canessae / altiro3d Goto Github PK

altiro3D is a free, extended C++ Library (-see releases) developed to reconstruct reality from a given single imagine (e.g., .png, .jpg) or flat video (e.g., .mp4). This is done generating a light-field (or Native) image -see "User's Manual"

The several altiro3D command lines provide the following features: • Create Native (i.e., 3D image) from Photo using MiDaS-small. • Create Native from Photo using MiDaS-large. • Create Native from a given original camera’s RGB image and depth image. • Create Native from a given Quilt (N×M). • Create Native from sorted N-views (i.e., sequential set of plain images) stored in a given directory. • Convert Quilt views to 2D video (.mp4). • Convert given 2D video to Native 3D video (.mp4). All these forms of Native images/videos consisting of light-fields can be correctly displayed on a specific slanted, lenticular screen such as the Looking Glass (LG) Portrait.

The latest version of the Debian ’altiro3D-x.x.x-Linux.deb’ package can be downloaded from "Releases" For details see "User's Manual"

	sudo dpkg -i altiro3D-x.x.x-Linux.deb

Set the path to the working directory

	source /opt/altiro3D/bin/setupvars.sh

	cd /opt/altiro3D/bin/examples
	./run_examples-1.sh 
		...

	sudo dpkg -r altiro3D

If this software (including any part of it) is useful for you, please cite

@misc{altiro3D:
  title = {{Scene representation from single image and novel view synthesis}},
  author = {E. Canessa, L. Tenze},
  doi = {10.48550/arXiv.2304.11161},
  url = {https://arxiv.org/abs/2304.11161},
  publisher = {arXiv},
  year = {2023},
}

Article to appear (2023) in Springer's International Journal of Information Technology (IJIT).