Is DD over-riding my camera settings?

So typically each time I begin a mission, I set the aspect ratio to 3:2. Then I take a pic in GO4 confirming the change from 4:3.

Once image capture begins, I have noticed it goes to 4:3. I hadn’t noticed this before the DD update.

This seems like a simple problem to fix. Do I just deselect something in the advanced settings so that it doesn’t cause me this problem. I want to map in 2:3. Why on earth would DD force image capture in a lower resolution after I have set the sensor to 3:2???

1 Like

Hey @TotalBeast. This has been discussed several times on the forum and it was determined a while back by DroneDeploy testing that 4:3 is better for quick image save, uploads, file storage overall pixel quality and processing. I don’t quite agree with a few of those, but pixel quality is a big deal. While the 3:2 ratio does allow for maximum resolution from the camera (20 vs 16) it provides more width and a less balanced front to side overlap ratio. The problem with the width is that you subjects at those extents become very oblique and also out of focus because they are so far from the metering. This points to the fact that there really are no advantages to using 3:2 and picture quality, upload speed and processing are all much better. The one thing that I really would like to see a progression on is the ability to shoot RAW. The 3:2 ration might become more viable at that point, but I still have my reservations because we are not shooting with cinematography cameras in a wider ratio and the pixels will still suffer. Maybe you can try some manual photos and different meterings to plead the case?

@Jamespipe mentioned that DD was going to give the full sensor images a try again which would reduce any given mission time by 10% (his number). I think it would be a good idea. But DD seems to have their hands full with a couple of nasty bugs right now so probably not a high priority.

I hope we get an option if that happens. I am happy with my flight time and don’t need an extra 12.5% of mediocre pixels on the edges. Obliques will be even worse. Maybe if there was a better matrix focus…

4:3 was what was originally used because the P3P’s sensor is 4:3 and that’s the drone everyone started with. Then the P4P came out with the 3:2 and intitially the engines couldn’t cope.

I’ve never seen any proof that images captured with the entire sensor are of any less quality than by cropping. Do you have a citation?

I understand that and agree that it should be an option, but if you take a photo of each and look at the 10% overage you can see the degradation of the pixels. At least I could when editing. Especially if there were GCPs. They are from such an oblique angle that they looked warped and blurry. This is why it is recommended by DD that GCPs not be used in obliques on the upload screen and why I do not perform missions if I want to use GCPs. Metashape tries to locate these and I have to remove a portion of them every time. While DD’s tagging process is much easier I assume that it will pickup these images as well and they may take precedence over better shots. This is one downfall to DD’s tagging is that you do not have the option to select any further than the chosen 8 and even if you search for more images your options will be lessened further.

Interesting. Sounds more like a field of view calculation error. If the individual images are sharp edge to edge, and the processor makes calculations (corrections) based upon the field of view, then the edge of the image should be just as good as anywhere else. James affirms this in his comment I linked.

If there was a mathematical reason why any image data a certain distance from the sensor center would have to be removed, it would apply in both axis and would be a mess going from camera to camera, I would think.

Perhaps the camera you were using was out of focus on the edge?

All I can share is what I have seen. Lenses are round so the further you get from center the further out of focus the subject will be and the higher the angle of skew will be. It does apply to both axis, just one more than the other. Choices are great and I think everyone should have a choice. Do what works for you.

Here’s some snippets from the same image. If you’ve ever tagged GCPs that were on the edge of the image you would have seen this a hundred times.

1 Like