GPU-based Light Field Rendering with CuPy, pyTOFlow
To store and calculate images, NumPy and CuPy modules are used. CuPy is used for calculation in GPU environment.
Our implementation is based on PyTorch 0.4.1, but we can use the other versions. The current environment is PyTorch 1.7.1. (python 3.6.12)
The part of interpolation is based on pyTOFlow, so we need modules and files in https://github.com/Coldog2333/pytoflow.
After After downloading the https://github.com/Coldog2333/pytoflow, insert LFRendering_TOFlow_0326.py and M3DLF_S3(LightField photo folder) and run it.
python LFRendering_TOFlow_0326.py --o path_out.jpg --gpuID 0
Because there were modifications in pyTOFlow, there are only two options.
- --o [optional]: filename of the predicted frame. default: out.png, saving in the same directory of the input frames.
- --gpuID [optional]: No of the GPU you want to use. default: gpuID = 0.
