I want to do it easily with Jupyter-lab + numpy / scipy.
It is not for people who do ordinary (?) Image processing in the world.
SSIM
In skimage
https://scikit-image.org/docs/dev/auto_examples/transform/plot_ssim.html
SNR
It seems to be in scipy, but the expression itself is easy, so it seems better to write it in numpy
https://github.com/scipy/scipy/issues/9097
https://www.codespeedy.com/calculate-signal-to-noise-ratio-in-python/
FLIP
ꟻLIP: A Difference Evaluator for Alternating Images https://research.nvidia.com/publication/2020-07_FLIP
Freshly made with HPG2020.
It seems that it can be used for calculating the error between CG and live action, so it is offered to users like us.
There are implementations in various languages, including implementations in numpy / scipy.
Detecting Bias in Monte Carlo Renderers using Welch’s t-test http://jcgt.org/published/0009/02/01/
For Monte Carlo rendering. There may be python code here as well.
Recommended Posts