georeferencing with transformation grid

[GGR]

Hello Daniel,
First of all, thanks for this great tool ! It helps every day !

I'ld like to use CC to georeference relatives point clouds. The current process consist of
- segmenting exteriors scans,
- picking points whose coordinates are known,
- reprojecting a subset, and then
- using the transformation grid to georeference the indoors scans

My problem is, applying a transformation grid does not modified the points coordinates. It does if, for example, I merge scans but it is not optimal in my process.

So, based on the georeferencing of a subset, is there a way to modify the coordinates of a an other ?

Thanks by advance
GGR

[daniel]

I guess in this case you'll have to (manually) copy the Global Shift information from one entity and transfer/paste it to another one (with 'Edit > Edit global shift and scale').

[GGR]

Hello Daniel,
Thanks for your answer,

What I've observed so far is, editing the global shift doesn't reproject datas. It just modify the global box center. Locals coordinates remains locals.

The second thing is, local datas are heading north. But, due to the kind of instrument used, it only can be based on magnetic north and not the geographical one. So the process I'm trying to find have to also allow rotation, what does perfectly a 4x4 matrix.

I keep on scratting my head to find my way throw, and avoid the merging phase.

[daniel]

So I was confused by 'transformation grid'. Do you mean the 4x4 transformation matrix?

But in this case I'm not sure to understand why applying this transformation matrix would not 'change the local coordinates' ... I must be missing something.

[GGR]

Hello Daniel,

Sorry for the delay for this response.

Indeed, the transformation applied is a 4x4 matrix. I maid further test to have a better understanding of the behavior of CC when transforming with a 4x4 matrix.

Because it can reproject natives locals coordinates based on GCP in global coordinates, we use the "Align" tool.

A 4x4 transformation matrix is produced after this Align process. This matrix could be used to reproject the other parts of the same scan but, it appears to be applied to the shifted box center and not to the global box center. So data are projected in the same local space, with different global coordinates. Witch make a significant difference for third parts softs using others importing methods.

To illustrate those differences, I roughly georeferenced a e57 structured data set available here https://webimaging.lillemetropole.fr/ex ... _scans.e57. It is composed of :

• "align_point_picking" set has been georeferenced using the Align tool (properties in the first attachment),

• "align_mat4x4" with the 4x4 matrix (properties in the second attachment)

Code: Select all

0.983088016510 0.172502502799 0.061488993466 799.030822753906
-0.173136740923 0.984884619713 0.005099871196 905.169494628906
-0.059679824859 -0.015659626573 0.998094797134 1.418987154961
0.000000000000 0.000000000000 0.000000000000 1.000000000000 

• "original1", "original2" and "original3" are native parts of the local registered data set (properties in the third attachment).

If it's the more appropriate way to georefence data sets based on 4x4 matrix, is there a way to apply a 4x4 matrix to the global box center ?