ODM for a Mac user

Hello all. I pilot a DJI Phantom 4 Pro and when making 2d photomosaic maps (rankin from 40 to 100 individual images), I have been using a program that is not ideally suited for this, PTGui. It is made for stitching panos. Fudging the settings often gets me very close, but I heard about the ODM software for Mac and wanted to try it. After browsing the web site for a bit, I don’t see an option to try before buying. I want to make sure that the program can stitch my images precisely (a la DroneDeploy, which I don’t believe does any free processing any longer). Can any of you tell me if ODM is indeed a suitable app to run locally on my Mac ( plenty of ram and storage) for making 2d maps and is it limited in how large a file it can produce? for this purpose and is there indeed a try it before you buy it option that I’m not seeing? I am not a programmer or engineer but am computer literate so I would need something that does not require terminal or other input that a lay user could not accomplish. Thanks very much for any help someone can be. ==========



I’ve successfully processed about 150 photos from my phantom 3 advanced, on a macbookpro (2011) with 8gb of ram and on macosx high sierra and the docker installation of webodm.
More ram speeds up the process / processes more photos.

Great flights!!

Carlos, thanks for the quick response. Much appreciated. I’ve got 32GB of RAM on a new MB Pro, so I think I’m covered there. I am mostly interested in if the program needs much help stitching in terms of manual adjustments of the images or if it does a seamless job. I had Drone Deploy process a couple freebies for me back when they did that and the PTGui pano program I purchased did okay but it was hit and miss since it wasn’t created for this purpose. Do you have an example of one that you created using ODM? I’m inclined to purchase it and give it a go if it’s easy to use and the stitching is not obvious. Thanks again very much.


Hey Zach, check out a few examples here: https://demo.webodm.org/

ODM does not need help in stitching, it works mostly like DroneDeploy.

Thanks Piero. I purchased a copy of the program for Mac ad after downloading the three files that come as part of the download and placing the WebODM_Setup folder in the Applications folder. I started the instal process. Docker loaded just fine, but I did to create a name and password for it yet as I do not do command line work. When clicking the WebODM manager application (91MB), it offers me a dialogue box asking if I tried to launch it from there downloads folder. I did not, but nonetheless, as you can see from this screen capture, the downloading of files doesn’t appear to be happening. I know it can take a while ( I have very fast internet) but I just get the spinning downloading icon at the bottom right of the screen grab and nothing shows up in the box for the last 15 minutes. Any idea what is keeping WebODM from loading all the files and being ready to use? Thanks again for the help.

Mm, do you have any antivirus/firewall product installed on the computer? If so, does it work if you disable them?

Also, could you share the version of MacOS you’re on?

Follow up. Rebooted Mac, made sure the Firewall protection was turned off (It stays on all the time and all other apps run fine) and now I’m seeing this error.

1 Like

Do you get the same error message if you move the WebODM Manager app to your Home folder’s Applications directory (or even just to your home folder)?

I repeated the process again, with the Mac’s built in firewall turned off and it installed just fine. Unsure why it didn’t work the first time I had the firewall off but this did it, thanks. I am running 10.15.2 Catalina by the way. Can you tell me if the Docker application is required to be loaded to make maps mosaics or is it just for command-line capable folks and developers? Now that I’ve got this running, I’m off to make some sample images that have the proper amount of overlap (65%). I appreciate your help.

1 Like

Glad to hear! Docker is required for everyone.

Well, I think I’m getting closer. I went out and shot some fresh phantom for images and uploaded them to the program 27 minutes ago. There are 99 of them at about 8 MB apiece so that’s just under 900 MB. First I thought the program had stalled but at the 27 minute mark it appears to be about 1/3 of the way done. Will know shortly what comes out of the wood chipper! Do you know if there is a member of the forum, if not yourself, you could visit with me about such things as where the computing is actually happening and the Laymans terms for how the images are being stitched together? I write occasional drone articles for AOPA, and they are interested in the success I have with this application. If I write them up a little article, I will need to visit with someone knowledgeable about how it works, again in layman‘s terms. I don’t want to keep bugging you just because you were good enough to help me get this installed. Perhaps there is a moderator or person within the form who I could visit with in terms of material to help me better understand what is going on under the hood. Looks like the bar is still progressing but looking at the timer it will be another 35 minutes at least for a total of over an hour before it spit something out. Does this sound typical? Thanks again,

A few hours are typical for processing a dataset of that size, on consumer laptops. Processing happens on your computer. We have a book at https://odmbook.com that has a full chapter explaining in simple terms how processing works in details (well, I hope it does at least. Disclaimer: I wrote it).

1 Like

Thanks, I’ll get there book for sure in the morning. In the meantime, here are some side by side comparisons from the resulting map that formed after I first wrote (took about 45 minutes) and the same set of 99 images processed through PTGui, a program I purchased some time ago for use in making panos. The attached combo image has a scaled down, annotated WebODM versus PTGui, and they are very close in many regards. The two items I highlighted in red on the ODM image might be easy to solve for if the program allows for it. You’ll see the tops of some trees by the edge of the walking trail look crisp on the right image, but almost like shadows or blotchiness on the left one. I suspect some sort of stitching or seam-drawing computation flaw. Also, on both images, the actual metal poles that make up the soccer goals sometimes don’t join in a straight line. I’ve determined how to adjust the stitching seams in PTGui to solve for this, but I don’t know if that is possible in WebODM. I’m going to sample-down the 99 images and put them in a dropbox link here in case you need them for analysis and have time to offer some guidance. In the meantime, I will purchase the book and start reading. As mentioned earlier, PTGui is not intended to deal with images except those shot from only one point, but you can fudge it and make a respectable map. It seems WebODM is actually designed to do this.

Piero, I just purchased your book and have it on the iPad now. It looks great in terms of being written at my level and I’ll be wading into it today. Thanks for the tip.


Hello all. Zach in Austin, here. An experienced drone pilot but very much a newbie when it comes to using Mapping software. You’ll see a few other posts from me looking for basic, getting started help. Piero got me out of the gate and after installing WebODM on my iMac and then today, my newer, more capable MacBook Pro, I have some things to share. And of course, I have lots of questions. My attempts to produce the “high resolution” orthophoto failed on the iMac and then at first, on the laptop. However, I threw 16GB or ram (the laptop has 32GB, and an 80core processor)) at the “restart” and 46 minutes later I had a respectable 2D map. You’ll see however in there attached photo, that when enlarged, the lines of the metal soccer goal frame don’t either line up exactly straight, or the fabric netting of the goal is distorted. I’d appreciate anyone’s suggestions for tweaking the adjustments to solve for this, if it can be done. I’ll attach a side by side photo showing both WebODM and another program PT Gui, both having difficulties with the soccer goals. I’ll also attach a screen grab of the project dashboard and an overall of the finished product. Also, a screen grab from DJI GSP which flew the route for me. ODM does very well with the images overall, but I need to see if I can sweat the alignment details that would be seen on enlargements. Wondering if perhaps I don’t have enough “features” selected beyond the 8000. Please throw anything a newcomer can learn at me, and thanks in advance.

One parameter change will make a big difference: set camera-lens to brown. It should look much better using the Brown-Conrady model.


1 Like

Thanks for the suggestion smathermather. I tried it and actually had new artifacts or misalignments introduced on top of the other. I may have too many variables at work, but I have time to experiment, so please keep the suggestions coming. One thing I did do was increase the min-number of features to 16000 rather than the 8000 and the image looked better after that. There are still some soccer goal issues present, and for the first time, I’m seeing that my image on the ground of me with the drone controller is represented twice, as I walked back and forth and the drone passed overhead on one of its parallel runs. I’m not sure how to deal with that either. Screengrab attached.

I do have several questions about the interface itself, and now that I’ve read Piero’s book 2-3 times, I realize the options have changed a bit since he wrote that useful guide. As he mentioned in the foreword, it’s tough to keep up in print with a product that is undergoing rapid change. Here are my questions though, for anyone who knows the software well enough.

  1. If you choose an option from the pop-up menu when configuring a project, such as “High Resolution”, or Default, or any other… but want to change one or two of the other many optional parameters (like “use-EXIF” or “min-num features”, does this retain all the other parameters of the menu choice, or does it toss them all out and expect you set every one from scratch? Since making any edits changes the task label changes to “Custom”, I can’t be sure what all I am effecting by choosing a pre-set but wanting to make a change or two.

  2. Where do you actually tell the program what final output resolution you want, in terms of xcm/pixel? If my drone shot at 1.3 cm/px, then I want to ensure that I obtain that kind of output resolution. I think that field might have had a name change from the time there book was written.

  3. I was assuming that on “Default” setting, the program was taking the EXIF gps info from the drone images as a matter of routine, but then I saw the “Use-EXIF” choice, which is NOT enabled by default. Isn’t that where the gps info is stored in a Phantom-produced flight and do you have to tell the WebODM to go find it? Actually enabling that on one of my recent attempts might have actually made things worse. Perhaps it was that setting, and not the Brown-Conrady model choice that made the last attempt look worse.