3/12/25, 9:00 PM Advanced Geographic Research: Agisoft Photoscan Crash Course (updated for version 1.1.

6)

More Create Blog Sign In

Advanced Geographic Research

Studying environmental geography with fun remote sensing tools.

Home Point Clouds Publications Software About Me

WEDNESDAY, JUNE 24, 2015 SEARCH THIS BLOG

Search
Agisoft Photoscan Crash Course (updated for version 1.1.6)
ABOUT ME
Finally, time for an update…this time to Photoscan Professional 1.1.6. In this tutorial, I am
James Dietrich
going to fly through the basics of getting a Photoscan project up and running without getting
to deep into the details. If you are interested in the details of specific tools, I will refer you View my complete profile

to the official documentation (Agisoft Photocan User Manuals).

CHECK OUT MY YOUTUBE CHANNEL
The instructions for the previous version 1.0 are still here… James Dietrich
I also have discussions on cameras and camera geometry. YouTube 999+

POPULAR POSTS
Jump to a section:
Agisoft Photoscan Crash Course (updated
Photo Prep | Geotagging | Photo Alignment | Georeferencing | Optimize the alignment for version 1.1.6)
Building Dense Point Cloud | Point Cloud Editing | Aligning the Bounding Box | Build Mesh
Build Texture | Exporting Point Clouds | Exporting DEMs / Orthophotos Converting DEMs to STL files for 3D
printing

Filtering LiDAR data by height above

Photo Prep ground
Eliminate any blurry, out of focus, or random photos from your photoset
New Off Nadir Resolution Calculator
I like to do this step by hand, just to ensure that everything looks OK
Photoscan also has a tool, “Estimate Image Quality”, which can take some of the
guesswork out of large datasets. BLOG ARCHIVE

Load your photos, select all of them, right-click on one and choose ► 2017 (2)
“Estimate Image Quality”. ► 2016 (3)
Switch to the ‘Detail’ view in the Photo pane to see the quality score ▼ 2015 (7)
from 0 to 1. I have found that images with scores less than 0.6 are
► December (1)
good candidates for removal (the documentation says less than 0.5).
► August (1)
Removing photos is
a bit of a double- ► July (1)
edged sword. If you ▼ June (1)
do not remove poor Agisoft Photoscan Crash Course
photos, you risk (updated for versio...
getting incorrect
alignments and by ► May (2)
removing photos, Getting the Details view ► March (1)
you risk not getting
a complete alignment. The hope is that your photoset has sufficient ► 2014 (12)

overlap to mitigate the effects of a few missing photos. ► 2013 (10)

► 2012 (6)

Geotagging LABELS
If you have geotagged photos (from a camera with automatic geotagging or from a UAV/UAS)
SfM 3D software cameras LiDAR
If you have a synced GPS track for you photos, use a program like GeoSetter to GIS UAV structure-from-motion ArcGIS CHDK
geotag your photos before processing. Inspire 1 Kinect Photoscan conference
With geotagged photos, the coordinates are usually Lat/Long (in degrees); use the projects remote sensing 3D Printing Google Earth
Python Raspberry Pi Rivers change detection
convert tool in the “Reference” pane to convert to UTM (or your coordinate

https://adv-geo-research.blogspot.com/2015/06/photoscan-crash-course-v1-1.html 1/13
3/12/25, 9:00 PM Advanced Geographic Research: Agisoft Photoscan Crash Course (updated for version 1.1.6)
system of choice). helikite making publications scale site updates
tutorial
Never use geotagged photos as the sole source of
georeferencing information. The error in consumer
grade GPS units (handheld, in-camera, or UAS) is not เลือกภาษา
sufficient for anything more than helping with photo ขับเคลื่อนโดย แปลภาษา
alignment.
Coordinate conversion in Photoscan

Photo Alignment
Research Gate Profile
This step generates the initial alignment parameters for the photos and a sparse point cloud
Load your photos
Use the ‘Add Photos’ button on the top of the
Workspace pane, through the menu option in the
‘Workflow’ menu, or by right-clicking in the Photos
pane
Align your Photos Add photos

On the top menu bar: Workflow > Align Photos

Alignment parameters:
Accuracy:
High: is best for smaller photosets
Medium: is best for larger photosets
Pair Selection
Disabled: default, works for most datasets
Generic: I use this one if I am having trouble getting photos to align
Reference: If your photos are geotagged, this will use the GPS positions
to speed up processing.
Advanced:
Key Point Limit: 40,000 is default. Smaller numbers of key
points can speed up processing, but you run the risk of not
having enough points to align photos.
Tie Point Limit: 1000 – 2000 is recommended. This is one
parameter to play with if you are having alignment issues.
You can even take it down to zero to have no limit.
Constrain features by mask: If you have created masks on
your photos, use this option to limit the point search to just
the areas you selected.

Alignment options

Fixing misaligned/unaligned photos

After the initial pass of photo alignment, there may be photos that are misaligned
or unaligned. You can force the program to try again, usually this works (but not
always).
Select all of the photos that are causing trouble, right-click and select
“Reset Alignment”
Right-click again, select “Align Selected Cameras”

https://adv-geo-research.blogspot.com/2015/06/photoscan-crash-course-v1-1.html 2/13
3/12/25, 9:00 PM Advanced Geographic Research: Agisoft Photoscan Crash Course (updated for version 1.1.6)
If that does not work,
select a smaller batch of
problem photos and try
again working in small
batches.
A last ditch solution for problem photos is to add temporary markers in
the images to help force Photoscan to match the misaligned photos
You will need start with least two aligned images that
overlap with the misaligned photo(s)
Open the aligned photos and find features that are in both
images and right-click on the aligned image and select
‘Create Marker’ to place a temporary marker in the image.
Do that for 3-4 features.
Once you have created a marker in one image,
switch to the second aligned image and use the
‘Place Marker’ option in the right-click menu.
Open the misaligned image and place the markers on the features.
Then you can select the misaligned photo(s) and run ‘Align Selected Cameras’
again.
You can continue the process for all the misaligned images
** If the markers do not seem to be helping (one-by-one), you can try to realign all
the images. Either by selecting all your photos and selecting “Reset Alignment”
then “Align Selected Cameras” OR by running the Align Photos tool again. MAKE A
COPY OF YOUR PROJECT BEFORE YOU DO ANYTING DRASTIC LIKE THIS.
If none of that works, there is not a whole lot you can do but disable or exclude
the photos that will not align.

Georeferencing
Prep and import your ground control point (GCP) data
Double-check your projection and vertical datum.
Photoscan has a converter, however I like to do any conversions outside
of Photoscan in
a fun little program called VDatum from NOAA/NGS
Load your data points into Excel and delete all but the NAME,
LAT(Northing), LONG(Easting), and ELEV fields to simplify things.
Save as a text file (tab delimited)

GCP format in plain text file

Use the Import tool in the Reference pane,

Set the coordinate system,
delimiter, and choose which
columns go to which values
Double-check the preview pane at
the bottom of the window to make
sure your coordinate are going to Importing GCPs
the right place (be careful…some
GPS outputs are Y,X,Z instead of X,Y,Z).
It will complain that it “Can’t find match…” , click “Yes to
all” to import all of the points in the file
If you import a larger file, you should delete any points that
are not present in the photoset.
If you used geotagged photos select all of the photos in the Cameras section of the Reference
pane and click the checkbox to “uncheck” all of the geotagged photos.

https://adv-geo-research.blogspot.com/2015/06/photoscan-crash-course-v1-1.html 3/13
3/12/25, 9:00 PM Advanced Geographic Research: Agisoft Photoscan Crash Course (updated for version 1.1.6)
If you do not do this, Photoscan will try to use both the geotagged images and the
in photo GCPs to georeference. This can lead to some interesting results as the
software tries to reconcile the less accurate GPS data with coordinates that are
more accurate.

Unchecking geotagged photos

Placing the GCP markers

Find the first photo with a ground control point
Double-click the photo to open it, zoom in on the control point, right-click on the
center of the control point marker, select “Place Marker” and select the
appropriate marker.

GCP marker placement

Find the next photo with the same GCP as the first, set the marker.
Once you have a marker placed in two photos the software starts to
guess where they should be in other photos.
Right-click on the marker you have been working on, select “Filter
Photos by Marker”. This will show just the photos that the marker is in.
Go through all of the photos and place the markers.
** You can easily get to the next/previous photo using the
PageUp/PageDown keys on your keyboard
If a photo has a marker flag in it, but the GCP is not visible (obscured):
You can leave it as a gray flag and it will not be used, or
You can right-click on it and “Remove Marker”
In the photos pane, select “Reset Filter” to get back to the full photoset.

Repeat the above steps for the next two GCPs

After you have three GCP markers set, you can “Update” the
georeferencing in the ground control pane

Updating the georeferencing

This will roughly georeference the model and make finding the
remaining GCPs easier
If your first three points are close together, this may not
work very well. In this case you will likely have to manually
find and mark an additional GCP (preferably further away
from the initial GCPs
Right-click on the next GCP in the list and “Filter Photos by Marker”
Run through all of the photos and place the markers
“Update” again to refine the georeferencing

https://adv-geo-research.blogspot.com/2015/06/photoscan-crash-course-v1-1.html 4/13
3/12/25, 9:00 PM Advanced Geographic Research: Agisoft Photoscan Crash Course (updated for version 1.1.6)
Repeat this for the rest of the GCPs
Double-check that all of your GCP markers are placed
Right-click on each of the GCPs and “Filter Photos by
Marker”
Double-check that all of the photos have little blue flags
above them in the photo pane.
Update one last time

Optimize the alignment

Click on “Optimize” in the ground control pane
Leave the default options
This optimizes the camera alignment and generates a camera calibration based on
the GCP locations

Check your errors for each GCP

You may need to expand the reference pane to show the error statistics for the
GCPs
You can also click on “View Errors” to view the X, Y, and Z components
of the error
If you have points with errors that are higher than you would like, you
can uncheck the box on the left to exclude it from the georeferencing
calculations
Update the georeferencing, repeat for any other bad points if
necessary
Re-optimizing in between updates will also help you improve the error
A note on the error values
The errors reported by Photoscan are the ‘root sum of squares errors’ and the
‘root mean squared errors’ for each point.
*** These values are only valid for the GCPs and should not be reported as the
accuracy of the entire model. To judge the accuracy of the model I highly
recommend that you collect independent control points that can be used to
compute the overall spatial accuracy of the model. Please refer to the ASPRS
POSITIONAL ACCURACY STANDARDS FOR DIGITAL GEOSPATIAL DATA for details
on accuracy assessments.

Explanation of the different error values in Photoscan

Building the Dense Point Cloud

https://adv-geo-research.blogspot.com/2015/06/photoscan-crash-course-v1-1.html 5/13
3/12/25, 9:00 PM Advanced Geographic Research: Agisoft Photoscan Crash Course (updated for version 1.1.6)
Use the “Resize Region” and “Rotate Region” tools on the toolbar if the bounding
box looks like it is going to cut off any parts of the model
Resize and Rotate Region tools

Settings
Quality:
High: great for smaller photosets
Medium: good for larger datasets
Lowest, Low: produce low quality results (not usually the goal)
Ultra High: I have only ever tried this once and my computer run out of
memory.
This is the super memory intensive part of the process,
photosets that are large (lots of photos) can take a long
time to complete. Too many photos cause the program to
run out of memory. The quality that is achievable and the
time it takes depend on your processor (CPU), RAM, and
video card (GPU).
Advanced
Depth Filtering: Aggressive (default), limits points that are too far from
the surface. For some vegetation applications, this may need to be
relaxed a bit.

Dense reconstruction options

Point Cloud Editing

If there are any points in the dense point cloud are outside of the
envelope you wish to export, you can manually delete them
Point selection tools
Use the “Rectangle Selection”, “Circle Selection”, or “Free-from
Selection” tools on the toolbar to select and delete errant points.
You can hold the Control key on our keyboard to add to a selection OR
hold the Shift key to subtract from a selection
Point Cloud Classification
Photoscan has a tool to automatically classify the point
cloud into “ground” and “non-ground” classes (similar to
lidar data). You can also select points and manually assign a
class.
Tools menu > Dense Cloud

Aligning the Bounding Box

For geography applications, a height field mesh would normally be modeled

orthogonal to the Z-axis of the coordinate system. In Photoscan, the height field is
actually modeled against the orientation of the bounding box (the light gray box
with a red bottom) of the model. You can run a Python script (**Align Bounding
Box Tool**) in the console or from Tools > Run Script… to align the bounding box to
the reference coordinate system.

https://adv-geo-research.blogspot.com/2015/06/photoscan-crash-course-v1-1.html 6/13
3/12/25, 9:00 PM Advanced Geographic Research: Agisoft Photoscan Crash Course (updated for version 1.1.6)
Download the Script here https://github.com/geojames/photoscan
There are scripts for Photoscan version 1.1 (PS110…) and version 0.9
(PS090…)
There are some other fun scripts there too.

Build Mesh
Mesh Settings
Surface Type:
Height Field: only builds a mesh surface orthogonal to the reference
system Z-axis (i.e. no undercuts will be modeled). This is the fastest
mesh procedure and excellent if DEMs are going to be your final
product.
Arbitrary: Will model every nook and cranny of the object. Best for
oblique imagery, when point clouds or complex mesh surfaces are going
to be the final product.
Source Data:
Dense cloud: default
Sparse cloud: I am not sure why you would use the sparse cloud, but
you could…
Polygon Count: The number of polygons that the model will be decimated to after
processing.
The software has High, Medium, Low values
Or you can specify a custom number (entering 0 will not decimate the
final model)
Interpolation
Enabled: default, will interpolate over small holes in the surface
Disables: no interpolation will be done
Extrapolation: will interpolate the surface and then extrapolate out to
fill the bounding box.
Point Classes:
If you used it to classify the dense point cloud, you could
specify which point classes you want use for the mesh.

Build Mesh options

Build Texture
This step is not totally necessary, unless you will be exporting the 3D model (as a PLY or OBJ)
to a 3D graphics program. I will give the settings I use the most here; there are tons of other
options. If you are interested check out the user manual.

Settings:
Mapping Mode:

https://adv-geo-research.blogspot.com/2015/06/photoscan-crash-course-v1-1.html 7/13
3/12/25, 9:00 PM Advanced Geographic Research: Agisoft Photoscan Crash Course (updated for version 1.1.6)
Adaptive Orthophoto: I recommend this mode for most 3D models that
you will be exporting. It gives the best results for Arbitrary meshes.
Blending Mode:
Mosaic (Default): This will create a mosaic from the input
photos
Texture size/count:
This determine the resolution of the texture
map. Higher means better resolution, but also
increased file size.
The default is 2046 and I usually double that to
4092.
Enable Color Correction:
I usually leave this unchecked.

Build Texture options

Exporting Point Clouds

From the File menu, you can export the sparse or dense point clouds.
I recommend that you export in the LAS format, keep the coordinate system that
you georeferenced everything in, and keep the default export parameters.
For super big point clouds you can enable the ‘Split in Blocks’ option to
tile the point cloud.
The LAS files can be viewed in software like CloudCompare or
FugroViewer
I would not recommend trying to load a dense point cloud
into a program like ArcMap (it really doesn’t like dealing
with datasets like this)

Exporting DEM / Orthophotos

File > Export DEM > Export TIFF/BIL/XYZ
Choose your projection
Crop invalid DEM: checked by default
No-data value: Use -9999 for ArcGIS
Pixel Size: Photoscan estimates an appropriate pixel size, can be changed to
match your requirements
Split in blocks: segment the DEM into X by X rasters (only needed for large
datasets)
Set boundaries:
Check the box and click “Estimate”, this limits extra no-data values on
the edges
Write world file: not necessary if you export a TIFF
Choose your output location, file name, and type

File > Export Orthophoto > Export JPEG/TIFF/PNG

https://adv-geo-research.blogspot.com/2015/06/photoscan-crash-course-v1-1.html 8/13
3/12/25, 9:00 PM Advanced Geographic Research: Agisoft Photoscan Crash Course (updated for version 1.1.6)
Choose your projection
Blending mode: Mosaic (default)
Enable Color Correction: not usually necessary, but if you need the color to be
evened out feel free to select this option
Pixel Size: same as above
Split in Blocks: same as above
Set boundaries: same as above
Write world file: not necessary if you export a TIFF
Choose your output location, file name, and type

Finally all done!!!

If you are taking your orthophoto outputs into ArcMap (or any other GIS/RS software) there
are some settings that Arc will default to that will make the orthophoto way to bright.

Open the Properties of the image, switch to the Symbology tab

Select 'None' for the stretch type
Uncheck 'Apply Gamma Stretch'

Posted by James Dietrich at 12:04 PM

Labels: 3D, Photoscan, SfM, software, tutorial

31 comments:

"M" September 30, 2015 at 2:17 PM

Great article. Thx for sharing.

Reply

Unknown October 3, 2015 at 5:20 PM

Best info I've found on the settings. Thanks, this really helped me a lot :)
Reply

Unknown November 25, 2015 at 11:42 PM

One of the best informative article i have found ever for the photoscan.
Reply

Unknown March 1, 2016 at 1:21 AM

It's a very helpful website! I'm very pleased to read this article!
Reply

Unknown June 2, 2016 at 6:02 PM

I'd like to thank you as well! Viewing the Agisoft's web site, tutorials, and help files was not
getting me the aid I needed in referencing my GCPs. Your article walked me through it much
better than theirs. Thanks again.

Reply

Unknown June 12, 2016 at 4:45 PM

Been Playing with PhotoScan for a while now.... I compliment you on a very Nice tutorial...
Reply

jay June 23, 2016 at 6:29 AM

https://adv-geo-research.blogspot.com/2015/06/photoscan-crash-course-v1-1.html 9/13