Good first contributions - alicevision/Meshroom GitHub Wiki

Looking for a place to start? Here is a list of various projects you are welcome to contribute:

:octocat: good first contribution

:star: special

👑 challenging

https://github.com/alicevision/meshroom/issues?q=is%3Aissue+is%3Aopen+label%3A%22feature+request%22

Ways to contribute without software development skills :octocat:

contribute to the https://github.com/alicevision/meshroom-manual (proofreading, feedback, writing own chapters,...) Open a new issue in meshroom-manual to discuss the details
create hq illustrations for the documentation or website
contribute to the sensordatabase
(once the documentation is finalized: translation)
share your own ideas

Add file picker to add images :octocat:

#461 (pull request: https://github.com/alicevision/meshroom/pull/867)

Add/Edit Exif info :octocat:

Manual addition of EXIF data within Meshroom would be a good first contribution, with options for selecting 1 or N specific images, or All images, to apply the data to. https://github.com/alicevision/meshroom/issues/642

Maybe add support for ExifTool?

check for image validity :octocat:

Show popup when a corrupted images is detected. https://github.com/alicevision/meshroom/issues/630 https://github.com/alicevision/meshroom/wiki/Images-cannot-be-imported https://github.com/alicevision/meshroom/pull/635

Notification sounds :octocat:

Notification sounds for success and error Details: #268

Set relative paths :octocat:

Change absolute paths to relative paths within Meshroom. Same for the Input and Output paths of the nodes #472

Node to merge 2 meshes :octocat:

https://github.com/alicevision/meshroom/issues/864 Details: #473

Hide "advanced" values for choice param in Meshroom.

hide "advanced" values for choice param in Meshroom. Example to hide: sift_ocv This could work well with alicevision/meshroom#676 (comment)

Camera calibration Lensefun

We could use the existing Lensefun database to get the distortion parameters for over 1000 lenses. The database is licensed under CC_BY-SA_3.0 and could be integrated similar to the sensor database. https://github.com/lensfun/lensfun https://wilson.bronger.org/lensfun_coverage.html (dropped 2016 https://github.com/alicevision/AliceVision/issues/10)

Remeshing & Retopology :octocat:

WIP https://github.com/alicevision/meshroom/issues/774

Meshroom has its own Remeshing/Simplification nodes. However, instant-meshes is more specialized.

https://github.com/wjakob/instant-meshes

Also interesting: http://bakemyscan.org/ https://github.com/norgeotloic/BakeMyScan

Estimate disk & ram space

(*and remaining processing time WIP: https://github.com/alicevision/meshroom/pull/842) Details: #426 https://github.com/alicevision/meshroom/pull/778

dense point cloud computation without meshing

https://github.com/alicevision/meshroom/issues/1354

manual gpu selection

restrict resource consumption

ram, cpu, gpu usage

Visual interface for node statistics #564

Add overheat protection for local computations (pause computation for cooldown)

Sketchfab uploader

IMPLEMENTED https://github.com/alicevision/meshroom/pull/712

Group photos that have feature correlation, but are not reconstructed

view of an image with matched features visible/highlighted #514

Removing a photo from a group leads to redoing feature extraction

#510

Contribution of single camera on SfM (and visa versa)

what cameras have contributed to a single point #512

Better handling of video files (KeyframeSelection)

When adding mp4 files to Meshroom, the Keyframe selection node is added to the Graph Once the node has been processed, the images are not automatically added as input. It should also be possible to use Augment reconstruction in some cases. Needs to be compatible with external computation. https://github.com/alicevision/meshroom/issues/232#issuecomment-540688506

Manage Node Graph variations

#361

Project Manager/Templates

#360

Manual image matching (method & GUI)

#494 #450 View and correct and create tie points https://groups.google.com/forum/#!topic/alicevision/DoVpPYzCQz0

Line reconstruction :octocat: :star:

(WIP, internal) It would be great to improve edges on Meshes

https://github.com/manhofer/Line3Dpp (MPL2) :star: (Many thanks to manhofer for changing the license)

https://github.com/abignoli/EdgeGraph3D (GNU) newer

https://github.com/ySalaun/LineSfM (MPL2&others)

Optical flow

reconstruction based on optical flow

https://github.com/alicevision/meshroom/issues/232#issuecomment-593305597

http://lmi.bwh.harvard.edu/papers/pdfs/gunnar/farnebackICPR00.pdf

https://vision.in.tum.de/research/optical_flow_estimation

https://www.mia.uni-saarland.de/Publications/schneevoigt-gcpr14.pdf

https://github.com/ipa-mah/3D-Reconstruction-using-Dense-Optical-Flow

https://github.com/jswulff/pcaflow

https://github.com/menandro/Reconflow

https://github.com/limjiayi/RECONSTRUCT

https://github.com/menandro/opensor

https://github.com/menandro/sor

https://www.cvl.iis.u-tokyo.ac.jp/data/uploads/papers/Menandro_OpticalFlow_WACV2018.pdf

https://docs.opencv.org/3.4/d4/dee/tutorial_optical_flow.html

https://nanonets.com/blog/optical-flow/

https://www.researchgate.net/publication/228543845_3D_RECONSTRUCTION_FROM_OPTICAL_FLOW_GENERATED_BY_AN_UNCALIBRATED_CAMERA_UNDERGOING_UNKNOWN_MOTION

Texturing

show images as 2D plane in 3d Viewer project images from camera position on the object #358

PointcloudRegistration

Combine point clouds using PCL https://github.com/alicevision/AliceVision/pull/425

Visualize point cloud density (sparse and dense)

Dense reconstruction/ dense point cloud Open Source

List of open source software

Dense reconstruction (CPU/(GPU))

provide a node to compute dense point cloud from sparse point cloud or implement another solution for dense reconstruction

also provided with some SLAM tools https://github.com/alicevision/meshroom/issues/266

Dense reconstruction from unstructured image collections (Schönberger et al.) CPU

Implementation as part of COLMAP https://github.com/mitjap/pwmvs https://demuc.de/papers/schoenberger2016mvs.pdf

White balance of images

provide a tool to correct the white balance of the images for a better final texture similar to https://lesterbanks.com/2018/12/easycorrect-nuke-gizmo-adjust-images/

Camera projection on surface

#513

3D Features viewer

similar to 2d features viewer but in 3d similar to depth map viewer

Images / Images Viewer / 3D Viewer

#272

Graph Editor

#269

Rotate / Scale model in Meshroom

https://github.com/alicevision/meshroom/issues/256

https://github.com/alicevision/meshroom/issues/545

IMU/Gyroscope Integration for camera rotation estimation from mobile devices

https://github.com/rogermm14/rec3D http://openaccess.thecvf.com/content_cvpr_2016_workshops/w14/papers/Muratov_3DCapture_3D_Reconstruction_CVPR_2016_paper.pdf

Gyroscope sensor data logger for mobile devices (txt/csv) Example apps: https://github.com/sztyler/sensordatacollector https://github.com/tyrex-team/senslogs https://github.com/e-lab/VideoSensors

Camera tracking using SteamVR

#529

OpenVR-Tracking-Example: A small c++ example on how to access OpenVR tracking data and controller states using IVRInput system https://github.com/Omnifinity/OpenVR-Tracking-Example

vive-diy-position-sensor: Code & schematics for position tracking sensor using HTC Vive's Lighthouse system and a Teensy board. https://github.com/ashtuchkin/vive-diy-position-sensor

HTC Vive Tracker Node for ROS https://github.com/moon-wreckers/vive_tracker

Photometric_stereo

WIP https://vimeo.com/411076799

https://github.com/alicevision/meshroom/issues/928

https://en.wikipedia.org/wiki/Photometric_stereo

Orthoimage support

#246 https://github.com/natowi/orthoimage_software_collection/blob/master/README.md

Real-time Visual Odometry

Get camera position and path from video footage. As alternative to cctags without markers Example: https://www.youtube.com/watch?v=F3OFzsaPtvI https://www.youtube.com/watch?v=2YnIMfw6bJY https://github.com/uzh-rpg/rpg_svo

Sequential Rig (4D)

(WIP)

Network processing client 👑

build-in client for Meshroom similar to Photoscan #357

GCP support 👑

#230

GPS support

Use GPS for camera positions and model scaling Support GPS EXIF tags including pitch/roll/yaw

https://github.com/openMVG/openMVG/issues/547

Coordinate System 👑

https://groups.google.com/forum/#!topic/alicevision/Pd7kwKxfbes

Real world coordinates

convert using https://proj.org/about.html https://proj4.org

GDAL + CGAL

Real-time reconstruction using SLAM 👑

There are several software libraries for monocular SLAM that can reconstruct 3D models in real-time while tracking the camera's position. https://github.com/raulmur/ORB_SLAM2 https://github.com/alicevision/meshroom/issues/232

https://github.com/ethz-asl/maplab

Image Masking 👑👑

#188 WIP & first DEV version: https://github.com/alicevision/meshroom/issues/566#issuecomment-527822623

Deeplab based masking 👑👑👑 https://groups.google.com/forum/#!topic/alicevision/GvX1rDMI3oQ

Software list: https://github.com/natowi/masking_tools/blob/master/README.md

CPU / non-cuda implementation of the DepthMap node 👑👑👑

#439

(hipcl or https://github.com/hughperkins/coriander) https://www.iwocl.org/wp-content/uploads/iwocl2017-hugh-perkins-cuda-cl.pdf

DeepLearning integration for 3D Reconstruction 👑👑👑👑

https://groups.google.com/forum/#!topic/alicevision/HQhqtJjGaQ0 https://groups.google.com/forum/#!searchin/alicevision/project|sort:date/alicevision/CS0Os345kio/APLjJWkDEAAJ

Use neural networks for create depth map https://github.com/alicevision/meshroom/issues/528#

GeoDesc replace SIFT

LiXudong found a new algorithm that seems to replace SIFT. This algorithm is called GeoDesc and I don't know if you know this algorithm. Its GITHUB address is: https://github.com/lzx551402/geodesc

Graph Structure from Motion (GSfM)

https://groups.google.com/forum/#!topic/alicevision/_5Eo6hqLBS8 There (on Github) is a more efficient than state-of-the-art open-source SfM system that surpasses the accuracy of OpenMVG (...) at the same time. The reconstruction system, named i23dMVS, ranks in the top 10 in tanks and temples dataset. Here the link to the project: https://github.com/AIBluefisher/EGSfM Since GraphSfM is partially based on (an early) version of OpenMVG and licensed under BSD 3-Clause, It should be possible to include this approach in Meshroom to accelerate large scale reconstructions.

Polarimetric Multi-View Stereo http://alumni.media.mit.edu/~shiboxin/files/Cui_CVPR17.pdf