Hello everyone.
Currently I am starting to research in dynamic novel-view synthesis with Neural Radiance Fields and (3D) Gaussian Splatting with deformation, time-varying and flow approaches. State of the art methods are mainly developed with Python and PyTorch, so there are few to none papers implemented natively with MATLAB.
There have been a lot of years since I used MATLAB, but recently I’ve been interested in resuming it to research dynamic NVS, particularly because I didn’t knew about the bi-directional integration with Python and thus with ML frameworks such as PyTorch. However, I do not know how mature is MATLAB for that, or, if it has the capabilities to research new methods via its native language and tools, or with the Python integration.
MATLAB is very powerful in digital signal processing and mathematical optimization, both important to dynamic NVS (or dynamic 3D reconstruction), so been able to combine all those capabilities through its Toolboxes with the "standard" in research (Python + PyTorch) sounds really good, but I don’t know if it is a good idea regarding ease of implementation, training with cloud, and all prepreocessing tasks, especially given the COLMAP dependency.
So i would really appreciate info, experiences, and perspectives about that. Thanks in advance.Hello everyone.
Currently I am starting to research in dynamic novel-view synthesis with Neural Radiance Fields and (3D) Gaussian Splatting with deformation, time-varying and flow approaches. State of the art methods are mainly developed with Python and PyTorch, so there are few to none papers implemented natively with MATLAB.
There have been a lot of years since I used MATLAB, but recently I’ve been interested in resuming it to research dynamic NVS, particularly because I didn’t knew about the bi-directional integration with Python and thus with ML frameworks such as PyTorch. However, I do not know how mature is MATLAB for that, or, if it has the capabilities to research new methods via its native language and tools, or with the Python integration.
MATLAB is very powerful in digital signal processing and mathematical optimization, both important to dynamic NVS (or dynamic 3D reconstruction), so been able to combine all those capabilities through its Toolboxes with the "standard" in research (Python + PyTorch) sounds really good, but I don’t know if it is a good idea regarding ease of implementation, training with cloud, and all prepreocessing tasks, especially given the COLMAP dependency.
So i would really appreciate info, experiences, and perspectives about that. Thanks in advance. Hello everyone.
Currently I am starting to research in dynamic novel-view synthesis with Neural Radiance Fields and (3D) Gaussian Splatting with deformation, time-varying and flow approaches. State of the art methods are mainly developed with Python and PyTorch, so there are few to none papers implemented natively with MATLAB.
There have been a lot of years since I used MATLAB, but recently I’ve been interested in resuming it to research dynamic NVS, particularly because I didn’t knew about the bi-directional integration with Python and thus with ML frameworks such as PyTorch. However, I do not know how mature is MATLAB for that, or, if it has the capabilities to research new methods via its native language and tools, or with the Python integration.
MATLAB is very powerful in digital signal processing and mathematical optimization, both important to dynamic NVS (or dynamic 3D reconstruction), so been able to combine all those capabilities through its Toolboxes with the "standard" in research (Python + PyTorch) sounds really good, but I don’t know if it is a good idea regarding ease of implementation, training with cloud, and all prepreocessing tasks, especially given the COLMAP dependency.
So i would really appreciate info, experiences, and perspectives about that. Thanks in advance. 3d, computer vision, digital image processing, python MATLAB Answers — New Questions