<liclass="toctree-l4"><aclass="reference internal"href="../installation/#step-3-check-memory-and-cpu-allocation">Step 3. Check Memory and CPU Allocation</a></li>
<liclass="toctree-l2"><aclass="reference internal"href="../installation/#basic-commands-and-troubleshooting">Basic Commands and Troubleshooting</a></li>
<liclass="toctree-l4"><aclass="reference internal"href="#prep-data-and-project">Prep data and project</a></li>
<liclass="toctree-l4"><aclass="reference internal"href="#resize-droplet-pull-pin-run-away">Resize droplet, pull pin, run away</a></li>
<liclass="toctree-l4"><aclass="reference internal"href="#after-it-finishes-assuming-you-survive-that-long">After it finishes (assuming you survive that long)</a></li>
<liclass="toctree-l3"><aclass="reference internal"href="../large/#getting-started-with-distributed-split-merge">Getting Started with Distributed Split-Merge</a></li>
<liclass="toctree-l3"><aclass="reference internal"href="../large/#understanding-the-cluster">Understanding the Cluster</a></li>
<liclass="toctree-l3"><aclass="reference internal"href="../large/#accessing-the-logs">Accessing the Logs</a></li>
<liclass="toctree-l3"><aclass="reference internal"href="../faq/#can-i-process-two-or-more-orthophoto-geotiffs-to-stitch-them-together">Can I process two or more orthophoto GeoTIFFs to stitch them together?</a></li>
<liclass="toctree-l3"><aclass="reference internal"href="../faq/#i-want-to-build-a-commercial-application-that-includes-odm-do-i-need-a-commercial-license">I want to build a commercial application that includes ODM. Do I need a commercial license?</a></li>
<liclass="toctree-l3"><aclass="reference internal"href="../faq/#are-there-other-licensing-options-aside-from-the-agplv3">Are there other licensing options aside from the AGPLv3?</a></li>
<p>Without any parameter tweaks, ODM chooses a good compromise between quality, speed and memory usage. If you want to get higher quality results, you need to tweak some parameters:</p>
<blockquote>
<div><ulclass="simple">
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--orthophoto-resolution</span></code> is the resolution of the orthophoto in cm/pixel. Decrease this value for a higher resolution result.</p></li>
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--ignore-gsd</span></code> is a flag that instructs ODM to skip certain memory and speed optimizations that directly affect the orthophoto. Using this flag will increase runtime and memory usage, but may produce sharper results.</p></li>
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--texturing-nadir-weight</span></code> should be increased to <codeclass="docutils literal notranslate"><spanclass="pre">29-32</span></code> in urban areas to reconstruct better edges of roofs. It should be decreased to <codeclass="docutils literal notranslate"><spanclass="pre">0-6</span></code> in grassy / flat areas.</p></li>
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--texturing-data-term</span></code> should be set to <cite>area</cite> in forest areas.</p></li>
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--mesh-size</span></code> should be increased to <codeclass="docutils literal notranslate"><spanclass="pre">300000-600000</span></code> and <codeclass="docutils literal notranslate"><spanclass="pre">--mesh-octree-depth</span></code> should be increased to <codeclass="docutils literal notranslate"><spanclass="pre">10-11</span></code> in urban areas to recreate better buildings / roofs.</p></li>
<h2>Calibrating the Camera<aclass="headerlink"href="#calibrating-the-camera"title="Permalink to this headline">¶</a></h2>
<p>Camera calibration is a special challenge with commodity cameras. Temperature changes, vibrations, focus, and other factors can affect the derived parameters with substantial effects on resulting data. Automatic or self calibration is possible and desirable with drone flights, but depending on the flight pattern, automatic calibration may not remove all distortion from the resulting products. James and Robson (2014) in their paper <aclass="reference external"href="https://onlinelibrary.wiley.com/doi/full/10.1002/esp.3609">Mitigating systematic error in topographic models derived from UAV and ground‐based image networks</a> address how to minimize the distortion from self-calibration.</p>
<p><em>Bowling effect on point cloud over 13,000+ image dataset collected by World Bank Tanzania over the flood prone Msimbasi Basin, Dar es Salaam, Tanzania.</em></p>
<p>To mitigate this effect, there are a few options but the simplest are as follows: fly two patterns separated by 20°, and rather than having a nadir (straight down pointing) camera, use one that tilts forward by 5°.</p>
<p>As this approach to flying can be take longer than typical flights, a pilot or team can fly a small area using the above approach. OpenDroneMap will generate a calibration file called cameras.json that then can be imported to be used to calibrate another flight that is more efficiently flown.</p>
<p>Alternatively, the following experimental method can be applied: fly with much lower overlap, but two <em>crossgrid</em> flights (sometimes called crosshatch) separated by 20° with a 5° forward facing camera.</p>
<ulclass="simple">
<li><p>Crossgrid overlap percentages can be lower than parallel flights. To get good 3D results, you will require 68% overlap and sidelap for an equivalent 83% overlap and sidelap.</p></li>
<li><p>To get good 2D and 2.5D (digital elevation model) results, you will require 42% overlap and sidelap for an equivalent 70% overlap and sidelap.</p></li>
<h2>Creating Digital Elevation Models<aclass="headerlink"href="#creating-digital-elevation-models"title="Permalink to this headline">¶</a></h2>
<p>By default ODM does not create DEMs. To create a digital terrain model, make sure to pass the <codeclass="docutils literal notranslate"><spanclass="pre">--dtm</span></code> flag. To create a digital surface model, be sure to pass the <codeclass="docutils literal notranslate"><spanclass="pre">--dsm</span></code> flag.</p>
<p>For DTM generation, a Simple Morphological Filter (smrf) is used to classify points in ground vs. non-ground and only the ground points are used. The <codeclass="docutils literal notranslate"><spanclass="pre">smrf</span></code> filter can be controlled via several parameters:</p>
<blockquote>
<div><ulclass="simple">
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--smrf-scalar</span></code> scaling value. Increase this parameter for terrains with lots of height variation.</p></li>
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--smrf-slope</span></code> slope parameter, which is a measure of “slope tolerance”. Increase this parameter for terrains with lots of height variation. Should be set to something higher than 0.1 and not higher than 1.2.</p></li>
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--smrf-threshold</span></code> elevation threshold. Set this parameter to the minimum height (in meters) that you expect non-ground objects to be.</p></li>
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--smrf-window</span></code> window radius parameter (in meters) that corresponds to the size of the largest feature (building, trees, etc.) to be removed. Should be set to a value higher than 10.</p></li>
</ul>
</div></blockquote>
<p>Changing these options can affect the result of DTMs significantly. The best source to read to understand how the parameters affect the output is to read the original paper <aclass="reference external"href="https://www.researchgate.net/publication/258333806_An_Improved_Simple_Morphological_Filter_for_the_Terrain_Classification_of_Airborne_LIDAR_Data">An improved simple morphological filter for the terrain classification of airborne LIDAR data</a> (PDF freely available).</p>
<p>Overall the <codeclass="docutils literal notranslate"><spanclass="pre">--smrf-threshold</span></code> option has the biggest impact on results.</p>
<p>SMRF is good at avoiding Type I errors (small number of ground points mistakenly classified as non-ground) but only “acceptable” at avoiding Type II errors (large number non-ground points mistakenly classified as ground). This needs to be taken in consideration when generating DTMs that are meant to be used visually, since objects mistaken for ground look like artifacts in the final DTM.</p>
<p>Two other important parameters affect DEM generation:</p>
<blockquote>
<div><ulclass="simple">
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--dem-resolution</span></code> which sets the output resolution of the DEM raster (cm/pixel)</p></li>
<li><p><codeclass="docutils literal notranslate"><spanclass="pre">--dem-gapfill-steps</span></code> which determines the number of progressive DEM layers to use. For urban scenes increasing this value to <cite>4-5</cite> can help produce better interpolation results in the areas that are left empty by the SMRF filter.</p></li>
<h3>Fieldwork planning<aclass="headerlink"href="#fieldwork-planning"title="Permalink to this headline">¶</a></h3>
<p>Weather conditions modify illumination and thus impact the photography results. Best results are obtained with evenly overcast or clear skies. Also look for low wind speeds that allow the camera to remain stable during the data collection process.
In order to avoid shadows which on one side of the stockpile can obstruct feature detection and lessen the number of resulting points, always prefer the flights during the midday, when the sun is at the nadir so everything is consistently illuminated.
Also ensure that your naked eye horizontal visibility distance is congruent with the planned flight distances for the specific project, so image quality is not adversely impacted by dust, fog, smoke, volcanic ash or pollution.</p>
<h3>Flight pattern<aclass="headerlink"href="#flight-pattern"title="Permalink to this headline">¶</a></h3>
<p>Most stockpile measurement jobs does not require a crosshatch pattern or angled gimbal as the resting angle of stockpile materials allows the camera to capture the entire stockpile sides. Only some special cases where erosion or machinery operations causes steep angles on the faces of the stockpile would benefit of the crosshatch flight pattern and angled camera gimbal but consider that these additional recognized features come at a cost, (in field labor and processing time) and the resulting improvements are sometimes negligible.</p>
<p>In most of the cases a lawn mower flight pattern is capable of producing highly accurate stockpile models.</p>
<p>Recommended overlap would be between 75% and 80% with a sidelap in the order of 65% to 70%. It is also recommended to slightly increase overlap and sidelap as the flight height is increased.</p>
<h3>Flight height<aclass="headerlink"href="#flight-height"title="Permalink to this headline">¶</a></h3>
<p>Flight height can be influenced by different camera models, but in a general way and in order to ensure a balance between image quality and flight optimization, it is recommended to be executed at heights 3 to 4 times the tallest stockpile height. So for a 10 meter stockpile, images can be captured at a height of 40 meters.
As the flight height is increased, it is also recommended to increase overlap, so for a 40 meter height flight you can set a 65% sidelap and 75% overlap, but for a planned height of 80 meters a 70% sidelap and 80% overlap allowing features to be recognized and properly processed.</p>
<h3>GCPs<aclass="headerlink"href="#gcps"title="Permalink to this headline">¶</a></h3>
<p>To achieve accuracy levels better than 3%, the use of GCP’s is advised. Typically 5 distributed GCP are sufficient to ensure accurate results.
When placing or measuring GCP, equipment accuracy should be greater than the GSD. Survey grade GNSS and total stations are intended to provide the required millimetric accuracy.</p>
<p>For further information on the use of GCPs, please refer to the <aclass="reference external"href="https://docs.opendronemap.org/gcp.html">Ground Control Points section</a>.</p>
<h3>Processing parameters<aclass="headerlink"href="#processing-parameters"title="Permalink to this headline">¶</a></h3>
<p>A highly accurate model can be achieved using WebODM high resolution predefined settings. Then you can further adjust some parameters as necessary.</p>
<p>If using ODM, these this reference values can help you configure the process settings.</p>
<h3>Measuring<aclass="headerlink"href="#measuring"title="Permalink to this headline">¶</a></h3>
<p>As almost 50% of the material will be found in the first 20% of the stockpile height, special care should be taken in adequately defining the base plane.</p>
<p>Dialog box will show the message “Computing …” for a few seconds, and after the computing is finished the volume measurement value will be displayed.</p>
<p>If you are using the command line OpenDroneMap you can use the dsm files to measure the stockpile volumes using other programs.</p>
<p>Also consider that once the limits of the stockpile are set in software like <aclass="reference external"href="https://www.qgis.org">QGis</a>, you will find there are some ways to determine the base plane. So for isolated stockpiles which boundaries are mostly visible, a linear approach can be used. While for stockpiles set in slopes or in bins, the base plane is better defined by the lowest point.
Creation of a triangulated 3D surface to define the base plane is advised for large stockpiles. This is also valid for stockpiles paced on irregular surfaces.</p>
<h2>Using Docker<aclass="headerlink"href="#using-docker"title="Permalink to this headline">¶</a></h2>
<p>Since many users employ docker to deploy OpenDroneMap, it can be useful to understand some basic commands in order to interrogate the docker instances when things go wrong, or we are curious about what is happening. Docker is a containerized environment intended, among other things, to make it easier to deploy software independent of the local environment. In this way, it is similar to virtual machines.</p>
<p>A few simple commands can make our docker experience much better.</p>
<p>This is likely to be unwieldy large, but we can use a pipe <cite>|</cite> character and other tools to extract just what we need from the logs. For example we can move through the log slowly using the <cite>more</cite> command:</p>
<p>Pressing <cite>Enter</cite> or <cite>Space</cite>, arrow keys or <cite>Page Up</cite> or <cite>Page Down</cite> keys will now help us navigate through the logs. The lower case letter <cite>Q</cite> will let us escape back to the command line.</p>
<p>We can also extract just the end of the logs using the <cite>tail</cite> commmand as follows:</p>
<h3>Command line access to instances<aclass="headerlink"href="#command-line-access-to-instances"title="Permalink to this headline">¶</a></h3>
<p>Sometimes we need to go a little deeper in our exploration of the process for OpenDroneMap. For this, we can get direct command line access to the machines. For this, we can use <cite>docker exec</cite> to execute a <cite>bash</cite> command line shell in the machine of interest as follows:</p>
<h3>Cleaning up after Docker<aclass="headerlink"href="#cleaning-up-after-docker"title="Permalink to this headline">¶</a></h3>
<p>Docker has a lamentable use of space and by default does not clean up excess data and machines when processes are complete. This can be advantageous if we need to access a process that has since terminated, but carries the burden of using increasing amounts of storage over time. Maciej Łebkowski has an <aclass="reference external"href="https://lebkowski.name/docker-volumes/">excellent overview of how to manage excess disk usage in docker</a>.</p>
<h2>Using ODM from low-bandwidth location<aclass="headerlink"href="#using-odm-from-low-bandwidth-location"title="Permalink to this headline">¶</a></h2>
<h4>Install<aclass="headerlink"href="#install"title="Permalink to this headline">¶</a></h4>
<ulclass="simple">
<li><p>Create a Digital Ocean droplet with at least 4GB of RAM. That’ll cost
about $20/month. Less than 4GB of RAM and the install will probably
fail. When we actually run the ODM process we’ll resize it to a much
larger—and more expensive—cloud machine, but between runs you can
downsize it between runs to the second-cheapest droplet which costs
only $10/month (the cheapest droplet, at $5/month, comes with such a
small drive that you can’t downsize back to it).</p>
<ul>
<li><p>Should be an Ubuntu 18.04 instance to ensure dependency
compatibility</p></li>
<li><p>Create a user with sudo privileges. <aclass="reference external"href="https://www.digitalocean.com/community/tutorials/initial-server-setup-with-ubuntu-18-04">Digital Ocean’s insanely good
documentation</a>
can help you figure this out. In our case we set up a user called
<codeclass="docutils literal notranslate"><spanclass="pre">odm</span></code>, so connecting to it is via the command
<codeclass="docutils literal notranslate"><spanclass="pre">ssh</span><spanclass="pre">odm@xxx.xxx.xxx.xxx</span></code> (where the x’s stand for the IPv4
address of your server). If you want to follow this example
closely, <em>do</em> use the username <codeclass="docutils literal notranslate"><spanclass="pre">odm</span></code>; then your install path
will be <codeclass="docutils literal notranslate"><spanclass="pre">/home/odm/ODM/</span></code> and will match all of the examples in
this document.</p></li>
<li><p>Go ahead and execute <codeclass="docutils literal notranslate"><spanclass="pre">sudo</span><spanclass="pre">apt</span><spanclass="pre">update</span></code> and <codeclass="docutils literal notranslate"><spanclass="pre">sudo</span><spanclass="pre">apt</span><spanclass="pre">upgrade</span></code> to ensure
your server isn’t dangerously without updates. Make sure to stay with
Ubuntu 18.04.</p></li>
</ul>
</li>
<li><p>Download and install ODM on it from the <aclass="reference external"href="https://github.com/OpenDroneMap/ODM">ODM
<li><p>If you do this from the default home folder of your user
(i.e. <codeclass="docutils literal notranslate"><spanclass="pre">odm</span></code>) the path to the install will be <codeclass="docutils literal notranslate"><spanclass="pre">/home/odm/ODM</span></code>
(abbreviated as <codeclass="docutils literal notranslate"><spanclass="pre">~/ODM/</span></code>).</p></li>
<li><p>There are some environmental variables that need to be set. Open the
~/.bashrc file on your machine and add the following 3 lines at the
end (From <aclass="reference external"href="https://github.com/OpenDroneMap/ODM">the ODM github</a>).
The file can be opened with <codeclass="docutils literal notranslate"><spanclass="pre">nano</span><spanclass="pre">~/.bashrc</span></code> (or whatever text
editor you use in lieu of nano). Be sure to replace <codeclass="docutils literal notranslate"><spanclass="pre">/home/odm/</span></code>
with the correct path to the location where you extracted
OpenDroneMap if you didn’t do everything exactly as in our example
(for example if you used a different username in your server setup):</p></li>
<h4>Resize droplet, pull pin, run away<aclass="headerlink"href="#resize-droplet-pull-pin-run-away"title="Permalink to this headline">¶</a></h4>
<ulclass="simple">
<li><p>Shut down and resize your machine to an appropriately monstrous
number of CPUs and amount of memory. I use the memory-optimized
machine with 24 dedicated vCPUs and 192GB of RAM (which costs about
$1.60/hr—which adds up fast, it’s over $1000/month). Restart, and get
to work quickly so as not to waste expensive big-droplet time.</p></li>
<li><p>Launch the ODM process via ssh using nohup (so that if you’re cut
off, processing will continue)</p>
<ul>
<li><p>Alternately you can use GNU screen to launch the process from a
screen session which won’t stop if your connection is interrupted;
launch <codeclass="docutils literal notranslate"><spanclass="pre">screen</span></code>, and use <codeclass="docutils literal notranslate"><spanclass="pre"><ctrl></span><spanclass="pre">a</span><spanclass="pre"><ctrl></span><spanclass="pre">d</span></code> to detach,
<codeclass="docutils literal notranslate"><spanclass="pre">screen</span><spanclass="pre">-r</span></code> to re-attach. But using screen won’t get you a log
file of all of the console output unless you do something specific
to capture that, while nohup gives you a file with all of the
console output, including error messages, for free.</p></li>
<em>not</em> to work; the <codeclass="docutils literal notranslate"><spanclass="pre">-i</span></code> or <codeclass="docutils literal notranslate"><spanclass="pre">--image</span></code> parameter causes a weird
error. So we drop the -i parameter, and rely on the project
directory line in the settings.yaml file to direct ODM to the
right place. Now using (including a split-merge):</p></li>
<li><p>This points ODM at the folder (in this example)
<codeclass="docutils literal notranslate"><spanclass="pre">/mnt/odmdata/myproject/</span></code>. Provided the image_groups.txt and
gcp_list.txt are in this folder, the images are in
<codeclass="docutils literal notranslate"><spanclass="pre">/mnt/odmdata/myproject/images/</span></code>, and the project path in
settings.yaml is <codeclass="docutils literal notranslate"><spanclass="pre">/mnt/odmdata/</span></code> it will not waste time and space
copying images.</p></li>
<li><p>Note that this assumes you have an image_groups.txt file. If not,
this <codeclass="docutils literal notranslate"><spanclass="pre">-split-overlap</span><spanclass="pre">0</span></code> will probably fuck things up, and the
<codeclass="docutils literal notranslate"><spanclass="pre">--split</span><spanclass="pre">1</span></code> is literally a random number that will be ignored after
the image_groups.txt file is loaded (I think it normally controls how
many groups it splits a set of images into, but in our case we’re
assuming the images are already grouped sensibly). If you don’t have
a large dataset (>1000 images), omit the <codeclass="docutils literal notranslate"><spanclass="pre">--split</span></code> and
<h4>After it finishes (assuming you survive that long)<aclass="headerlink"href="#after-it-finishes-assuming-you-survive-that-long"title="Permalink to this headline">¶</a></h4>
<ulclass="simple">
<li><p>As soon as processing is done, shut down the machine and resize it
back down to the inexpensive minimum capacity.</p></li>
<li><p>Start the machine back up, and log in via ssh.</p></li>
<li><p>If you want to save download bandwidth, you can compress the
orthophoto using GDAL. Don’t add overviews, do that on your local
machine to avoid making the file bigger before downloading it.</p></li>
<p><aclass="reference external"href="https://github.com/opendronemap/docs#how-to-make-your-first-contribution">Learn to edit</a> and help improve <aclass="reference external"href="https://github.com/OpenDroneMap/docs/blob/publish/source/tutorials.rst">this page</a>!</p>
