my modells does not fit anymore - google earth changed terrain

[Utta Hagen]

Google earth did change the image and now my modells are not anymore in the right position. What can I do? I have over 100 modells in that area.
Thanks for your answer!

[sketchutta]

Am Dienstag, 10. April 2012 19:54:22 UTC+2 schrieb sketchutta:

Google earth did change the image and now my modells are not anymore in the right position. What can I do? I have over 100 modells in that area.
Thanks for your answer!

 

Here are my models:
sketchup.google.com/3dwarehouse/cldetails?mid=c6e1ae5989e0b93ea1635aae7b53091a&ct=mdcc&prevstart=0
or have a look to sketchup warehouse: sketchutta museumsdorf bayerischer wald

[matthiasbasler]

Hi Utta,

I fear that it will be your task to move all the buildings, that is,
for each building remove the georeferencing information, then import
the terrain again, move the building so it fits onto the terrain,
check in GE and, if OK, re-upload the model using the same model ID.

If you need more details on this "updating" process and model ID
business, have a look at the last chapters of my tutorial here:
https://docs.google.com/Doc?id=dgft4kkq_0f8fd79dh

I can only recommend to store several neighbouring buildings in one
model file and upload them as components if you haven't already done
so. This way, whenever the imagery changes, you can move all models
within a file at the same time and you only have to update the terrain
only once for every file, not once for every building. (Saves some
time.)

P.S. Very nice collection!

Greetings from Weißenburg, Bayern.
Matthias Basler

[Google Guide Jordan]

Hi Utta,

To give some background on imagery shifts, we have been working for a solution to solve this problem since we started the 3D buildings layer many years ago.  As you can see with your models, there is not a uniform shift meaning that models are misaligned by different amounts and in different directions.  This is often due to the fact that different imagery sets are not captured from the same angle.  Since the earth is not flat (hills and slopes) there is slight transformation of the imagery in an unpredictable way.  In the past we tried to identify these shifts and move models independently of one another.  This occasionally returned good results, but it also caused some bad alignments.  This was most noticeable in cases where models should be aligned precisely to one another such as row houses.  We have since abandoned this approach as it appeared to create more problems than it fixed.

The other approach is to simply move all the models in an area along a uniform shift.  As you can see from your example, this approach would correct some models, but many others would still be misaligned.  For those models that can't be aligned with this method, a manual approach must be taken.  However, if the new models are close (under 5m off) to being aligned, they would still be acceptable per our acceptance criteria and you wouldn't have to manually fix the models unless you felt so inclined.

In the end, some amount of manual editing will always be required to correct the alignment due to new imagery.  I completely understand how unfortunate this is.  After all, you spent many hours creating and aligning your models, and we go and update our imagery and cause this problem for you.  I am trying to be honest and open with you by saying that we are far from having a good solution for this problem.

If you want, I can run a uniform shift for all of your models.  Please know that this will not correct every model, but would hopefully improve the majority.  I also can't promise that I can get to this for 1-2 weeks, which would also mean that any new models you add to this location would also get shifted along with the rest.  Please let me know if you want to go this route.  Your other option would be to manually edit all the models as Matthias mentioned in his post.

Sincerely,

Jordan

[matthiasbasler]

Hi Jordan,

thanks for your honest answer on this.

I completely understand how difficult it would be to re-align all
buildings accurately, and I also agree that if an area has a not very
uniform shift then manual work is necessary from either side. Shifting
different buildings in different directions doesn't seem a good idea
to me either.

I just want to suggest that, since a 100% pleasing solution is
obviously not coming soon, that least following partly solution can be
implemented if possible:
1. Users get notified of imagery and/or terrain changes. (I have
registered to get notified but wasn't notified the last times :-( )

2. For cities or other areas with many buildings average shift vectors
are computed and a *uniform* shift is applied similar to what you
suggested above. (In my town of Weißenburg the shift had been rather
uniform (3.5m at 120° SE) and even if not perfect it would have
reduced the average offset from ~4m to maybe <1m, which would have
been hardly noticeable.

3.a Users can opt-out for automatic shifts for (all of) their
buildings by switching a flag in their user account if they don't
trust the system.
OR
3.b Ideally the user would get one mail for all buildings affected by
each vector, something like:
-----
Due to new imagery N models in the area of XYZ will be moved 2.7m to
SouthEast. Please check in Google Earth if this a suitable correction,
then click on [this link] to confirm the change. If we don't hear back
from you within two weeks the change will not be performed.
Kind regards, Google.
-----

I have no idea how much effort this would be on your side, but it
should solve maybe 80% of the cases and this would already be a big
step forward.
Thanks again for the feedback.

Matthias Basler