WED Coordinates in MCX

[JoePilot]

Hi all,

I'm trying to accurately place my ground objects into P3D, my method at the moment is using WED (X-Plane tool) to accurately place it & then obtain the Lat & Lon coordinates.

However, inputting those coordinates into the MCX ground wizard doesn't align it the same as it was in WED. I understand coordinate types can vary not sure if that's causing issues here.
When loading P3DV4 it's not aligned to the orthoimagery as it was in WED.

Maybe @arno will know why WED coordinates don't seem to work with the MCX ground wizard?


Appreciate any help!

 

[arno]

Maybe I miss the point, but in your screenshots the coordinates seem the same.

 

[svdm13]

Hi, I am having the same problem with definitely different values. The WED locations are the correct ones.
I do know the data in the dsf files are compressed which could be part of the problem.

Did you find a solution?

Some examples

 

[arno]

Hi,

No, I have never been able to debug this issue. If you can provide a sample DSF file plus some information like the screenshot of what the coordinates should be I can have another look.

 

[svdm13]

Hi Arno,

I have attached a reduced sample airport with a few objects to demonstrate.
The WED xml file is OK and shows the right location (also verified with googlemaps), the .dsf I don't know as it is not readable.
I read it is being compressed what could be part of the issue...

Note as mentioned in another mail earlier I had some issues with converting textures with MCX to Aerofly.
As work around I now always use the standard textures but converted to ads. This gives me a predictable result not ideal but workable :
- When the DDS flip flag in MCX is true it looks OK in the preview but flipped after conversion in Aerofly
- When flag is No it is wrong in MCV preview but OK in in Aerofly

I really appreciate your help and all the efforts you have put in this great tool !
Really a petty IPAC lacks behind on offering decent tools to their customers like WED and MCX alike

Cheers

 

[arno]

Hi,

I don't see an attachment to your post....

Those DDS textures are tricky indeed. X-Plane has them flipped compared to FSX/P3D, that's what the flip option does.

I think when converting to AF2 I usually used a format like BMP and let the AF2 tools compress the textures again. But I guess I need to add a flip option for the export as well, depending on what the target simulator is.

 

[svdm13]

Sorry Arno,

Not sure what went wrong with the upload. It is a zip file holding all data.

Regards,

 

[arno]

Thanks for the file. When I checked it and compared the XML coordinates with those shown in the MCX object placement editor, I don't see a big difference. The small difference is probably rounding due to the accuracy with which it is stored in the DSF file.

I did find a small bug though that when just opening the object placement editor the wrong placement could be shown. But after cycling through the objects that did not happen anymore. That bug will be fixed in the next development release.

 

[svdm13]

Hi Arno,

I have attached the result of the complete airport in the actual simulator and as you can see the errors are really significant. It might also depend on the size of the complete airport.
A few objects close to each other might have less error. I have even updated the objects and moved the origin to the center (as I read this might have influence as well). Afterterward I had to re-position them
again in WED but still same outcome :-( Wouldn't be better to import the WED.xml file as this is 100% accurate and not compressed.

 

[arno]

Hi,

The rounding error between the WED coordinates and what MCX reads from the DSF file are quite small. I just calculated it, but that gives you a maximum offset of 10 cm.

Since your reference is the photoreal, are you sure that the photoreal is accurate in both X-Plane and AF2? Did you cross check if the runway thresholds are at the same coordinates in both sims for example.

 

[svdm13]

Hi,

I have just imported the DSF file back into an empty WED airport and all objects from the DSF file line up perfectly. So the data seems to be correct even after compression and decompression.
Not sure what could be the reason...

Regards,
Stefan

 

[arno]

Did you take the screenshot where the buildings are not placed over the photo in X-Plane or is that in AF2 after converting?

Like I wrote before I would first check if the photoreal is positioned right, else you use a wrong reference.

 

[svdm13]

Hi Arno,

The last picture I sent was taken from a helicopter in the actual simulator. As you can see we are not talking about small errors.
WED has its own background taken from ESRI on which I aligned my own generated photoreal taken from Aeroscenery that is also used in the simulator.
When I toggle between my reals and the ESRI ones they 100% match. When I verify the WED coordinates with the same location on google maps they also 100% align.

The simulator shows the objects with a similar error as shown in MCX position page (using google maps) When I correct them here it is OK in the simulator as well/ Unfortunately we are
talking about 2000 objects...

It seems to be the translation between DSF coordinates and MCX. Not sure how this is done. It uses indeed a lot of decimals with quite some impact on the scene.
If I manually copy the WED coordinates in the MCX spot it aligns perfect.

I have the impression that the error changes the bigger the airport becomes which makes me think on a kind of earth rounding correction. The few objects in the airport I sent
showed indeed smaller errors.

Regards,
Stefan

 

[arno]

Hi,

Let me take another look later. In the test object you send me I only saw offsets of centimeters, which is normal due to the way dsf files store coordinates.

 

[svdm13]

Hi,

I have now manually copied the locations of some buildings from the WED xml file directly into the TSC file with the following result.
note: I still need to add a ground mesh to smoothen the ground surface.

 

[arno]

Hi,

I had another look at the test file you provided, but if I look at those objects in the MCX object placement editor they seem to align quite well with the Google Maps imagery. Would you be able to provide an example that has a big offset as well? Maybe I can spot something from it.

 

[svdm13]

Hi Arno,

It is strange. When I make a smaller scene with eg. 10 objects (same objects and on same background) the errors are indeed small. However when I use the real scene that holds 2000 objects some are right spot-on and others are seriously off. even the heading might be 2-3 degrees off. I Have paste some examples of the real scene, all compiled in one go using the object import and the .dsf file.

- Sathub 3 is perfect 0,03 meter off. This is one of the first objects in the TSC file. => OK
- Safran object 1953 in the TSC file => 35 m off
- Sabena technics2 object 1951 => 35 m off
- Beltech object 1432 => 38 m off
- Garage 1 object 897 => 10 meter
- Garage 2 object 896 => 10 m off

It is also strange X and Y offset are similar and it seems to get bigger when the object count goes up. Not always the same value though so it is not a simple shift of the background image.
(even using the built-in background from WED which should be accurate)

For your reference I have added some screenshots from the real simulator to show the actual displacement and from the coordinate converter tool I used to calculate the distance between the WED coordinate and TSC file.

I hope this might give you an indication.

Regards,
Stefan

 

[arno]

Hi,

Sounds like I need to test with a more complex DSF file then, if the bug does not show with the test file you send me last time. But that will make debugging a lot harder as well

 

[svdm13]

Hi Arno it took some time but I have been able to trace down at least one strange behaviour. I rearranged my objects in WED and grouped all my buildings and moved them on top of the list. The same objects seem to be
much better converted now. So I made a sample airport with just cones on the airport of brussels Zaventem. I duplicated each cone 96 times and divided them in 4 groups. When you look in the WED xml file you will see the same coordinate 96 times where as in the converted . TSC you only find the same coordinate only 24 times (=number in one group) It looks like every group gets an offset on the 5th digit behind the decimal. When you look at the position in MCX you will also detect 4 groups of cones on each location.

Conclusion : the more object groups you use in MCX the bigger the position errors. Identical objects within a group seems to be at the same locations independent the number of copies.
I have attached the test case.

I hope this helps to find the bug.

Regards,
Stefan

 

[arno]

Hi,

Thanks for the test object. I'll see if I can find the bug with it. I'll have to check how groups are stored in the dsf file, might be the bug is in that part.