Sign in to follow this  
HumptyDumpty

India Mesh.

Recommended Posts

Hi,

 

 I am making a mesh for India but mostly would be the state of Maharashtra as this is my home state.  May extend to the South. Won't be doing the Northern region because that is already covered by Alpilotx's Himalaya Mesh.

 

 I am using .hgt files from viewfinderpanorama which to my knowledge is a 30 DEM tile which should be good enough, though there is government data but the .tiff from there has a lot of spikes and I don't have any idea how to rectify it. 

 

 I have made four tiles :

 

 +19+072  - Bombay - this tile has some issues hence I may remove it  or look for other data.

 

 +18+072  - South Bombay  - covers Butchers Island and the Elephanta Caves (still needs some editing in openstreetmap) but you won't see the caves :)

 

 +18+073 -  SE of Bombay heading towards Pune and Aamby Valley. Aamby Valley OSM data is almost NIL atm. But many are editing so in time it should be   populated.  

 

 +17+073    This is the area of the western ghats and the Sahyadri Mountain Range.  The area looks pretty good. But I am not able to locate the Mountain called as   Table Land in Panchgani , which is a flat mountain and almost hollow. So may need some editing again here in openstreetmap .

 

 To me the changes are noticeable.

 

  The objects like buildings and other stuff is derived from openstreetmap and the coverage is very good for the island of Bombay heading East .  For this I am using World2Xplane to generate them.  There are quite a few National Parks in and around Bombay. The one in Bombay is the Sanjay Gandhi National Park which has two lakes with crocs.  But these two lakes are not showing up in the +19+072 tile atm.

 

     I shall also be uploading the .pbf file so if users want to fly in this area can generate the objects using W2XP.   

 

   The airports and the helipads are in the thread "Juhu Aerodrome" which has VAJJ Juhu Airport and three Helipads (Actually there are four around Bombay but i haven't yet made the fourth one as yet which is the Naval Radar Station ) which shall be uploaded with this package once I am satisfied. 

 

This is my first effort and to be honest Xplane scenery development is quite easy even for a layman a little searching and advice from Tony has helped. At least now I can fly around in a replica of Bombay which was never like this in any sim i have used.   Xplane needs a a few more objects to make it look better and hope the devs who make them do go further.

 

  I am also using custom water textures and scripts which to my sight looks very nice. I have not made these textures and scripts , got them from .org. I am not a developer or a graphics guy it's beyond me.

 

    I hope you rascal pilots like it.   

  • Upvote 2

Share this post


Link to post
Share on other sites
Help AVSIM continue to serve you!
Please donate today!

+1, great to see more people getting into large scale scenery development

Share this post


Link to post
Share on other sites

+1, great to see more people getting into large scale scenery development

 

That's how Xplane progresses :) 

This is Aamby Valley airport VAAV (not there in Xplane)

 

Default Mesh

 

Don't get worried looking at the fps counter , that is due to the clouds.

 

DefaultMesh.png

The mesh I made same airport

 

HDMESH.png

Share this post


Link to post
Share on other sites

Hello humpty,

 

Wow thats realy cool. Would be great to use some tiles that people created. Made some tiles covering the Netherlands myself and also the south of florida which is absolutely flat... Where would you like to upload your tiles?

Have a nice day!

Jozeff

Share this post


Link to post
Share on other sites

Hello humpty,

 

Wow thats realy cool. Would be great to use some tiles that people created. Made some tiles covering the Netherlands myself and also the south of florida which is absolutely flat... Where would you like to upload your tiles?

Have a nice day!

Jozeff

 

Hi Jozeff,

 

 I will be uploading it to dropbox but not this week as I still need to figure out that one tile which is not showing the water like the lakes.

 

 I thought Florida was covered by the HD Mesh.

Share this post


Link to post
Share on other sites

I will use the Cartosat 1-R2 data and check if the terrain comes out right, there is also data called as "Land and Terrain" on the government site and don't know what it contains.

Share this post


Link to post
Share on other sites

Well the DEM for India are a mess, certain .hgt have missing points so it does not create and the .tiff from the government site is also a mess.

 

I am mixing working .hgt and working .tiff and will see how it shows up.

Share this post


Link to post
Share on other sites

Damn 3:30 a.m. and tomorrow work but can't help it , my mesh is going good and again thanks to Tony.

 

Just checking out the Sahyadri mountain range Yes to me it looks different now.

Share this post


Link to post
Share on other sites

What's the resolution of the mesh, you can make it more detailed using the mesh_specs command?

Share this post


Link to post
Share on other sites

What's the resolution of the mesh, you can make it more detailed using the mesh_specs command?

 

I have no idea about the resolution, The data from the government site which contains the shape and the tif file , well the tiff file has very bad problem especially near the coast or where the lat long meet. So I have to use .hgt from viewfinder and to my knowledge that is 30 m.

 

Is mesh_specs a new tool ?

 

going for lunch now be back in 30.

Share this post


Link to post
Share on other sites

It's an option in mesh tool to increase the density of the mesh

Share this post


Link to post
Share on other sites

Ok I just saw it @.org as I am using that same command, I will try the mesh_specs but I what I would not know are the numbers that I need to assign.


I am trying this now

 

mesh_specs 200000 4 in the text file.

Share this post


Link to post
Share on other sites

India's Mesh is a Mess :-) ( said Zulfi.... not me! )

Share this post


Link to post
Share on other sites

India's Mesh is a Mess :-) ( said Zulfi.... not me! )

 

India is sorting itself now :)

Share this post


Link to post
Share on other sites

Important correction: Viewfinderpanoramas.org does mostly only provide 90m (3sec) mesh for the entire planet. Only a few areas (which I already all covered with UHD Mesh Scenery v1) have 30m (1sec) elevation data ... You can see those few on this download map:

Instead you might try to use 30m (1sec) data from NASA SRTM (which has been released in the mean time) ... but even there, you can expect some "holes" and other mess etc. .

 

Let me give you a short explanation on DEM problems:

 

Elevation data is a bit more complicated science than many imagine. Especially with higher res data, you can experience a lot of strange data "errors", like for example spikes in cities (this was a big problem for the UHD Mesh coverage in Japan for example). Those spikes - especially in cities - are a result of the way this data is collected. Most of the time - especially with all large scale projects like SRTM etc. - they rely on some kind of aerial (most of the time satellite) based measurements (usually radar or laser / lidar). And those measurements do reflect from the surface(!!) ... what ever the surface is at a given point! In cities, that surface is mostly defined by human infrastructure like buildings, roads etc. With high skyscraper or big highways you can easily get some "nice" bumps in those surfaces. Thus, in big cities, you can get "big" spikes (with easily having 100m offset to the ground) due to many high rising buildings. And those spikes can ruin any mesh scenery.

 

But this is usually the nature of most freely available elevation data. If you want real terrain model, where the ground elevation is in the DEM, you - usually - very quickly enter an area of very expensive products (simply for the reason, that its far from trivial and usually very labor / cost intensive to create them). You only alternative might be to do some kind of smoothing on yourself (and realize the complexity of this topic) ... In Japan for example I used the landclass data to determine where the cities are, and only there did I apply some smoothing (which gave a usable result ... but for sure isn't a scientifically perfect solution).

 

Add to this all the holes in the landscape. They are usually a result of measurement errors. For example having bad weather in the way or more often (and very commonly in mountainous regions) depending on the angle of the radar beams hitting ground, some parts of the landscape are shaded by others (the radar not always sees the deep valleys etc.) ... And without correct data, the data processing (because even the "raw" data you get on the net is not really raw ... it is already a complex interpretation of "really raw" radar / lidar measurements) either introduces nasty errors (another reason for spikes) or rather decides to stay with "no data" (resulting in holes). Over the time, some agencies or project groups introduce their own hole filling (if you ar lucky) ... or you can try to do it - again - yourself by back filling the holes from other source. Again, for japan - as an example - I had to back fill the fantastic (but full of holes) Japanese AW3D 1sec elevation data with a bit less detailed SRTM 1sec data AND even had to back fill SRTM 1sec data (because it sill has holes too) with SRTM 3sec data.

 

By the way, Wikipedia has a nice page about elevation data (and the implications I noted above):

So you see ... even though we nowadays are very very lucky to have access to fantastic (and a few years ago unimaginably detailed) data on the internet, we should never expect (even if we wish :) ) it to be perfect and often need to find ways around those problems (or accept them to some extent if the work needed would not be justifiable). And yes, I really mean this to be true for ALL types of data (as ana example ... whoever worked more deeply with Openstreetmap data will immediately understand what I mean)!

Share this post


Link to post
Share on other sites

Important correction: Viewfinderpanoramas.org does mostly only provide 90m (3sec) mesh for the entire planet. Only a few areas (which I already all covered with UHD Mesh Scenery v1) have 30m (1sec) elevation data ... You can see those few on this download map:

Instead you might try to use 30m (1sec) data from NASA SRTM (which has been released in the mean time) ... but even there, you can expect some "holes" and other mess etc. .

 

Let me give you a short explanation on DEM problems:

 

Elevation data is a bit more complicated science than many imagine. Especially with higher res data, you can experience a lot of strange data "errors", like for example spikes in cities (this was a big problem for the UHD Mesh coverage in Japan for example). Those spikes - especially in cities - are a result of the way this data is collected. Most of the time - especially with all large scale projects like SRTM etc. - they rely on some kind of aerial (most of the time satellite) based measurements (usually radar or laser / lidar). And those measurements do reflect from the surface(!!) ... what ever the surface is at a given point! In cities, that surface is mostly defined by human infrastructure like buildings, roads etc. With high skyscraper or big highways you can easily get some "nice" bumps in those surfaces. Thus, in big cities, you can get "big" spikes (with easily having 100m offset to the ground) due to many high rising buildings. And those spikes can ruin any mesh scenery.

 

But this is usually the nature of most freely available elevation data. If you want real terrain model, where the ground elevation is in the DEM, you - usually - very quickly enter an area of very expensive products (simply for the reason, that its far from trivial and usually very labor / cost intensive to create them). You only alternative might be to do some kind of smoothing on yourself (and realize the complexity of this topic) ... In Japan for example I used the landclass data to determine where the cities are, and only there did I apply some smoothing (which gave a usable result ... but for sure isn't a scientifically perfect solution).

 

Add to this all the holes in the landscape. They are usually a result of measurement errors. For example having bad weather in the way or more often (and very commonly in mountainous regions) depending on the angle of the radar beams hitting ground, some parts of the landscape are shaded by others (the radar not always sees the deep valleys etc.) ... And without correct data, the data processing (because even the "raw" data you get on the net is not really raw ... it is already a complex interpretation of "really raw" radar / lidar measurements) either introduces nasty errors (another reason for spikes) or rather decides to stay with "no data" (resulting in holes). Over the time, some agencies or project groups introduce their own hole filling (if you ar lucky) ... or you can try to do it - again - yourself by back filling the holes from other source. Again, for japan - as an example - I had to back fill the fantastic (but full of holes) Japanese AW3D 1sec elevation data with a bit less detailed SRTM 1sec data AND even had to back fill SRTM 1sec data (because it sill has holes too) with SRTM 3sec data.

 

By the way, Wikipedia has a nice page about elevation data (and the implications I noted above):

So you see ... even though we nowadays are very very lucky to have access to fantastic (and a few years ago unimaginably detailed) data on the internet, we should never expect (even if we wish :) ) it to be perfect and often need to find ways around those problems (or accept them to some extent if the work needed would not be justifiable). And yes, I really mean this to be true for ALL types of data (as ana example ... whoever worked more deeply with Openstreetmap data will immediately understand what I mean)!

 

 

Hi Andras,

 

 Thanks for this fantastic explanation and about the different SRTM data.  I will definitely try it out.

 

Yes I have those spikes and very bad heights in the .tif from the government website usually on the coast. It's a big learning process and want to make it as close as possible.  

Share this post


Link to post
Share on other sites

I followed Andras advice and checked out NASA SRTM  1 arc-second from here http://earthexplorer.usgs.gov/ the same area tiles I used earlier and at least till now they are good. The main is the +17+073 lat long as that falls under the Sahyadri Mountains so let's see what I get. 

 

If they are good I may extend the mesh more to the south.

Share this post


Link to post
Share on other sites

The tiles 

 

+19+072

 

+19+073 

 

 have issues because I am not getting the water areas proper hence won't be touching at the moment.  

 

These are the tiles that I have done :

 

 

+18+072  -  South Mumbai

 

+18+073  Lonavala + Pune

 

+17+073  Area of western ghats 

 

+09+076   Cochin - Kerala

 

+10+076   Cochin - Kerala

 

+10+077  Western Ghats Kodai Kanal Kerala

 

 All are NASA SRTM and ASTER GLOBAL DEM 1 arc-second which to my knowledge would be 30 DEM.

 

 Still doing some flights to check the data.  I would need a lab rat now.

  • Upvote 1

Share this post


Link to post
Share on other sites

I have covered the western coast of India heading south so the tile starts from :

 

 

+18 and ends at +08. These are only 1 degree tiles. I will go east once I am sure the mesh is OK.  

Share this post


Link to post
Share on other sites

Made a big blunder, some of the ASTER DEM's are from 2011 and the 2014 NASA SRTM few of the tiles have holes and have no idea how to fix them, So I will not touch for now , will await better data.

 

Only if I could get the forests with the mesh areas I have done would be nice , but again I am a layman here

Share this post


Link to post
Share on other sites

I don't know the if mesh will be done because lot many SRTM have holes and I have no idea how to fix those.

Share this post


Link to post
Share on other sites

Well, usually with problems like this, you end up with the need for some kind of GIS tools to aid you.

The - I think - best solution with SRTM 1sec holes is, to back fill them with lower resolution - but still good enough - SRTM 3sec data.

 

In the following blog post, they describe an approach which uses the open source (command line) GDAL toolset to achieve this:

Even though I rely on GDAL a lot too (for a lot of different, other tasks) I never tried the above solution (I used some freaky, Laminar internal tools for it ...).

Share this post


Link to post
Share on other sites

Well, usually with problems like this, you end up with the need for some kind of GIS tools to aid you.

The - I think - best solution with SRTM 1sec holes is, to back fill them with lower resolution - but still good enough - SRTM 3sec data.

 

In the following blog post, they describe an approach which uses the open source (command line) GDAL toolset to achieve this:

Even though I rely on GDAL a lot too (for a lot of different, other tasks) I never tried the above solution (I used some freaky, Laminar internal tools for it ...).

 

Thanks Andras , appreciate it. will have a look and hopefully it hits my gray matter.

 

Will try it tonight.

Share this post


Link to post
Share on other sites

I am trying what Andras has suggested but its complicating. But will keep trying till I get it.

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
Sign in to follow this