Oriented Imagery from Insta360 (or similar) camera

[DaleHarris3]

Hi,

I am investigating the best way to use a cheap 360 camera to create our own Street View like web map for our company. Does anyone have experience with doing this through ArcGIS Pro / AGOL. I haven't got a camera to test the workflow, can anyone supply a couple of photos so I can understand how the photos store there GPS coordinates, etc.

I have seen some other workflows, but they seem to send the viewing part to other applications. 

Any help would be appreciated.

Regards,

Dale

[ToddW_stl · Accepted solution]

@DaleHarris3 - have you seen either of these demos/tutorials? These would be a great starting point.

• Crash Course in Oriented Imagery
• How to create an ArcGIS Online map with 360° images

It would also be helpful to understand how/why/where(what apps) you'd like to use these images.  For example, a couple months ago I was trying to get oriented images into the new ArcGIS Indoors Mobile 2.0 application for users to orient themselves with office spaces prior to visiting, booking a conference room, or reserving a hotel office.  Without a camera that captures a 360 image, I achieved this using:

• ArcGIS Pro to create and publish an Oriented Imagery dataset as a service (along with sharing the mobile map package)
• an iPhone with the 360 Photo Cam app (free account creation as of April 2025 allowed free 360 equirectangular image downloads to my device)
• Survey123 to add the selected image from my device to the Oriented Imagery service, in the appropriate location (with additional required attribution to get it to work/display in the Indoors Mobile app)

I was originally given an Insta360 X2 spherical 360 image to work with, that was supposed to have GPS coords included, but I could never find it.  I also didn't look/didn't notice if it was included (or could be included) from the free 360 app mentioned above.  We're currently waiting to see if we'll get access to a better camera that may capture GPS this for us next time. 

@MorakotPilouk and @amanda_stanko provided me support as I struggled through understanding how to work with oriented and 360 images, and may have more insight on upcoming tools/releases that may streamline this type of workflow.  Let us know if you have any questions!

[amanda_stanko]

I played around with 360 Photo Cam app today for Android and built an Experience Builder example of the output with an Oriented Imagery widget. @ToddW_stl  found a great workflow!  The quality of the app output is not as crisp as an actual 360 camera, but it's cool that this is possible to build without any special hardware. 

[CameronRex1 · Accepted solution]

@DaleHarris3 - We've spent quite a bit of time building scalable Street View-style workflows using both professional systems (like the Mosaic 51) and more accessible gear like the Insta360 One X, One X2, and One RS 1-Inch. If you're looking to build your own Oriented Imagery Dataset using ArcGIS Pro and ArcGIS Online, here's a field-proven approach that might work well for you.

Camera Setup: Insta360 + Interval Mode

The Insta360 line of cameras are solid 360 cameras for this purpose. When collecting images:

• Use Interval mode to capture an image every few seconds. 
• We typically start capture from the camera (not the app) to avoid connectivity issues.
• Note that images are saved in INSP format, so you'll need to use Insta360 Studio on a desktop to convert them to JPG.

Time-based intervals can be inconsistent
If you stop, slow down, or speed up during capture, you'll end up with too few (less likely) or too many images. You'll need to plan your capture speed and interval carefully. Still, some redundancy is inevitable.

We typically handle this by filtering or thinning the image set afterward, keeping only the most relevant images for the final dataset.

GPS Logging: Use a Dedicated Logger

We strongly recommend not relying on the Insta360 app or the Insta360 GPS Remote for GPS tagging/logging -- it only works if the app remains active, which is unreliable in real-world conditions.

Instead, use a dedicated GPS logger: We've had success with:

• Bad Elf Pro + and Bad Elf GNSS Surveyor (both discontinued)
• Bad Elf Flex / Flex Mini
• Emlid Reach RS3

(But any quality GPS logger that exports a GPX track should work.)

Once your GPS track is recorded:

• Use a tool like GPicSync (or a similar tool) to synchronize your track with image timestamps and write Latitude/Longitude/Elevation into the JPGs' EXIF/XMP metadata.

Oriented Imagery in ArcGIS Pro

After geotagging your images:

• Use ArcGIS Pro's Oriented Imagery tools to create an Oriented Imagery Dataset (OID) and load the images. This generates one point per image, with some of the necessary attributes. 
• After the OID is created and loaded, you'll need to populate or adjust attributes as needed (either on a per image or dataset basis. Mostly importantly:
  • Image Heading: This won't be included automatically.
  • Use Python to calculate heading from each point to the next in sequence.
  • Optionally: Use this step to filter redundant images by spatial interval. (This can be accomplished using Python as well.)
• Generate the OID Footprint

Hosting & Publishing

To publish your OID for use in ArcGIS Online or Enterprise Portal:

• Upload your final JPGs to a static cloud host (e.g., AWS S3 or any public-accessible file host).
• Update your OID paths to reflect the cloud-hosted image locations.
• Use the Oriented Imagery toolbox to publish the dataset.
• In ArcGIS Online, create a web map with the OID (points and footprint).
• Build a viewer using Experience Builder and the Jewelry Box template with the Oriented Imagery widget.

That is the basic workflow. It is a long, multi-step and multi-software process.

Automating the Workflow

We've recently developed a custom ArcGIS Python toolbox that automates much of this workflow. It's currently optimized for the Mosaic 51, but we're actively considering adding support for Insta360 as well -- especially since much of our earlier capture work was done with various Insta360 models.

Once we finish cleanup, we plan to release the toolbox on GitHub. It won't support Insta360 out of the gate, but I'll make sure to post here when it does.

If you are interested, I also just wrote an article on what we're doing with 360 imagery capture and Oriented Imagery:

From Capture to Corridor Intelligence: How RMI Valuation Is Unlocking New Value with 360° Imagery

Here I explore our progression from early 360 cameras to the Mosaic 51, the automated Python toolbox and what we use it for.

I'll check if I have Insta360 imagery and a GPX log that I can share -- if not, I'll try to capture some soon, and share here.

Always happy to share resources or answer any questions.

Best,

Cameron Rex
RMI Valuation, LLC

[SamuelAdamson]

"Use Python to calculate heading from each point to the next in sequence"

This used to be a function within the GP tools of the original Oriented Imagery install. Is there a new version of this tool anywhere? The old GP tool no longer works since oriented imagery as been incorporated into ArcGIS Pro .

[SamuelAdamson]

import arcpy
import math

# Set workspace and feature class
arcpy.env.workspace = r"C:\Users\sdada\Documents\ArcGIS\Projects\Street_view\Street_view.gdb" # replace with FGDB containing OIC #
feature_class = "Streetview_ExportFeatures" # Replace with actual feature class name #
heading_field = "CameraHeading" #replace with camera heading field #

# Add Heading field if it doesn't exist
if heading_field not in [f.name for f in arcpy.ListFields(feature_class)]:
arcpy.AddField_management(feature_class, heading_field, "DOUBLE")

def calculate_initial_bearing(lat1, lon1, lat2, lon2):
"""Calculate the bearing between two GPS points."""
delta_lon = math.radians(lon2 - lon1)
lat1, lat2 = map(math.radians, [lat1, lat2])

x = math.sin(delta_lon) * math.cos(lat2)
y = math.cos(lat1) * math.sin(lat2) - math.sin(lat1) * math.cos(lat2) * math.cos(delta_lon)

bearing = math.degrees(math.atan2(x, y))
return (bearing + 360) % 360 # Normalize to 0-360 degrees

# Read features and compute heading
gps_data = []
with arcpy.da.SearchCursor(feature_class, ["OBJECTID", "SHAPE@XY"]) as cursor:
for row in cursor:
gps_data.append({"ID": row[0], "Lat": row[1][1], "Lon": row[1][0]})

#check camera heading field exsists#
fields = [f.name for f in arcpy.ListFields(feature_class)]
print(fields) # Check if "CameraHeading" exists

#check gps data #
print(gps_data) # Debugging check before processing

# Update heading field
updates = {gps_data[i]["ID"]: calculate_initial_bearing(
gps_data[i]["Lat"], gps_data[i]["Lon"], gps_data[i + 1]["Lat"], gps_data[i + 1]["Lon"]
) for i in range(len(gps_data) - 1)}

with arcpy.da.UpdateCursor(feature_class, ["OBJECTID", heading_field]) as cursor:
for row in cursor:
if row[0] in updates:
cursor.updateRow([row[0], updates[row[0]]]) # Assign heading

print("Heading calculation complete! Check the updated feature class.")

[DaleHarris3]

@CameronRex1 and @ToddW_stl thank you both for the amazing feedback. Cameron you have given me enough detail that I am happy to go ahead with testing. We were after a Insta360 like solution because it would be easier to teach someone from our remote sites to do the capture and update it on a regular basis.

@CameronRex1 the post on LinkedIn was an excellent read. Thanks for sharing.  

[ClarkSelby]

Here’s a 360° Photo Experience Builder widget that uses a photos-to-points approach:
View on Esri Marketplace

We use several Insta360 cameras for both indoor and outdoor projects. The photos are easy to integrate into ArcGIS Enterprise or Online, and the tool is intuitive—even for users without a GIS background.

There is also a 360 degree and standard Video Experience Builder widget in the Esri Marketplace: https://www.esri.com/en-us/arcgis-marketplace/listing/products/6cd2d165020c4b63a1e9ba5830f88c85.  He...

[amanda_stanko]

I just signed up for the Vidgeo trial this week! Looking forward to testing the 360 video app that's there.

[ClarkSelby]

Thank you - let us know if we can walk you through the functionality or help with the installation.