Selecting points within the point cloud by color would be great.

Some of my point clouds have a lot of noise, especially around leafless trees.  The points of the noise are almost always white.  If I could clip the processing area and select all the white points by their color, the point cloud editing process would be magical, instead of the mind numbing, time guzzler that it is.

16 Likes

Awesome idea. I often fiddle with the same problem. Around trees trying to filter the cloud - and in fact they are most of the times white. just filter the cloud by color would be awesome!

best, thomas

Agreed, this would also be extremely helpful when trying to remove the blue points from the sky.

Hello,

We have passed this suggestion to our colleagues in charge of the product development.
They will consider it for future release.

This could indeed be very helpful to have some kind of classification according the color of the points.
Thank you for the feedback.

Best regards,

I am curious if there have been any further discussions about this request at Pix4D. This would indeed be an extremely useful feature.

1 Like

I would also like to have this feature, do you know if this will be added in a near future?

I moved this post to our new Product feedback topic, where it fits better.

For the moment it is not in the development pipeline for the near future. However, feel free to add votes at the top of the post (thumbs up or down) to show that this would be useful to you.

Você pode me dizer se essa função realmente funcionará?

Hi Cleiton,

For the moment, our users cannot select points based on the color (manual intervention)

Our software does an automatic Point Cloud Classification based on color and geometry, classifying the points into the following categories:

  • Ground
  • Road Surface
  • High Vegetation
  • Building
  • Human Made Object

Please reconsider this feature once more…
Clipping a color range would be just as good.  

Hi Daniel,

Your feature request has been reported to our team as well.

Any feedback about your experience or information about features you would like to see in a future release is much appreciated. Thank you!

Pix4D Development Team -

I have ran into a product named VRMESH which has this feature for classifying point by selecting a point’s color and includes parameters to find similar colors. Also Global Mapper has such a tool we are exploring. I would prefer to stick with Pix4D for point cloud editing but it does not offer major advancements to speed up the editing process.

We want to be able to select a point by color and add to point group of our choice. Please help us out with this tool so we can maintain loyalty! We hope you take this to high consideration.

Regards

2 Likes

Hi Aaron,

We really appreciate your contribution to this topic. Your voice was passed on to our team.

Thank you very much

1 more for edit by color!  I am trying to make a 3d model of a dredge crane that is in water.  If I could eliminate blues it would be amazing!

Hi William, 

Your interest in implementing that feature is registered as well!

Thank you

Best

same here. Shiploader in the water… Photogrammetry doesn’t work. A lot of blue noise :slight_smile:

Is there any workaround? would it be possible to delete the water in e.g. Photoshop before the photogrammetric processing? How is Pix4d handling true black, or any other homogeneous colour?

1 Like

Hi Philipp,

You could try by annotating images in Pix4D. Please have a look at those articles:

Let me know in case you have some questions ;-) 

1 Like

Would be great to have this feature somewhere in the near future. Tomorrow would be good… :)  So also a vote from our side for this feature

Confirming that annotating (creating a “mask”) for water regions in photos improves the results.  (Should probably run step 1 alignment again after masking)

I had a peer project in which the water was creating lots of bad points.

  • Masking the water helped a lot.

  • But it was also necessary to create several tie points in the thin structures  

  • Had to use double size in “Initial Processing” -> “Keypoints Image Scale” (this would increase the static areas and create more automatic tiepoints in them) 

But I belive that if the peer is floating (moving), it would still be weird.

Cloudcompare has an excellent tool for converting colours to a scalar field.

This can then be used to sort the cloud by histogram intensity on the red, blue, green and composite bands.