Control Gizmo for Chunk Orientation, Position and Scale

[Infinite]

Hello all fellow Agisoft users, I have spoken to Alexey about the idea of a 3DSMax style Gizmo control for Agisoft scenes, specifically for Characters. But I wanted to see if others though it might be useful. This is also useful for buildings, objects and human capture in general.

At the moment there is no real easy way for users to specify custom Chunk Orientation, Position and Scale ready to export for external applications as we currently have no extrinsic calibration data. Also many external application have their own ways and scales for importing, which is a pain.

Such a Gizmo UI control would be so powerful, as it would allow users to set the position in space, the orientation and scale ready to export correctly to any applications. We can then use the various Chunk Alignment tools to re-align other similar chunks or .psz scenes, using markers or camera alignment.

i.e.



At the moment it is very arbitrary, scale can be set using Markers, which is OK but orientation and position is very hard to do. Position impossible I think.

Perhaps the 3 axis (x,y,z) Gizmo control would have known scale sizes for each axis, or we can set it's scale from centre to end points as 1cm, 10 cm, 50cm etc Or we can physically scale the Gizmo in size to match reference in the scene, i.e. markers or known geometry size and the Gizom is set to a certain size like 1cm, 10cm, 50cm etc

Thoughts?

[Bendytoons]

This feature would be a godsend for me. Right now I am bringing my models into 3DCoat for clean up, and it is of the opinions that they are all 2 inches tall. I have to supercharge the voxel resolution of the models to get detail to show up. Also it would be nice to be able to set orientation in PS instead of later in a 3D editor.

[Nico]

+1
That would be extremely useful - provided that it also moves all the cameras with it, so that exported cameras still match the scan.

[Infinite]

+1
That would be extremely useful - provided that it also moves all the cameras with it, so that exported cameras still match the scan.

That's the idea. A full Chunk change for position, orientation and scale. So all Cameras also move.

I also had the idea of exporting the mesh (decimated) to an application, like a 5000k mesh. Scale, Orient and Position as you need to. Export back to Agisoft (making sure the point order doesn't change) and have an option to make the Chunk match the imported Mesh's Scale, Orientation and Position using the Meshes Point Order as reference. A bit of a hack but another option.

[Alexey Pasumansky]

Hello,

Thank you for the suggestions.

In Professional edition the similar operation could be performed using Python scripts to orient axis parallel to the sides of the bounding box (provided that it is correctly oriented) and move coordinate center to the Bounding box center.

[Infinite]

Hello,

Thank you for the suggestions.

In Professional edition the similar operation could be performed using Python scripts to orient axis parallel to the sides of the bounding box (provided that it is correctly oriented) and move coordinate center to the Bounding box center.

Hi Alexey, related to this thread and the possible update of how the bounding box is displayed, i.e. control handles and also center point visual cue to help aid with using the bounding box to change position and orientation of Chunk.

It would be good to be able to move the whole bounding box as one, instead of just using the corner handles as it can be quite tricky to position the bounding box as the center point for the model without visual cue and just using handles to move it's position. They tend to jump some times when changing positions.

Handles are great, so is Rotation control but it would be nice to move the whole bounding box as one on each different axis x,y,z

[erwanl]

Hi,
I'm digging up this topic, which I found from ir-ltd.net
I've been testing PhotoScan along with a few other softwares for photogrammetry use, in order to choose which one to teach to students in a VFX school.
While Photoscan gave me the best results in terms of quality of dense point cloud and texture extraction, the fact that I can't use a regular "translate/rotate/scale" gizmo is a deal breaker.

I did see the workarounds using python scripts to hack the region position, but we're not going to use that. The school will not buy 30x pro licenses just so we're able to position our scans properly, using a weird workaround, for our 12 hours of photogrammetry training/year.

Or is there an actual method by now that I'm missing?