another page ), Bing, Google Earth? Off your own computer as I have a feeling you have Terabytes of Earth photo's on storage? It's a joke or course, or is it 🙂

[GlideBy 324]

6 hours ago, in2tech said:

Cavarzere, Italy this page and others

Bucharest, Romania page 27 this Thread, I think!

Italy the Bing image looks greener than the MSFS image, at least to me.  But you are 100% right the Bucharest image is green.  In fact so green that the roads are green.  Maybe the AI misidentified the roads as canals with green water, or grass.  So my guess would be an AI goof.  I would think they would want to cross reference the road data used for routing to the image data. 

There are other mistakes in the Bucharest image.  Like the buildings are not symmetrical down the middle road, the roof colors don't match the satellite image, and there is a big building in the middle of the parking lot.  But it is Alpha, so we expect issues. 

[Shack95 2,658]

11 hours ago, in2tech said:

In a few of the Alpha tester's screen shots the whole image almost has a Green tint to them? Is it the Azure, Bing data getting computed wrong, a Alpha testers screen settings, etc...

I was wondering about that, too. In Cavarzere the source image already has a green tint so I guess their AI (or whatever) for some reason couldn’t correct the colours properly. There are quite a few areas in Bing Maps that have this green tint.

What I find interesting about the Bucharest shot is that here it’s the other way round. Instead of toning down the green tint it was increased. I don’t know enough about the technical side of it to say why.

11 hours ago, in2tech said:

And where did you get the REAL image of the Canal on page 33

I googled it. It’s from https://www.alpenpaesse-wasserscheiden.at/italien/das-alpine-einzugsgebiet-der-adria-excl-po-gebiet-innerer-alpenbogen/brenta-gebiet-flussverlauf/

Edited March 23, 2020 by Shack95

[Claviateur 789]

On 3/21/2020 at 4:50 AM, Shack95 said:

I agree. Comparing it to the photo it actually looks rather good for a non-photogrammetry place that hasn't been edited manually. Well, apart from that grass strip that should really be a canal maybe...

In the language of actual scenery development, the canal is missing what we call a "water mask" I suppose. It's a sort of Alpha channel in a land texture that reveals water in the terrain layer underneath.

If Asobo are using the OSM data (as it seems the case in few other locations), then as far as I could see, this area in OSM has the canal vector drawn. 

However, it could be that for "Water masks" they are not using the OSM data.

Or their engine uses techniques different from the actual ones to display inland water surfaces.

Edited March 23, 2020 by Claviateur

[Dominique_K 10,138]

The Fiume/Canale Gorzone is in P3D+Vector.

The scattering of trees and the bland autogen  need some work.

[Shack95 2,658]

35 minutes ago, Claviateur said:

If Asobo are using the OSM data (as it seems the case in few other locations), then as far as I could see, this area in OSM has the canal vector drawn.

Also the placement of autgen buildings doesn't seem to be based on OSM here. I wonder why there are still some inaccuracies. Maybe their footprint data is older or newer than the imagery. Or they’re using yet another source.

[Claviateur 789]

2 hours ago, Shack95 said:

Also the placement of autgen buildings doesn't seem to be based on OSM here. I wonder why there are still some inaccuracies. Maybe their footprint data is older or newer than the imagery. Or they’re using yet another source.

Do you mean by inaccuracy the types of autogen rendered at a specific spot or the placement?
 
But I can only guess based on my curiosity in this field and I suppose only, based on my observations that they could have, perhaps, for now at least, a mixed bag of data they work with. They might want to process for the Alpha and render the procedural world so it can be populated and available for testing.

Then let's not forget that the Alpha screenshots come from users specific settings. I do not know if all users set their simulator to have the maximum graphic options or capability. This could explain the inconsistent coverage in a specific area (i.e: some infrastructure are not rendered could be the result of a lower autogen setting on a specific User's setup). 

Edited March 23, 2020 by Claviateur

[Shack95 2,658]

1 hour ago, Claviateur said:

Do you mean by inaccuracy the types of autogen rendered at a specific spot or the placement?

I meant the placement. But yes, as you say, there can be many reasons for this.

[Claviateur 789]

4 minutes ago, Shack95 said:

I meant the placement. But yes, as you say, there can be many reasons for this.

Yes for sure, the mixed bag of data and the margin of errors in the automatically generated footprints + the purely procedural ones to fill the gaps (Japan style). And at the end, the type of infrastructure placed there or extruded from the footprint using the tags or with random or procedural values.

I believe it's the Beta that tells us for sure, what is the final data we will have in the R1 of MSFS.