Anyone else noticed this? There was a thread quite a while back, but so much has changed since then that I thought to start fresh.
I noticed on my last couple of flights that the yaw of the drone is slightly skewed from the flight line. Sometimes it runs in that manner the entire flight as if there is a defined heading and sometimes I can see it turning.
I always thought it was due to wind. It should have no impact on your map data.
I thought about that as well, but winds have been under 10mph all week and I have never seen this before. Don’t you think it would affect overlaps in some manner? You can see the area of coverage above and how consecutive photos have less overlap and of course the adjacent flight lines will add coverage, but it’s not the same so I wonder how the machine sees it? This is a visual of a 75/75 overlap with normal orientation and the skew that I experienced. I have not calculated the exact skew and if it is consistent when it happens so this is just a visual estimate. It becomes really hard to calculate the skewed overlap areas.
Thank you for surfacing Michael. Just to be certain - we will verify with internal testing.
With 75/75 overlap outcomes should be unchanged. Only if your frontlap/sidelap have very large deltas (e.g. linear flight) could there be some impact.
Thanks Andrew! My actual configuration was 80/65 so I may be a little bit more of an issue, but I will see how the final product comes out.
I just flew and here is one of the images. I did notice some rotation from the endpoints where it seemed to be a gradual rotation with the worst skew being in the middle of the leg. This one was pretty close to the middle.
Here is what I get stitched with ICE.
Michael I experience this too quite often. I found that usually if you turn off obstacle avoidance it will fix it. Also before I take off I do what I call a “figure it out flight”. I put the drone up in the air do a couple 360’s each way, move forward, back, left, right, and land. This seems to also help. I don’t know if it actually accomplishes anything but my theory is that with the IMU and accelerometers there needs to be some data before it will really work like a charm.
I don’t use obstacle avoidance, but that was a good thought. I’m not glad you have experienced it, just that someone else has… 
Mine is damn near 1/4 of my NADIR flights. Might try switching gimbal mode to FPV rather than follow and try to have your flight heading 0-360,90,180, or 270.
Here is some data from the example project above. The overlaps do look pretty good except for the edges which still aren’t much worse than a flight with the correct heading. You can even see the skew in the overlaps. Not part of this thread, but look at the poor timing in the shots. The real problem comes in with the SFM values and the Camera Calibration assessment. Note the Y-Axis (north to south) is so much further out. Yikes on the distortion map…
The timing doesn’t look great - unclear why the camera does that. But high overlap eliminates this as a problem. Interesting on the distortion - have you seen this before on this drone?
I had 80% front lap so the map is fine, but I would still really like to figure out the cause. I have seen it with 3 different P4P’s and 4 different mobile devices, one of them being an Android.
As for the distortion, no I have never seen that before. It has always been a consistent gradient and symmetrical.
A flight line example of the skew behaviour. As you can see the line running north to south yawed to the west (right) as it progressed and the line running south to north the opposite way (left) to also yaw to the west. I have calibrated the drone each flight and the behaviour continues.
I did a full calibration of the P4P this morning out in the middle of nowhere and confirmed the orientation with a digital compass. Since this has just shown up I feared it was something wrong with the drone, but after having it happen on two different units and another user experiencing the same thing I think not. It happened all all four flights this morning including one linear flight. As I said above it seems to be correctly aligned at the ends and is skewed the most at the midpoint. I now know that it is affecting my overlaps on the edges of the map because it is only capturing a triangular portion and is missing allot of GCP images as a result. I usually keep my GCP’s to where they can be seen in 2-3 images laterally which provides 8-12 images per point depending on the front overlap, but I am now only capturing 4-6 on the same map that I have flown before this started happening.
20ft into the line
200ft further down the line
I have noticed this more with my Mavic 1 than my Mavic 2. I always attributed it to that particular drone. When flying around in DJI Go, pushing the stick straight forward would always drag to the side a bit and not fly straight. I could fix this by calibrating the sticks in the app, but if I were to load something like Dronedeploy and then later go back, the calibration would be lost and need to be redone.
More data. Distortion maps from the processing report before (3 months ago) and after the skew started happening. @Jamespipe @Andrew_Fraser 
Noticed - The top one appears to be a Phantom 4 Pro v2 the bottom one is a Phantom 4 Pro.
It certainly looks like you can see the diagonal rolling shutter in the distortion map.
Rolling shutter? V1.0 or V2.0 are they not the same camera? It happens with both models regardless.
Yes. Mechanical shutter needs to be specified in camera settings though. I’m not saying it wasn’t, just that it is possible for electronic shutter to be set.
I believe mechanical is the default, but will check that before my next flight. If it is set for DroneDeploy control it will use the correct one right? I don’t think it has anything to do with this and is a result of the skew and the improper overlaps that it is creating. I’ll be back.
