FreeMesh X Global (May 29)

[YoYo]

Check this: http://forum.avsim.net/topic/467933-...-announcement/

Im user of 5th m of payware FS Global Ultimate series from FS PILOT but it looks interesting if You use default FSX mesh (there are here ; ) ?) and its free. 42 Gb of data, distribution by Torrent.

[roger-wilco-66]

Wow, sounds good.
I'll compare it to the meshes I have created, if Freemesh is as good I'm out of the mesh creating business!


Cheers,
Mark

[YoYo]

Wow, sounds good.
I'll compare it to the meshes I have created, if Freemesh is as good I'm out of the mesh creating business!


Cheers,
Mark

Here is 38m so is better than default FSX/P3D mesh, but not very accurate like 5m for example, business will not die .

[arrowmaker]

Just been reading the avsim thread. Looks really interesting. Essentially this freeware project offers slightly better mesh (LOD 10 instead of LOD9) than FSGlobal, with certain exceptions. These exceptions include places like the Alps where FSGlobal has a higher LOD than it provides for the rest of the planet. However it looks like the FreeMesh team are going to release higher resolution mesh's for these areas in future updates.

All in all it looks like this will become a pretty major contribution to the FSX freeware scene. I know it's early days, but I can imagine this having a pretty severe impact on payware vendors like Pilots. Personally speaking, the only reason I would consider upgrading to another paywhere mesh, from the FSGlobal 2010 I currently use, is if it provides significantly better compatibility with Orbx products.

[Penz]

Super nice global mesh in every way in my opinion.

I also think the release has been well thought through, with one big global release instead of trying to do it in bits as has been attempted previously on more than one occasion, it's been done in a way that modern day "FS consumers" (the locust swarm!) can understand.

Like already pointed out, the team is already moving to provide further improved resolution of specific areas of interest.

I hope the project will be hugely succesful, it shows what FS community is really about (and where it came from.).

- JP (seeding ...).

[Dumonceau]

Does anyone have experience with the combination of FTX global, vector and Open LC?

Cheers,

Johan

[hairyspin]

I can't compare FTX or any other since my tightwad tendencies - and modelling preoccupations - have left me on FSX standard scenery for a long time. Now, the autogen objects and ground textures are still bog standard with Freemesh X Europe, but I've just taken the Milviz Sabre on a 300knot VFR flight over country I know very very well and was able to accurately distinguish landmarks and particular river valleys, ranges of high hills and bodies of water where the family recently wandered. In fact, I didn't get lost!

Greatly looking forward to buzzing the west coast of Scotland! Gotta dash, they've refuelled me now...

[Bjoern]

Still downloading the NA mesh...

[Dumonceau]

Still downloading the NA mesh...

Eagerly awaiting your remarks Bjoern!!

Johan

[Bjoern]

Once I get out of the "Do tons of crap only two people care about" loop for other projects maybe.

[Ferry_vO]

Great stuff, hard to believe it is freeware!

[Bjoern]

My place of choices for pictorial comparison were total duds.

http://i.imgur.com/tKP5lv2.png
http://i.imgur.com/9GE2Ao7.png

N42° 3.05', W73° 58.15'

Wanna guess what's what?
(Hint: Both meshes are at 38m resolution. One is default, one is Freemesh.)


Southern Germany.
Default resolution 76m.
http://i.imgur.com/lfUCb2R.png
FreemeshX. I think I see a mildly higher valley floor and mildly more prominent peaks.
http://i.imgur.com/XAjEUV4.png



Default mesh resolution:
USA - 38m
Europe, Mexico, Canada - 76m (resampled from 90m, which is bad)
Everywhere else - 304 or 608 m (awful or really bloody awful).

If you never stray across the borders of the USA, save the bandwidth. Everybody else, go for it.

[Penz]

If you never stray across the borders of the USA it's your own fault!

(At least in the simulator.).

Try a Cambodia comparison, like the Angkor area NE of VDSR Siem Reap.

Also, patch 1.01 is out, correcting two problem tiles in Italy/Balkans. Nice support.



(I'll have to re-do my homegrown Angkor Wat now ...).

[WhiskeyEcho]

The US is to be LOD11 and was not included in the base North America.

[Ferry_vO]

Seattle area:



Swiss Alps:



Tenerife, clearly visible is the large crater around the top of the Teide volcano:

[roger-wilco-66]

I did some checks and it seems I'm not out of the mesh business :-) At least for those parts of the world I work on.
I also found that using two active custom meshes (e.g. FreemeshX at a lower priority and another one like mine on a higher priority introduce periodic stutters) which I noticed in FSX as well as P3D. So if you fly an area and you notice that the simulator stops every 2-3 seconds you might want to disable the lower prio mesh for the time being.

A very notorious area for mesh errors or anomalies are the Himalayas. The images below are shot around Fort Hertz in northern Burma / Myanmar.

Here's the default mesh (P3D), which is unacceptable to me because it is far from reality:




Here's FreemeshX Asia. I expected that since it's SRTM V3 source. These spikes and holes are present all the way down the hump.



Here's my mesh from the ASTER source. The anomalies were worse than SRTM in the source file, but I corrected them (a very laborious task) and the mesh resolution is higher than SRTM V3.

[Meshman]

Here's my mesh from the ASTER source. The anomalies were worse than SRTM in the source file, but I corrected them (a very laborious task) and the mesh resolution is higher than SRTM V3.

They are suppose to both be 30m source, AFAIK. How it comes out is another thing.

[AviatorMoser]

I did some checks and it seems I'm not out of the mesh business :-) At least for those parts of the world I work on.

Here's FreemeshX Asia. I expected that since it's SRTM V3 source. These spikes and holes are present all the way down the hump.
Here's my mesh from the ASTER source. The anomalies were worse than SRTM in the source file, but I corrected them (a very laborious task) and the mesh resolution is higher than SRTM V3.

Ha! Nobody is going to go out of business! There are certainly artifact prones in FreeMeshX, some of them source, and others sometimes from me (that I'm patching fast to correct)! That's why FreeMeshX is an evolving project, where corrections and new content will added in patches and expansions. The Himalayas for sure have their problems (as do some areas of Oceania as well). These will be corrected in time, free of charge of course. I actually started making my first meshes with ASTER v2 data, but it just had so much noise that I had ditched it once the new 30-m SRTM data was released in late 2014.

They are suppose to both be 30m source, AFAIK. How it comes out is another thing.

Yes, they are both 30-m.

[roger-wilco-66]

I didn't want to step on anyone's toes :-) I think the Freemesh initiative is really a great thing and know exactly how much work it is to resample and correct the source data.
As far as I'm concerned I'd really be glad if I could stop resampling large areas myself for these parts of the world where I roam around or develop for.

The reason why my meshes are partly in a higher resolution is because I resample them to a slightly higher density. I also like to "resurface" airfield meshes and the areas around them in the Daylon Leveller because I don't like the huge flat surfaces and imprinting effects that are caused by large flatten polygons. Further, I want to structure the ground like in the image below (berms, sloped runway), and for that you definitely need a custom mesh. So for as long as I make airfields like that I always will concern myself with local meshes.


Cheers,
Mark

[stovall]

Ha! Nobody is going to go out of business! There are certainly artifact prones in FreeMeshX, some of them source, and others sometimes from me (that I'm patching fast to correct)! That's why FreeMeshX is an evolving project, where corrections and new content will added in patches and expansions. The Himalayas for sure have their problems (as do some areas of Oceania as well). These will be corrected in time, free of charge of course. I actually started making my first meshes with ASTER v2 data, but it just had so much noise that I had ditched it once the new 30-m SRTM data was released in late 2014.

AviatorMoser, as one who has dabbled in the mesh department I am always looking for ways to improve. As you mentioned starting with ASTER v2 data, I am still using that program but will definitely be looking at SRTM data once I find it. My question is what do you mean that the ASTER v2 data had so much noise? I have noticed that from time to time there are spikes and areas that need flattening to remove what looks like an artifact. Is this what you are referring to?

[AviatorMoser]

AviatorMoser, as one who has dabbled in the mesh department I am always looking for ways to improve. As you mentioned starting with ASTER v2 data, I am still using that program but will definitely be looking at SRTM data once I find it. My question is what do you mean that the ASTER v2 data had so much noise? I have noticed that from time to time there are spikes and areas that need flattening to remove what looks like an artifact. Is this what you are referring to?

Noise is simply some high frequency artificial spatial data. Take this 90-m SRTM source:

http://i.imgur.com/5nhXlqM.png

And then compare it to the ASTER source:

http://i.imgur.com/A0DUC8j.png

You can see the small pits and bumps that is prevalent in the ASTER data. Now, you can filter this stuff out. Take the CGIAR SRTM void-filled 90-m source data, resample to 30-m, create a difference map, and then use median and mean filters to help remove the noise. This is a similar approach I will take with the artifact prone areas, such as Papau French Guinea and the Himalayas in FreeMeshX.

Don't get me wrong though. The ASTERv2 data is fine once filter out that noise. I just found the SRTM 30-m data just a better starting point as most of it is corrected. There are certainly some anomalous data zones though.

Here is the new SRTM 30-m data. http://e4ftl01.cr.usgs.gov/SRTM/ (under SRTMGL1.003)

[stovall]

Here is the new SRTM 30-m data. http://e4ftl01.cr.usgs.gov/SRTM/ (under SRTMGL1.003)

Thanks for the explanation and the link to SRTM. I am definitely a beginner when it comes to Mesh but learning.

[AviatorMoser]

I am definitely a beginner when it comes to Mesh but learning.

So am I! I've been only doing this for less than a year. You can see my inexperience in some of the FreeMeshX areas with artifacts that slipped through our review.

[AviatorMoser]

Because there are some coarser, but error-corrected data out there, fixing the artifacts like those in the Himalayas become a lot easier if we let the coarser data guide us on correcting the finer data.

Tonight, I've been working on getting rid some of the most apparent errors along the Hump. Not too shabby.

Before:

After:

[roger-wilco-66]

Looks good, the holes are gone
I think that method (using the coarse data as reference) is an advanced technique. When I experimented with that I found some negative effects though, on landscape features that are naturally extreme, like canyons / cliffs / cone shaped mountains.
Correcting raw DEM data by filtering or running it against a second dataset is a difficult but interesting matter. Dissertations have been written about that.
In older SRTM meshes NASA seemed to have done the corrections manually, it looks like they did the corrections by directly __painting__ the geotiff greyscale patterns.

I'm going to check the Solomons next. Besides making localized meshes for the airfields, I used the ASTER mesh there which was horrible until I fixed it. One area was not useable at all because it was totally shifted.

Good work!
Mark