Manual Augmented/Live Reconstruction?

[Wilco Boode]

Good day all,

I have used meshroom quite a bit, including the Live Reconstruction function, as I find it really handy.

I was looking into a way to automate this process, basically using the CLI to manually replicating the Live reconstruction method using  images send to my home server.

However, every time I try to do this, I get stuck at various steps of the pipeline. I even tried creating 2 meshroom files, one with the basic pipeline, and one where I replace ImageMatching with the ImageMatchin Multi SFM (basically replicating Live Reconstruction inside the Meshroom GUI). 

It doesnt matter what I do, There is always a point where it fails. Right now I get an error in the Depthmap step that it cannot find the images belonging to certain viewpoints.

Is there anyone, or any way, that this could be properly achieved?

Simply said I want to:

1. Make photos on a location

2. Send those to my home server

3. When x amount of images are gathered, send them to a separate folder and start meshroom on CLI.

4. Run the meshroom pipeline over it, where every consecutive run uses the sfm data from the previous run (just like Live Reconstruction does) using imagematchingmultisfm.

I would be extremely happy if someone is able to help me achieve this indicated pipeline.

Kind regards,

Wilco Boode