Can images with limited depth of field still be used in RealityScan?

General Feedback & Requests
, ,
1

On 24 May 2026, using a Samsung Galaxy S23 and its native camera app (v16.5.02.37), I shot several detailed close up photos of stone fence posts lying horizontally on the ground intended to augment the mapping mission flown by the drone. They were all shot on the same day as the drone images and under the same overcast lighting conditions. Many of the images include in frame the RTK surveyed points.

After looking at the cell phone photos, it appears that Samsung’s camera app had added its own processed “enhanced” depth of field adjustment. The DOF issue seemed to have happened, or noticeably worsened, after a recent UI update. Under Camera Settings, Camera Assistant, Picture Softening, it is now turned off, but in general, can images with limited depth of field still be used in RealityScan?

For example, in this image we see GNSS2, a nail in root that was set in 2024 in the immediate foreground, and GCP3, the photo target, a nail and orange ribbon in a 15cm black paper plate with 1" white duct tape + set for the drone mapping mission about 2 meters away in the middle ground which has been blurred by the Samsung camera app to appear out of focus.

I later learned after I had taken the photos off of the camera, the faux DOF “enhancement” is reversible, but only as long as the images remain on the phone.

20260524_145112
20260524_1451121920×1440 1.12 MB

2

Hello @vkbellis , the photos with limited depth of field can still be used, even though they are not ideal. RS checks for blurry parts during calculations.

Modern phones have manual modes where you can adjust your aperture, shutter speed, and ISO. I would try to use that. Try using a smaller aperture (a higher f-stop number). These small sensors and lenses should actually give a broad DOF, and most of the DOF is artificial. If the native camera app doesn’t support that, you can try a 3rd party camera app.

Also, disable any artificial enhancements and AI filters, those can only mess up the alignment.

Becasue of the limited DOF, you should adjust the way you scan objects. Make sure that each part of your scanned object is sharp in at least two, but ideally more, images. There needs to be an overlap between the sharp parts so RS can do reliable matching. Basicaly take more photos with higher overlap.

3

Excellent answer Jacub, thanks so much!