Point Density setting

Hello Team,

On Pix4D mapper software, is it correct to say that when creating point clouds the three point density options High, Optimal, and Low use Original Image size, 1/2 image size, and 1/4 image size respectively when creating point clouds?

Thank you as always.

 

Henok 

Hi Henok,

In the processing options, you will see image scale and point density are different.

Image Scale : The image scale defines the scale of the images at which additional 3D points are computed.  From the drop-down list, it is possible to select:

  • 1/2 (Half image size, Default): Half size images are used to compute additional 3D points. It is the recommended image scale
  • 1 (Original image size, Slow):   The original image size is used to compute additional 3D points. More points are computed than with half image scale, especially in areas where features can be easily matched (e.g. cities, rocks, etc.). This option may require four times more RAM and processing time than when choosing the default value 1/2 (half image size), and usually, it does not significantly improve the results
  • 1/4 (Quarter image size, Fast):   Quarter size images are used to compute additional 3D points. Fewer points are computed than with the half image scale. However, more points are computed in areas with features that cannot easily be matched such as vegetation areas. This scale is recommended for projects with vegetation
  • 1/8 (Eighth image size, Tolerant):   Eighth size images are used to compute additional 3D points. Fewer points are computed than with the half or quarter image scale. However, more points are computed in areas with features that cannot easily be matched such as vegetation areas. This scale is recommended for projects with vegetation.

Point Density : This parameter defines the density of the densified point cloud. The point density can be chosen from the following options:

  • Optimal (Default): A 3D point is computed for every (4/ Image Scale) pixel. For example, if the _Image Scale _is set to 1/2 (half image size), one 3D point is computed every 4/(0.5) = 8 pixels of the original image. This is the recommended point cloud density.
  • High (Slow): A 3D point is computed for every Image Scale pixel. The result will be an oversampled Point cloud that requires up to 4 times more processing time and RAM than optimal density. Usually, this point cloud option does not significantly improve the results.
  • Low (Fast): A 3D point is computed for every (16/Image Scale) pixel . For example, if the _Image Scale _is set to 1/2 (half image size), one 3D point is computed every 16/(0.5) = 32 pixels of the original image. The final point cloud is computed up to 4 times faster and uses up to 4 times less RAM than optimal density.

I have used the information from here: https://support.pix4d.com/hc/en-us/articles/202557799-Menu-Process-Processing-Options-2-Point-Cloud-and-Mesh–Point-Cloud

 

1 Like

Bonjour,

Lors de nos traitements sous PIX4D, nous utilisons les options suivantes :

  • Image scale : 1/2 (Half image size, Default)
  • Point Density : Optimal (Default)

Nous souhaiterions savoir combien de point sont créer dans le nuage à partir des pixels de l’image.
Dans l’exemple de la notice de PIX4D il est indiqué que dans notre cas nous devons faire 4 / 0.5, ce qui nous donne un point 3D tous les 8 pixels. Nous ne comprenons pas d’où vient le 4 qui est utilisé dans la formule pour obtenir le résultat d’un point 3D tous les 8 pixels. De plus, cette information vient en contradiction avec la première description de l’option « Optimal » qui indique qu’un point 3D est créé pour chaque pixel de l’image.

Pourriez nous fournir plus d’information sur le calcul que nous devons faire pour savoir combien de point 3D sont créés à partir des pixels avec les options que nous utilisons.

Merci d’avance de l’attention que vous porterez à notre requête,

Cordialement,

GUERRY Julien

Bonjour @julien.guerry ,

The number of points can be read a posteriori after processing Step 2 from the quality report.

The factor 4 or 16 is defined by Pix4D in order to decide how many additional points must be created. I remind you that the purpose of the densification is to densify the cloud of points generated during step 1, those we refer to as Automatic Tie Points. You can find additional information at https://support.pix4d.com/hc/en-us/articles/203269885-How-many-Points-are-generated-during-step-2-Point-Cloud-and-Mesh.

For your case, it is correct to say that one 3D point is computed every 4/(0.5) = 8 pixels of the original image. Could you point me to the place where it says that with the “Optimal” option a 3D point is created for each pixel of the image?

1 Like