In this video we look at some aspects of using Meshroom to do 3D-Reconstructions. We start by discussing the first part of the pipeline, up to the StructureFromMotion node. We may discuss the remaining nodes in a future video. Then we look at automatically scaling our model with CCTags, followed by what changes when we use a turntable. Finally, we look into using known camera positions to make our reconstructions more reliable. You can find the CCTags under: (see folder markersToPrint). The github issue regarding known camera poses can be found here: For the Dockerfile to build Meshroom with my adjustments see here: Do you need help on a project? Whether it's about theoretical analyses, planning or implementing Python/C++ code, I can help. Contact me through the form on my website on ~0167c8e428f0643acf?mp_source=share. Contents: 00:00 Intro 00:50 Pipeline - CameraInit 02:57 Pipeline - FeatureExtraction 04:50 Pipeline - FeatureMatching 05:26 Pipeline - ImageMatching 08:09 Pipeline - StructureFromMotion 11:08 CCTags 14:53 Using a Turntable 19:13 Known Camera Positions for ImageMatching 22:21 Using Node ImportKnownPoses 1 23:41 Meshroom's Coordinate System 25:50 Using Node ImportKnownPoses 2 28:34 Outro