Select to view content in your preferred language

Resolution of mesh model generated from Site Scan

1074
6
Jump to solution
02-23-2022 01:09 AM
st_123
by
New Contributor II

May I know what factors will affect the resolution of a mesh model generated from the SURE engine in Site Scan for ArcGIS? Sometimes, I choose "High" in the options but the results are not in high resolution. May I know the reasons behind this issue? 

Tags (2)
0 Kudos
2 Solutions

Accepted Solutions
GordonSumerling
Esri Regular Contributor

@st_123 It is one recommendation to improve accurcay. I would also fly lower to increase resolution. As for the sinking below the surface. This blog article is probably relevant

View solution in original post

0 Kudos
NicoBonnafoux
Esri Contributor

Though the initial question about improving resolution was answered by @GordonSumerling  (thank you), @st_123 reached out to us through Technical Support channels and provided additional details. It was determined that the challenges that led to this question were caused by the data capture and affected all the outputs, not just the mesh. 

I'll share our findings and the tips we provided here in case they can help others in the community: 

  • Not all photos were captured the same day. All images in a mission should be captured on the same day, and ideally close in time to ensure that environmental conditions remain similar, particularly lighting.
  • Photos show blur, and the data was not captured using Site Scan Flight. We recommend using Site Scan Flight to capture data when using a supported drone like the P4R used here because it optimizes flight speed and camera settings for mapping, which many other apps (including DJI's) do not. At the least, increase the shutter speed to ensure photos are sharp, ideally 1/640s or faster, with Shutter Priority enabled. 
  • Overlap was excessive in certain areas, which can lead to long processing time and server timeouts on a larger datasets like this one. Using Site Scan Flight would also help here, some other apps are known to incorrectly calculate overlap.
  • Flight elevation was inconsistent throughout the different data collection flights. Use Site Scan Flight with Terrain follow, this will ensure that flight height above ground remains constant, even if the pilot needs to reposition for different flights.
  • Drone/camera hardware: 
    • Ensure the lens is clean before each takeoff.
    • Takeoff from a hard surface and ideally clean takeoff pad to avoid dust pickup during takeoff.
    • In warm and humid environments, let the drone hardware acclimate to ambient temperature for about 5 minutes before taking off (avoid going from cold and dry air conditioning of a car/truck directly to outside). This can also help prevent condensation appear on the lens.
  • GCPs: make sure to use targets that can be very precisely measured in the images, such as checkerboards. Painted "X"s or visible identifiable features like corners cannot be measured as precisely. 
  • Some images were captured in cropped resolution, 4864x3648, rather that the full 5472x3648. Site Scan Flight would set this automatically to the maximum, but if using another app be sure to use full resolution for best results.

 

View solution in original post

6 Replies
GordonSumerling
Esri Regular Contributor

Hellos @st_123 I'm afraid we will need to understand a little more about your imagery your attempting to process. Can you please consider the following

1) What is the ground resolution of the source imagery

2) What overlap do you have in the source images.

3) What does the output Orthomosaic look like resolution wise

4) What is the source drone/camera

Cheers

Gordon

0 Kudos
st_123
by
New Contributor II

Hi @GordonSumerling ,

Thanks for your response. The resolution of the orthomosaic is 1xGSD (2.76 [cm/pixel]) . The DTM resolution is 5x GSD (2.76 cm/pixel) . The camera model is Pix4Dreact :  FC6310R_8.8_4864x3648(RGB) with average ground sampling distance 2.76 cm / 1.9in. 

 

The overlap images are about some river, sites and green area ( ~1.4 km2). The overlapping image will also sink in the map and the level of DSM shown in the group around -40m.

 

 

 

 

0 Kudos
GordonSumerling
Esri Regular Contributor

Hello @st_123 

So youare using a Phantom4 RTK. The Resoltion indicates you were flying somewhere between 70 and 80m. This wouls indicate that you should get a resonable resoltion mesh model, if the image overlap is good. We normally recommend 80% in the forwards and 70% in the side. Do you have this overlap?

0 Kudos
st_123
by
New Contributor II

Thanks @GordonSumerling ,

As we are currently setting is 75 % end and side laps rate, are you recommended that we should have 80% in the forwards and 70% in the side to ensure the high resolution of the model? 

Moreover, may I know is that overlapping is the main reason that affects the resolution of my mesh model? Is there any other possible reason that will affect the resolution of the mesh model?

Moreover, I found that my mesh mode will sink in the map , the level of DSM is showing around -40m , may I know the possible reason behind this issue? 

0 Kudos
GordonSumerling
Esri Regular Contributor

@st_123 It is one recommendation to improve accurcay. I would also fly lower to increase resolution. As for the sinking below the surface. This blog article is probably relevant

0 Kudos
NicoBonnafoux
Esri Contributor

Though the initial question about improving resolution was answered by @GordonSumerling  (thank you), @st_123 reached out to us through Technical Support channels and provided additional details. It was determined that the challenges that led to this question were caused by the data capture and affected all the outputs, not just the mesh. 

I'll share our findings and the tips we provided here in case they can help others in the community: 

  • Not all photos were captured the same day. All images in a mission should be captured on the same day, and ideally close in time to ensure that environmental conditions remain similar, particularly lighting.
  • Photos show blur, and the data was not captured using Site Scan Flight. We recommend using Site Scan Flight to capture data when using a supported drone like the P4R used here because it optimizes flight speed and camera settings for mapping, which many other apps (including DJI's) do not. At the least, increase the shutter speed to ensure photos are sharp, ideally 1/640s or faster, with Shutter Priority enabled. 
  • Overlap was excessive in certain areas, which can lead to long processing time and server timeouts on a larger datasets like this one. Using Site Scan Flight would also help here, some other apps are known to incorrectly calculate overlap.
  • Flight elevation was inconsistent throughout the different data collection flights. Use Site Scan Flight with Terrain follow, this will ensure that flight height above ground remains constant, even if the pilot needs to reposition for different flights.
  • Drone/camera hardware: 
    • Ensure the lens is clean before each takeoff.
    • Takeoff from a hard surface and ideally clean takeoff pad to avoid dust pickup during takeoff.
    • In warm and humid environments, let the drone hardware acclimate to ambient temperature for about 5 minutes before taking off (avoid going from cold and dry air conditioning of a car/truck directly to outside). This can also help prevent condensation appear on the lens.
  • GCPs: make sure to use targets that can be very precisely measured in the images, such as checkerboards. Painted "X"s or visible identifiable features like corners cannot be measured as precisely. 
  • Some images were captured in cropped resolution, 4864x3648, rather that the full 5472x3648. Site Scan Flight would set this automatically to the maximum, but if using another app be sure to use full resolution for best results.