Regular Orthomosaic in Version 2023.1.1 ?

[EmmaSchultz1]

Hi D2M Users, 

I just downloaded the most recent version of D2M (2023.1.1) - I have an advanced license. 

I regularly process orthomosaics with this software, however, I noticed with the new version that there is now a "True Ortho" option under the 2D Template. I think I will definitely use True Orthos in the future, but my workflow only requires regular orthomosaics. No where in  D2M can I find where to check off regular orthomosaic as a product. By using the default checked off options while processing, it generates a 3 band true orthomosaic, but it does not produce a regular orthomosaic anymore from what I can see? 

In the D2M documentation, it says you are still able to produce a regular orthomosaic with the 2D template: https://doc.arcgis.com/en/drone2map/latest/help/processing-options.htm 

When checking off products in D2M however, there is nowhere you can specify a regular orthomosaic. The options are: 

- Image Collection

- True Ortho 

- DSM

- DTM 

So does the regular orthomosaic autogenerate in tandem with one of these other products? 

I require the old orthomosaic option as they are much quicker to generate and all of our organization's orthomosaics are housed in a single MDS, which we want to keep standardized. 

Thank you in advanced. I attached a couple of screenshots from my D2M options. 

Emma

[MarkBarker1 · Accepted solution]

Hi @EmmaSchultz1 ,

The orthomosaic processing option that you are mentioning is only available on 2023.1.x when you are using multispectral data. This is because we have yet to implement the True Ortho for that type of imagery. I am adding a note to the mentioned documentation to help clarify this for the next major release. 

In answer to your other question, you can still obtain the same orthomosaic output as you would in Drone2Map 2022.1 by exporting out the image collection. Steps would be as follows:

1. Within processing options under the 2D Products tab set an orthorectification method for the image collection.
   
2. Check the boxes for Color Balancing and Seamlines.
3. Run the processing. (You only need the image collection on)
4. Add the Image Collection to the map. You can use the drop-down arrow next to the product in the Manage pane to do this quickly. Check the box for “Add to Map”.
   
5. Click on the Image Collection in the Contents pane.
6. At the top of Drone2Map a new mosaic layer tab will appear, click it.
7. Use the Sort button to change the setting to Seamline.
8. Use the Resolve Overlap button to change the setting to Blend.
   
9. On the Home tab, under the Post-Processing group click Export Raster.
10. Define any settings you want and click Export.

The resulting image should be the same as 2022.1. The benefit of this workflow is more customization can be configured on your end in terms of the exported raster.

Let me know if you have any trouble with the workflow.

Regards,
Mark

[EmmaSchultz1]

Hi Mark, 

Just tried out the above workflow and it worked. I apologize if this was mentioned in the documentation somewhere. So essentially there are capabilities to orthorectify the image collection and suit the workflow to your needs. Good to know. And assuming ESRI is pushing for the True Ortho option to be used. I will see if the True Orthos will work with our data organization standards. Thank you for your help! 

Best, 

Emma 

[CodyBenkelman]

Thanks to Mark for posting the workflow.  If you're interested, we've published a blog discussing the different ways of compiling images into the orthoimage outputs - True orthos, the traditional orthomosaic, and also the dynamic mosaic (a.k.a. image collection).  See https://esriurl.com/3orthoTypesArcGIS  

[EmmaSchultz1]

Hi there, 

Just wanted to add to this thread as I have been doing some testing and want some insight in the differences in products. I have been getting some funny results. 

Background: 

Drone imagery captured with a Phantom 4 Pro. 538 image captured. Part of the project area is heavily forested and there was some wind on flight day that caused some blurry pictures over the trees. The drone also flew over water for a portion where I expect there to be holes above the water and trees. For these reasons, only about 409/538 images were calibrated when creating a True Orthomosaic. See the adjusted positions image from the processing report: 

After processing, my traditional orthomosaic did not look good at all. Even in the areas where the images calibrated fine. Here is my comparison between D2M products in addition to processing the same project in Agisoft Metashape. 

1. D2M True Orthomosaic: 

- Checked off color balancing, checked off enhance true ortho. Everything else default. I expected there to be holes over the trees, but the gravel road and infrastructure pad did not stitch together well at all. After processing, I realized the initial image scale was set to only 1/8th of the size, but I thought it was supposed to be the original size image for the 2D full template. But with changing this parameter I ended up with the same image : 

2. Traditional Orthomosaic completed with @MarkBarker1 suggested Workflow: 

- I then processed a traditional orthomosaic with the above workflow. Only thing I changed was setting the initial image scale to 1. In addition, when I exported the raster at the end I did not compress it at all. It was set to be full resolution - which gave me a 31 GB TIF in the end - crazy huge file size. But this time, the gravel road/ pad were clear, but huge holes were in other areas of the dataset: 

3. Traditional Orthomosaic, LZW Compression: 

- Next, I wanted to try to export the same traditional orthomosaic, but didn't want a 31 GB tiff. I chose to use the LZW compression when exporting and got the following output which was only 6.7 GB Tif, and was  by far the best product so far: 

4. Lastly, I wanted to compare these products against an orthmosaic created in Agisoft Metashape with using the default workflow. The end product was about a 2 GB TIF, which looked the best out of all the options: 

My questions are: 1. would some of the more blurry images have made a true orthomosaic not possible in this instance? If so, the traditional orthomosaic is the more favourable option for these types of sites. 2. How to make a traditional orthomosaic with a smaller file size? The Agisoft product was only 2 GB, while the compressed traditional orthomosaic was > 6 GB. Any suggestions to reduce the file size? 

Thank you!

[MarkBarker1]

Hi Emma, 

I am sorry to hear you are still having some reconstruction issues. Blurry images can make it harder for Drone2Map to perform photogrammetry properly. It is possible that with those heavily vegetative areas not many tie points are being generated or the ones that are generated are above the default tie point error threshold so they are getting filtered out before generating products. I would check those settings first if you haven't already https://doc.arcgis.com/en/drone2map/latest/help/processing-options.htm#GUID-1CCAC1A3-E7D3-4709-984B-.... Increasing the matching neighborhood size also can help with those areas to allow for more potential tie points and matches. I would like to get a look at the flight if you are able to share the data so I can give some more specific recommendations. Feel free to email me if you can share it and I will setup a shared location. (mbarker@esri.com)

Regarding the compression, do you know what type of compression Agisoft uses? Feel free to share the outputs you mentioned above to me as well so I can compare.

Thank you,
Mark