Agisoft Metashape

Agisoft Metashape => Feature Requests => Topic started by: wojtek on October 18, 2020, 11:09:49 AM

Title: Gradual Selection -> Model Confidence
Post by: wojtek on October 18, 2020, 11:09:49 AM
We should be able to select parts of the model using the confidence values.
Title: Re: Gradual Selection -> Model Confidence
Post by: dpitman on October 18, 2020, 04:45:50 PM
I don't think the sparse cloud has a "confidence" value associated with points. But it is probably similar to one of the other criterion in the gradual selection tool.  It would be nice if the manual had more detailed information on criterion values. What is good and bad.

You can filter the dense cloud by confidence.

Tools > Dense Cloud > Filter by Confidence

You could then select and delete points based upon your filter setting if desired.
Title: Re: Gradual Selection -> Model Confidence
Post by: wojtek on October 19, 2020, 03:49:47 PM
I don't think the sparse cloud has a "confidence" value associated with points. But it is probably similar to one of the other criterion in the gradual selection tool.  It would be nice if the manual had more detailed information on criterion values. What is good and bad.

You can filter the dense cloud by confidence.

Tools > Dense Cloud > Filter by Confidence

You could then select and delete points based upon your filter setting if desired.

Model (vertex) confidence, not point cloud.
Title: Re: Gradual Selection -> Model Confidence
Post by: dpitman on October 20, 2020, 02:57:20 AM

Model (vertex) confidence, not point cloud.

Interesting.  "Confidence", with regard to points, is in relation to overlap of photographs and the # of images that see each pixel from what I understand.

How is the "confidence" of either vertices or polygons within the mesh defined?
Title: Re: Gradual Selection -> Model Confidence
Post by: wojtek on October 20, 2020, 10:34:08 AM

Model (vertex) confidence, not point cloud.

Interesting.  "Confidence", with regard to points, is in relation to overlap of photographs and the # of images that see each pixel from what I understand.

How is the "confidence" of either vertices or polygons within the mesh defined?

It represents the confidence with which the model was reconstructed (interpolated areas have 0 confidence for example then low overlap areas etc).