Monday, November 16, 2026

Production

I have learned in recent weeks there is a significant difference between my Mavic 3 Pro drone and the Mavic 3 Enterprise. The enterprise edition comes with a different controller that can do automated mission planning, among other things. The enterprise model also supports an RTK module which provides very precise positioning needed by engineers and surveyors. The RTK module is an add-on, $890 from the reputable dealers and less on eBay. That additional cost is in addition to the $1,700 price difference between the similarly-named drone models.

But what about using the Pro version for a roof inspection? Centimeter accuracy isn't needed just to get pictures of a roof, right? One of the job aggregators that I patronize keeps sending me such missions, so I took one and flew it a couple of days ago. It was a small office building with a gently-sloping roof 20-23 feet above the ground. I downloaded the KML map from the company, imported it into web-based Litchi Hub, set the parameters for ortho and oblique missions, and exported the two KMZ files. (Ortho is the camera pointing straight down; oblique is at a downward angle of 45 degrees.) Then I followed the laborious procedure of:

  • Setting up two dummy waypoint missions on the Mavic controller.
  • Using Windows Explorer, copying the KMZ files to the controller's SD card.
  • Using the controller, copying the KMZ files from the SD card to the folders on the controller's internal storage containing the dummy missions. This couldn't be done in Windows Explorer because it does not see the file structure of the internal storage.
  • Copying the names of the dummy files, then deleting them.
  • Renaming the new files with the names of the deleted dummy files.

With an enterprise drone and controller with integrated missions planning, you avoid all of this dummy file nonsense.

But despite being cumbersome, it works. I also uploaded the missions to my new Mini Pro 5, which is a somewhat easier process. It combines the final four steps into a few clicks. I flew the missions first with the Mini, but quickly realized that it does not have the processing power of the Mavic and couldn't take the photos as fast as was needed. Despite being easier to program, the Mini will not be my mapping drone. It did come in handy when I wanted to measure the height of a tower that was very close to the flight path. Without using any Mavic battery, I was able to determine that the top was about 10 feet below the planned oblique flight (58 feet) and well below the ortho flight (115 feet). Doing initial on-site scouting and saving the Mavic's battery may be the Mini's role.

I flew both missions with the Mavic at a global speed of only 2 mph, but found that the drone accelerated to 33 mph when going from one leg to another, throwing it out of alignment for the first couple shots of the next leg. I reshot the ortho mission after slowing travel between all 28 waypoints down to the global speed and completed that with no problem. I had used up two batteries and was down to my third and last. I found a power outlet on the building and plugged in the charger, just in case.

I reshot the oblique mission at the slower global speed and ran out of battery power. Compared to the ortho mission, it had a much longer distance to travel and 124 waypoints. I thought I had enough to just make it, but I came up two waypoints short. With the meter showing 7%, 2.5 minutes and flashing red, the drone said it was going to land "immediately" and I had to quickly steer it from over the roof to a safe landing area. It wasn't quite Neil Armstrong avoiding a lunar boulder field, but it was more exciting than I had planned. By then the other battery was recharged to 93%, and I was able to reshoot the mission in its entirety with no problem at a global speed of 4.7 mph. Litchi Hub does suggest a drone speed based on all the parameters you feed it, and its suggestions were 7.5 for the oblique and over 10 for the ortho. I had lower numbers in there because I needed them for the Mini, and that got me into trouble.

Another fun fact I found when researching differences is that the Enterprise model has a mechanical shutter, which is preferable for missions like this versus the Pro drone's electronic shutter. The Pro was able to snap decent images at 4.7 mph without detectable smearing that might have been evident at a faster flying speed or slower shutter speed. Litchi Hub suggests a speed based on all the inputs, 7.5-10 mph depending on the different parameters, and I'll have to see if those are usable for the Mavic.

The point of this excruciating detail is that the Mavic 3 Pro is not optimal for automated roof inspection flights, and it took me a while to figure out some of the settings, but it can be done. The job aggregator accepted my files, and there are two more mapping jobs I can accept from them. This aggregator is weird in that they put a job out at a low price, then raise it every day or so for the next few weeks until someone takes it or they reach their limit and give up. I figure I'm a week away from taking the two jobs, if they are still available.

Perhaps the best part about taking these missions is I don't have to merge the images. Someone else does all of that. I did a quick merge of the ortho images in Photoshop and it did a truly awful job. So I will show just one unmerged, cropped image from the oblique mission that may be revealing a couple of problem areas on the roof.

Here's that same section of roof from the ortho image. It was shot from a higher altitude, thus the smaller apparent size.

Conclusions:

  • It is possible to do automated ortho and oblique waypoint missions without spending extra for the Enterprise drone/controller, but the cheap way has its challenges and limitations.
  • There is a point where greater precision of the Enterprise software and hardware are needed. I'm not ready to take that plunge because my volume of work does not justify it.
  • I did take the plunge of ordering a fourth Mavic battery. Having the drone demanding to land immediately helped drive that decision. It's not a light decision because these batteries cost more than $150 from reputable dealers.
  • Litchi Hub (free) allows you to split a mission into separate KMZ files if it is too big to fly on one battery. I could have split the oblique mission into two and life would have been less tense.

Friday, November 6, 2026

More roofs

First, I want to offer a mini-review of the DJI Mini 5 Pro drone. Except for the camera and image quality (which is a big deal), I believe it has all of the important capabilities of my Mavic 3 Pro, and it is easier to export Litchi KMZ missions to the Mini than the Mavic. I'm happy I got the 52-minute Plus batteries instead of the 36-minute regular batteries. It increases efficiency not to have to change batteries as often. And my spouse has commented that the little drone is not as noisy as the big one. On the down side, I can tell the Mini is smaller when it is in the air because it gets blown around by the wind more than the Mavic.

The assumption with DJI's "Mini" designation is the drone is under the 250 gram FAA limit and does not require registration. That was the case with the Mini 4K I passed on my grand nephew, just 242 grams. Even with the regular battery, I have heard that the Mini 5 Pro is above 250 grams. With the big battery installed, my kitchen scale said my new Mini is a hefty 308 grams. (BTW, for comparison, the Mavic weights 944 grams.) Since I am a certified Part 107 pilot and I was planning to register the Mini anyway as a backup to the Mavic, that is not a big deal to me. For mere mortals, it might be perceived as a problem. But c'mon man, registration only costs $5 and you can fly for 16 minutes longer. In summary, the Mini 5 Pro is a much more capable backup drone than the Mini 4K was, and I am happy with it so far.

Today I was offered a mission to shoot a commercial roof inspection. But it is hundreds of miles away in North Dakota and the money offered is about 1/3rd of what makes sense for me. Based on previous experience with this company, they increase the offer every day or two, so maybe in a week they will be where I need to be. There's the risk that the company could find someone else, or just decide to cancel it, but the current amount is just not worth it.

With the outside possibility of making that drive, I did some more testing of the Mini 5 Pro, inspecting the roof on our house using the parameters in the mission profile. It was a snap to lay out the mission in Litchi Hub and transfer it to the Mini.

At the end of the mission, I would just send the images to the company without stitching, but for my own amusement I merged the 15 ortho (-90 degrees) images in Photoshop. The program did something weird (again) with the south wall, with different perspectives rendered for the garage (lower left) and the rest of the house. But at least it did right on the copper-colored canopy over the front porch. (I know, we call it copper even though it is orange.)

The image posted above is reduced to about 1/10th the size of the original, so I believe the full-sized images are detailed enough for a roof inspection. But I would want to fly the mission with the Mavic instead of the Mini. It is more difficult to program, but the Mini paused a couple times during the test shoot. My guess was it was adjusting its position against the wind, which was not that strong and would be handled easily by the bigger Mavic.

I also shot an oblique mission with the gimbal at -45 degrees, and it produced images with good detail. The original image was 8192x6144 pixels, and this is an unresized crop of 1800x1200, 4.3% of the pixels available.

And then, even though the mission parameters didn't require it, I took closeups of parts of the roof. There's no way I'm ever getting up on our roof to see this up close.

I also shot the ortho mission using the Mavic. Confirming my earlier observation, loading the KMZ file to the Mavic controller was much more tedious than the three-click procedure for the Mini. I believe the Mini images would be perfectly acceptable, but a direct comparison of the image below to the composite at the top of this post leads me to conclude that the Mavic wide camera is noticeably better than the Mini's only camera. The Mavic handles shadows MUCH better. It is possible to lighten shadows in Photoshop, but the result with the Mini images had a lot of color noise. Maybe some of the difference was due to the time of day (three hours difference) and sky conditions (a few more clouds later on), but I don't believe those factors can explain it all.

It's also worth noting that even though the two drones were loaded with the same waypoints, the images were not framed the same. The Mini 24mm lens has a slightly wider field of view than the Mavic 24mm lens, but even accounting for that, the framing was a just bit different. As mentioned in the Rabbit Hole post, there is a difference between "Relative Accuracy" and "Absolute Accuracy." What we have here is Relative Accuracy, which is good enough for a roof inspection.

That's enough of drones shooting roofs. I swear, the next image posted will be of a critter caught on a trailcam. (Note: That turned out to be a lie.)

Thursday, November 5, 2026

Roofs

There is a construction project going on in town that I plant to shoot every week or so until it is done in mid-2028. A wide angle lens of 24mm inherently is more distorted at the edges than a medium lens of 70mm. The Mavic has cameras with those focal lengths, and a third at telephoto range of 166mm. This first image is an orthomosaic of 18 images shot at 70mm. Despite my theory about less distortion than a stiching of wide-angle images, Photoshop still choked on the green roof. I replaced it entirely with the roof from one of the images.

In looking at a single image shot with the wide angle, I don't know whether I really gained anything with the mosaic except making more work for myself.

This oblique image sets the context for building site, which is for a three-story condo building. This is looking northeast, with the next street over being the main drag, Broadway. Red Lodge High School is off in the distance at upper left. For some reason I find this type of photography with a flying camera to be fascinating. Maybe not as much as shooting moose with a telephoto, but still. BTW, I'm picking up the lingo. "Ortho" is pointed straight down, "oblique" is shot from the side.

Wednesday, November 4, 2026

Rabbit Hole

This originated as a blog for my trailcam images, but now it is really "Tom's MotionAndRemoteTriggeredCam Central." In addition to motion-triggered images, both trailcam and DSLR, I also post images taken with remotely-triggered devices. This includes DSLR/mirrorless cameras with wired and wireless triggers, an app-controlled telescope, and flying cameras, aka drones.

As with my terrestrial photography, I don't make nearly enough income from drones to support the fambly, but I get enough to report to the IRS. For the past two years I've had two drones, a DJI Mavic 3 Pro that I used for my contracted projects, and a DJI Mini 4K that I used for...what? Even though it weighs less than 250g and doesn't need to be registered for recreational use, I registered it with the FAA so I could use it on a project if I had to. The only time I needed it to complete a project was when I flew my Mavic into a tree. (I haven't used Sport Mode on the Mavic since then.) The Mini 4K is one step up from a toy, but it is not a big step. To me, its primary deficiency is it does not do waypoints.

I have become somewhat proficient using waypoints on the Mavic. A recent mission required some video in addition to stills. (I can't say what it was because I had to sign an NDA, which is a first.) Rather than shooting video while flying manually, it is much easier to get smooth video by programming waypoints and having the drone fly the program. For that assignment, I framed the image while flying, set a waypoint, and repeated that 5-6 times for each video. The controller shows the projected time of the flight, so I adjusted the speed of the drone to fit each video in 60-90 seconds.

One problem with doing this is I have to fly the drone on the same route twice, once to set waypoints manually and once to shoot the video automatically. Multiply this by five videos plus some stills and I was getting nervous about whether three batteries was enough. I had the Mini 4K along as a backup, but since it can't do waypoints it only would have been useful for stills. Fortunately I made it through with about half a battery remaining.

But it got me thinking about what I really need in a backup drone, and the Mini 4K is not it. I received delivery of a DJI Mini 5 Pro a few days ago, and sent the Mini 4K to my great nephew. He is in high school, and last I heard he aspires to be a pilot.

The new drone does do waypoints and is easier to program than the Mavic because it has a more advanced controller (RC2 vs RC). In order to plan missions on a desktop computer rather than a tiny controller screen, I experimented with a couple of programs and for now have settled on Litchi Hub. It produces files with a KMZ extension which I can then transfer either to the Mavic (with great effort) or the Mini (more easily). DJI could make transferring KMZ files to the various controllers much more seamless, but they choose not to. They want you to buy their more expensive enterprise drones and their planning and processing software.

As I did when I was learning about gardening and greenhouses, I binge-watched YouTube videos about drone waypoints and ended up going down the rabbit hole of orthomosaics. These are stitched images shot with the camera pointed straight down. Once you get more than a couple of acres it might be impossible to get high enough (legally) to get the entire property in frame for a top-down shot, and stiching images also avoids the distortion at the edges that a single shot would have. As I was watching the videos, I learned that there are different levels of precision with orthomosaics, and the more precise you need to be, the more it is going to cost for equipment and software. On one of the videos, Dave Young of Drone Launch Academy calls these levels: Tier 1 - Relative Accuracy, and Tier 2 - Absolute Accuracy. Dopes like me who just want to stitch together a few images are Tier 1; professional surveyors and engineers are Tier 2.

Complicating the matter, the US government has decided that DJI and other Chinese drone companies are national security risks, and will not approve new models for use in this country. Previously-approved models are still allowed, but DJI no longer sells them directly and (more importantly) does not offer the warranty program DJI Care. Even though existing models are not totally banned, they can't be used on US government or critical infrastructure projects. There are some US-made enterprise drones that are rather expensive (Skydio, Parrot, Freefly), but as far as I know there are no US-made affordable prosumer drones like the Mini 5 Pro.

The drone I received probably is a South Korea gray market item, and there is a warranty but it may be difficult to enforce it. When I crashed my Mavic two years ago and knocked the camera gimbal out of alignment, it was covered by DJI Care. All I had to do was send it in (to Texas, not China) and they sent me a different one, probably refurbished. It cost $175 in addition to the Care cost, but I shudder to think would it have cost (in time and money) if I had to get it repaired myself. So I'm going to be careful with my new drone.

Back to orthomosaics, it is possible to do them with the Mavic and the new Mini using low-cost (or free) planning and processing software such as Litchi. Litchi Hub is free, but if you get a drone that can be controlled directly by their software that it a one-time payment of about $26. The Mini 4 Pro is such a drone, but its successor does not have that capability. For processing, I experimented with several programs, including RealityScan, WebODM, and plain old Photoshop. RealityScan, believe it or not, is free software brought to us by Epic Games, the developers of the game Fortnite. I followed along with a YouTube video to process this mosaic of a trailer park I shot last year, and came up with something presentable after a 30-step process.

Results were somewhat less successful with imported video I shot a few days ago. And my third attempt with a new set of images was a total disaster, somehow resulting in a low-resolution black-and-white image. I went looking for something that didn't require a 30-step checklist and found WebODM. Operating the program is ultra-simple, but what I found is that setting up the program is not. The output was very low resolution. Supposedly there is a way to increase the resolution but the internet advice I found did not match what I was seeing on the screen. So why not just use Photoshop to stitch the images together?

You give up some of the capabilities of orthomosaic software such as the ability to produce a Digital Terrain Model, which shows elevations, but most of the time I just want a stitched image. I fiddled with the Photoshop Photomerge function and found that with buildings in the target area it is best to check the box "Geometric Distortion Correction." I processed 66 images to come up with this, and it took about an hour quietly churning away. The orange roof over our front porch did not render properly, but overall this is the best stitched image I got for the least effort. RealityScan choked on the roofs even worse, and the WebODM image was so small and fuzzy that it made no difference.

This is an 1800x1200 crop of the back of the garage, showing the detail available in the stitched image.

After a couple day of experimenting, orthomosaics were giving me a headache so I kept it simple with today's testing. For properties up to about five acres, it is easy to go up to 400 feet and take one shot. There may be less detail and more distortion than an orthomosaic, but processing takes seconds instead of hours.

Today's testing also included some videos. To avoid jittery video, it is recommended that for 30 frames per second, the shutter speed should be 1/60th. That is slow, and the drone's camera with its fixed aperture of f/1.8 can't avoid overexposing without some help. It is necessary to use a neutral density (ND) filter to cut down the light. The package I ordered from Adorama included ND filters, and I used the densest one today in the bright sunshine. The Mavic can stop down to f/11, which cuts the light almost (but not quite) enough to avoid using an ND filter. I recently got a set after struggling through that NDA mission mentioned above.

Tuesday, October 27, 2026

Gap

I haven't been out in the woods since last May, so there was some catching up to do. It was a little bit of everything, including something I don't believe I have seen before at this location in the Custer National Forest south of Red Lodge, a bighorn sheep. It's not a good photo, but it undoubtable what it is. I also mixed in a few images from over the summer from the greenhouse. Lots of moose.

As mentioned, I pulled my best cameras out of the woods in May so these images were taken with my oldest Brownings. There has been no more interference with the cameras from the dumbass criminal vigilante, so I am putting my newest Reconyx back out for the winter. I will secure it as well as I can. All it was getting this summer from this camera was deer eating lilac leaves in our back yard...through a cage.

Browning #6 had a gliche and lost the date/time. Through the sequence of moose images I was able to determine that it happened about 12:41pm on August 14. That's why the elk image has a date/time that looks different.

Back end of a mountain lion:

Transient bighorn sheep:

I count the drone as a remote camera:

Saturday, May 16, 2026

Catching up

I haven't checked the cameras out in the woods for six months due to knee surgery and chronic procrastination. If I had gone out earlier I would have seen this HUGE mountain lion from October sooner.

Three coyotes, a bobcat, a couple moose, and an evasive rabbit also made an appearance.

Unfortunately I felt the need to pull all of my good cameras out of the woods. Some dumbass vigilante thinks it is illegal to put trail cameras in national forests, WHICH IT IS NOT as long as you aren't using them to track game in real time. For wildlife observation, they are legal. But this idiot stole the SD card out of one of my cameras, and I'm not the only one he has targeted. Even though my other four cameras were not tampered with, I don't want this dumbass looking further than his usual base of operations and stealing my cards from them also. I put my two crappy Brownings out there just to maintain a presence. Unfortunately, the cable lock only secures the camera, not the SD card, so this criminal could conceivably steal those. Anyway...

Greenhouse snapshots

Two images from April at the greenhouse that I'm finally getting around to posting. These are from the crappy Coleman camera. For better or worse, I pulled my good cameras out of the woods for reasons explained in the next post, so I now have my best Browning in this spot and I am considering going all in with my best Reconyx. These images prove it still snows here in April, and if the forecast is correct we will see some snow in mid-May also.

Saturday, February 28, 2026

Impulse purchase

I needed a camera for my greenhouse, and decided to get the cheapest camera I could find on Amazon. It carries a famous brand name, Coleman, but obviously it is a Chinese no-name with the Coleman logo pasted on it. It cost $39.34. It did come with a nice mount to screw it to a wall, which I am planning on using inside the greenhouse...with a different camera (the Gardepro). The Coleman images seem especially soft on the left side, and I don't think it is just because that's the distant part of this image.

This is the latest shot of the Greenhouse interior, taken with the Gardepro. To minimize the constant glare, I'll try mounting it higher up/further to the right and point it down and back toward the middle. Basically it is a security camera. The outside camera will capture all the deer, turkeys and (maybe) moose and bears. For that, I'm planning on using my best camera, the newest Reconyx.

Unless there is a snowstorm, next week I will collect my five cameras from the woods. After I clean them and make sure all the batteries are charged, I'll put three of them back out there. That gives me seven cameras I can use in town, but there's this crappy Coleman and the always-crappy Primos, and two extremely old Brownings included in that number. Here's my preliminary plan, with the best cameras in bold:

  • Woods
    • Reconyx #7
    • Reconyx #2
    • Browning #11
  • House
    • Browning #15
    • Browning #5
    • Browning #6 aka the Melted Browning
    • Primos #4 (sometimes useful for video, not stills)
  • Greenhouse
    • Reconyx #19 (outside facing the front)
    • Coleman #20 (outside, off to one side)
    • Gardepro #14 (inside)

Monday, December 22, 2025

The Field

There is an empty 4-acre lot to the west behind our house. I suppose the owners are planning to build their retirement home someday, but until last summer they hadn't even mowed it since 2020. The HOA started sending out threatening letters telling everyone to mow their lots, otherwise it would be done for them and they would get a bill.

I used to have a camera pointed into that field, but when it was unmowed for years, the tall waving grass would give thousands of false triggers. The field was mowed a few months ago, and I finally put a camera up a few weeks ago. There have been very few false triggers. There's nothing out of the ordinary besides the usual deer and a fox. Apparently I need to level the camera, which is the old Melted Browning #6. It still has the same old problem of the color balance varying between the center (red) and the edges (blue-green).

Also included here are a few shots of deer and turkeys at the greenhouse, which is about a mile from the house, closer to town. The camera is the Gardepro close focus. I didn't use the Gardepro at all last winter because I didn't need the close focus capability, but here with limited space between the camera and the greenhouse it seems to be the right choice. Moose have been known to pass through the area, but the camera hasn't gotten any of those yet.