File: Screenshot from 2025-11-0(...).png (152 KB, 871x1494)

152 KB PNG

Finally i've completed the 3D proyective reconstruction from both 2 and multiple views, now im debating between uprading to euclidean reconstruction from the recovery of the absolute cuadratic or by exploiting the calibration matrix as an intrinsic parameter, either way i've decided that this process of reconstruction will not lead to an subsequent process of dense reconstruction due to it's depencence of non ocluded points and small baseline, instead ill only use this method for camera pose reconstruction and with the movement of the camera obtained ill use volumetric space carving with boxels to estimate the shape of 3D objets, thanks to a book i've found that describes the process 'A Framework for Realtime 3-D Reconstruction by Space Carving Using Graphics Hardware', Im confident this will allow me to achieve an efficient-highly deterministic 3D reconstruction, also I'm planing on optimizing it using euristic shape prediction that will allow me to do fast estimation on the 3D shape based on little info about the object's shape using scene context. Any comments or recomendations will be welcommed. Have a nice day g-fag fellas