[request] using depth-camera information (Galaxy S9+)?

PhinPhin commented 5 years ago

Is your feature request related to a problem? Please describe.

Not a problem per se, but I noticed there is that 3D-photo function for Facebook that uses the depth-camera data generated on the Samsung Galaxy S9+ (and probably newer ones and the corresponding aged Apple products).

Describe the solution you'd like

In Portrait/LiveFocus-mode you can take photos and blur the background, the photos are stored with depthinformation of the image. I assume this information could be used to solve issues if an angle isn't clear enough? Right now I have no clue how accurate the depthinformation is. Anyone who has a better insight who could point us to some indepth-information about the technology thes use in stated phone?

Describe alternatives you've considered

Alternatives? None that I know off aside from using expensive LIDAR- or StructuredLight scanners... recommendations for large stuff (buildings) and small (like closeups of coins or small machinery parts)?

Additional context

I try to scan bones and objects for 3D-printing and putting it in scientific displays and it takes AGES to render on my trusty i5 2500k....

natowi commented 5 years ago

You better ask this question in the https://forum.xda-developers.com/ forum.

Tools https://github.com/jpbarraca/dual-camera-edof https://github.com/zekeadam/EDOF https://github.com/panrafal/depthy

fabiencastan commented 5 years ago

@PhinPhin Could you provide some datasets?

stale[bot] commented 4 years ago

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.

fire commented 3 years ago

https://store.intelrealsense.com/buy-intel-realsense-lidar-camera-l515.html Is a consumer accessible LIDAR scanner.

What data sets would you want and would you recommend patching meshroom to support it?

Intel® RealSense™ LiDAR Camera L515 $349.00 USD

Edited:

See https://github.com/IntelRealSense/librealsense/tree/master/examples/align-advanced

natowi commented 3 years ago

@fire most affordable lidar scanners are 2d lidar scanners. True 3D lidar scanners are more expensive. There are DepthCameras and lidar cameras like the intel realsense, but they are for industrial applications or only supported by a few software packages for specific use cases. A ready to use combination would be the proprietary software https://www.recfusion.net/ + realsense => both for 520-850$. Open Source package RTabMap supports the linked realsense camera https://github.com/introlab/rtabmap/issues/574 but although Rtabmap supports multiple Depth cameras and sensors (see http://wiki.ros.org/rtabmap_ros) it is far from being as user friendly as Meshroom.

Maybe Meshroom will get support for other Depth-Sensing devices in the future, but not in the foreseeable future. (Most of the time cheap Lidar scanners come only with some basic tutorials, scarce documentation and additional hardware requirements - so completely unsuited for the end-user and a hassle for developers as many sensors rely on their own API who commonly change and break as can be witnessed in past RTABMAP issues.) So at the moment the best open source software for non photogrammetric devices for depth sensing can be found in the ROS community (like RTABMAP). Meshroom is at the moment more focussed on photogrammetry and traditional RGB images (with the potential to support depth images in the future).

alicevision / Meshroom

[request] using depth-camera information (Galaxy S9+)? #623