Georeferencing: confusion of coordinates

Started by hvdosten, January 04, 2019, 05:02:07 AM

Previous topic - Next topic

hvdosten

Hi guys, I need your valuable hint... I don't understand the following:

What I want to do: To drop a small aerial picture on a part of the terrain.

Why I am confused? I have downloaded some DEM data (ALOS World 3D). First, everything works fine

[attachimg=1]

But.... how do these coordinates come about? 

[attachimg=2]

Thanks a lot!!!

Harry

Oshyan

I'm not sure I understand what you are confused about. What seems wrong about the coordinates? And what image are you trying to load/align with this terrain? Does the image have georeferencing data too? Is it important that the data be correctly positioned on the planet, or do you just need the terrain and imagery to line up?

- Oshyan

cyphyr

The coordinates shown in the bottom picture are only showing where the mouse is hovering above.

They are not the coordinates of the loaded DEM data and they are not meant to be.

If you want the DEM to be at the top of the planet un-tick "Georeference" and add a Heightfield operator "Heightfield resize" with the dimensions copied in from the XY size (in meters) from the top image.

Hope this helps
www.richardfraservfx.com
https://www.facebook.com/RichardFraserVFX/
/|\

Ryzen 9 5950X OC@4Ghz, 64Gb (TG4 benchmark 4:13)

hvdosten

#3
Hi,

thanks a lot for your replies.

Okay, what I want to do is to drap an image on the terrain:


[attach=1]

The image is georeferenced:

[attach=2]


so, everything should be perfect. But it isn't, the image is not on the correct place. So I don't know how to correct this, because I have no idea how to get information about the location of this image...


Again, thanks a lot,
Harry


EDIT: I have verified all the coordinates with Google Maps. They are okay...

Oshyan

So you are trying to *manually* georeference the image? It does not have embedded georeferencing data? (I ask because you have auto georeference disabled here). What is the source of the image data?

Something does not seem right with the Lat/Lon numbers to me. I don't think your NE and SE corners should be the same Longitude, for example. Perhaps the numbers are reversed?

Also, try going to the Info tab and see what size it says the image is in meters and whether it is the correct size that you expect this image to represent.

- Oshyan

hvdosten

#5

Harry


hvdosten

#6
   ---

Oshyan

Strange, you removed the most recent reply, but I was in the middle of composing a response. Did you solve the problem?

By the way image alignment, etc. does not work properly in the forum here, so you should just use the inline image attachment functionality. The way you are posting images keeps breaking things and I have to correct it. No blame, of course you did not know, but just to let you know, you can just use e.g. [ attachimg=1 ] to get a full-size in-line image.

- Oshyan

hvdosten

#8
Hi Oshyan,


oh sorry, I was totally frustrated... But now I am really very happy to hear that you wanted to response to my problem... So I will repeat my answer now.

No, I did't solve it. If I would have been able to solve it, I would have told it and the way how I did manage it, so that other users can use the information if they will also have this problem some day. Moreover, yesterday I tried it again and I feel even more confused:

The coordinates (corners of the satellite image) presented in the next figure are not correct (I changed them to be able to see the image within the area of the heighfield. If I  use the correct coordinates I can not see that image...). They should be:

NW  41.251106     35.387016
NE    41.251106     35.485490
SE    41.186399     35.485490
SW  41.186399      35.387016


so the difference of this area should be (very roughly) north-south: 0.07 degrees and east-west: 0.10 degrees (same differences as for the image draped in this figure:)


[attach=1]

but according to this figure, this is not correct. The area covered by the image is pretty too small. Moreover I had to exchange the latitudes of the northern and southern corners to get the correct orientation of the image. The orientation of the heighfield is correct. Why? What I am doing wrong?


Well, here is my removed reply:


Yes, I try to georeference manually a jpg. It has no embedded georeferencing data. The source of the image data are Ground Penetration Radar data.

The heighfield data were downloaded from http://opentopo.sdsc.edu/ as well the corresponding satellite picture (just to be able to control the results):


[attach=2]


I georeferenced this image using:


[attach=3]


This is the jpg I want to drap:

[attach=4]


and geoereferenced this jpg using:

[attach=5]


But the result is pretty strange (location and orientation of the satellite image is not correct):

[attach=6]


The correct position of the jpg would be (please note the circle):


[attach=7]



I really would like to know what I am doing wrong....


Again, thanks a lot for every small hint...


Harry




Oshyan

I can only say that I am pretty sure your coordinates are off. The point you reference on the map is at 41.207409, 35.429966 according to Google Maps:
https://goo.gl/maps/BdDcpzePTiQ2

If I plot the box you laid out with your coordinates in Earth Explorer it's much larger than you want your image to be:
[attach=1]

Now of course the decimals aren't exactly the same in my image above, but they are close enough (even if you change the whole number, for example 35.485490 to 35.500, it only shifts it a little bit).

If I draw a box in Earth Explorer around the smaller area indicated for your intended placement, this is approximately what I get:
[attach=2]

So the question becomes: where are you getting the coordinates that you think should give you correct placement for your image? What is the source of those numbers?

- Oshyan

hvdosten

#10
oh the coordinates are correct. Sorry I should have given you some more information. Here they are:

Google Maps show you a small hill next to the village of Oymaagac with some blue plastics on the right side:

[attach=1]


This is an excavation of a hethitic town. The first measurements (geomagnetic gradiometry) show you the townwall and especially the hethitic temple:

[attach=2]


Some years later GPR delineated the stone structures more clearly:

[attach=3]

A part of the temple is now excavated (you can see this in Google Maps), the areas of excavation are marked as white areas:

[attach=4]


Well, now for the final publication I wanted to presentate all these results of my geophysical mappings in a realistic landscape, made by Terragen...




EDIT:

For the preliminary report I used another software (Natural Scene Designer), the result is pretty nice:


[attach=5]

but doing this with Terragen should be much more impressive... ;-))


Oshyan

Apologies, I was confused because you are trying to georeference *two* images here and I was not clear which one was having issues. Matt mentioned to me your image may need to be flipped, so that's the current theory, and you should try that (on your main satellite image).

But more generally I am wondering how you have approached the development of this scene. You may have done this, but here's how I'd go about it. I would first load the terrain/heightfield, which I presume you have done. Then make sure the satellite image from the data site lines up with the terrain. If *that* does not line up, getting the other to line up will be impossible. So solving that issue would be the focus. Once you have your satellite data lining up, the rest should work fine.

So it is probably an issue of the satellite data not lining up, which may be solved by flipping it on one axis or another (on the Flip tab of the Geog Image Map Shader).

- Oshyan

hvdosten

#12
Hi, my only problem is, not to understand why I can not see the draped image on the terrain:

[attach=1]


while the coordinates of the corners of this image (it is the figure no 4 of my answer from January 11) should be within the corners of the terrain:

[attach=2]


And yes, it should really be within the first image shown in this post.


It is pretty misplaced:

[attach=3]


The black arrow demonstrates the correct position:

[attach=4]

and this demonstrates the difference between the position calculated by TG (white arrow) and the real position (black arrow):

[attach=5]




Oshyan

OK, I think at this point the only way we can really diagnose this further is if you can provide all of your test files. Taking a direct look at your entire setup may give us more info so we can identify whether there is a bug, or just some setting(s) that need adjustment.

You can use Export Gathered Project (file menu), then compress that (zip or .7z) and send to support AT planetside.co.uk

- Oshyan

hvdosten