Export point cloud depth map of each picture

[aaronfhd]

I was wondering if there was a way to export a depth map of each picture from the point cloud in the view port. I would need it to export at the same resolution of my image

This will be extremely useful for the project I am working on. 

[SAV]

Hi aaronfhd,

Right click on an image/photo in the photos pane and then choose EXPORT DEPTH...
You can then select if you want to export the diffuse map, depth map and/or normal map of all cameras, selected cameras, entire workspace or current photo.

I hope that's what you were looking for.

Regards,
SAV

[aaronfhd]

Unfortunately that does not work as you need to build a mesh to do this.

I need pixel for pixel accuracy and the mesh adds more stuff.

[chrisd]

I believe CloudCompare (open source) might be able to do this, it doesn't work with images however, so the point cloud would need to be exported. I'm not sure if scripting the view based on camera locations is possible.

[Alexey Pasumansky]

Hello aaronfhd,

Do you mean points of the sparse or dense cloud? Currently there's no access to the dense cloud points via Python. In principle, however, it is possible to export the dense cloud to XYZ file and use it on input, but it may take quite a long time to check, if the point appears on the certain image and estimate it's distance to the camera center.

[aaronfhd]

Hello aaronfhd,

Do you mean points of the sparse or dense cloud? Currently there's no access to the dense cloud points via Python. In principle, however, it is possible to export the dense cloud to XYZ file and use it on input, but it may take quite a long time to check, if the point appears on the certain image and estimate it's distance to the camera center.

The dense cloud,  I need be able to render the point cloud depth for each image. so for example I have a image with trees.

I remove all the points except the trees and then render a depth map for each image.

[aaronfhd]

any update on this.  I want to render a depth image of the Dense point cloud from each image.

Right now the only way is to export from the model.  This creates a issue with trees. Would like just to do it from the dense point cloud.

Thanks.

[aaronfhd]

Going to give this a bump as this would be great feature to have if we can do it.

[Alexey Pasumansky]

Hello aaronfhd,

We are considering an implementation Python API function similar to model.renderDepth, but for the dense cloud. In a couple of days I will let you know, if this function will be available in the next pre-release version update.

[Alexey Pasumansky]

Hello aaronfhd,


We are planning to add dense cloud render options to Python API in the next update of the version 1.6.0 pre-release. The feature also allows to define the size of the rendered point (in pix) and to add the transparency (alpha channel). Please let us know if any other additional parameters might be useful for your task.

[Alexey Pasumansky]

Hello aaronfhd,

We have updated a pre-release version of 1.6.0. Now you can use chunk.dense_cloud.renderDepth() method for your needs.

[jedfrechette]

I haven't had a chance to test it yet, but am looking forward to this new functionality.

Maybe it's already possible, but it instead of just exporting depth maps as single channel images it would be very useful to export them as deep images [1] with multiple channels. Undistorted exrs with R, G, B, Depth, and alpha channels would pack a lot of useful data in to one package.

[1] https://wikipedia.org/wiki/Deep_image_compositing

[lyhour]

Dear Sir, I want to know that you can solve it already. I need the dept image or depth information corresponding to RGB image as well. When I export the depth image from the Agisoft software, It is not well corresponding. What should I do to solve this problem ?