High quality SfM but poor depth mapping [question]

[Swirl-droid]

Problem
A dataset of around 400 images of an outdoor statue was taken. When imported to Meshroom, the SfM data looks great and can be meshed directly to produce a reasonable result. However, after depth mapping and filtering, the end result is unrecognizable. Looking at the depth mapping visualizer, I found that its a little noisy, but it looks alright otherwise. Can anyone help me to fix this problem?

Screenshots

Dataset
https://drive.google.com/file/d/1dD5AeiU28fLzlz5niuxl9h-ya2Y_LTnb/view?usp=sharing

Other information

• OS: Win 10
• Meshroom version: 2021.1.0
• Images taken with a Nikon D7000. focal length: 200 mm, f-stop number: 32, Metering: Matrix, Shutterspeed: Automatic (around 1 sec)

[tkircher]

This has been a known issue for ages. There is a very significant amount of noise in the depth map step; the only reliable output is from the SfM step. It's actually better to have more points there and skip the depth map.

Edit: Here's a link to the 'main' post about this problem. It does not offer any solutions, but most commercial software computes all the depth values directly from SfM and then generates a mesh.

#566

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[stale]

This issue is closed due to inactivity. Feel free to re-open if new information is available.