Python Image and Video Read/Write Examples

Functions to write multipage images and make videos from Numpy arrays. Also includes functions to test Matlab and Python OpenCV codecs.

Install

python -m pip install -e .

Scripts

HDF5_to_AVI.py converts HDF5 video data to AVI. Typically used in biomedical and science imaging, where they use HDF5 instead of TIFF, since HDF5 is a very widespread, fast file format that can store arbitrarily large datasets and metadata. It is recommended to researchers to use HDF5 instead of TIFF, FITS, CDF or proprietary formats for large data streams, including lossless video.

image_write_multipage.py demonstration of writing multipage TIFF from Numpy arrays, using tifffile and freeimage (tifffile is 20x faster and much more featureful)

imageconv.py converts large directory of files with same prefix to multipage TIFFs

Audio

The audio_*.py scripts show several methods for generating audio output from Python.

Python lossless AVI writing

You can write AVI from Python without axes labels (openCV) or with labels/axes (Matplotlib)

OpenCV AVI writing

Demo_OpenCV_writeAVI.py demonstrates using OpenCV to write video, with a lot of codecs to try. This does not insert any of the axes stuff that Matplotlib does, so it can be useful for machine vision work. You can optinally overlay dynamic text on the video.

Matplotlib AVI writing

matplotlib_writeavi.py writes axes labels AVIs lossless using Matplotlib. Note that file-based matplotlib.animation.writers['ffmpeg_file'] is used for better reliability, even though it takes twice as long as the pipe-based ffmpeg. On some computers, pipe-based matplotlib.animation.writers['ffmpeg'] results in very distorted, scrambled output for some reason, with Matplotlib 1.5 and 2.0. In that case, I have empirically found ffmpeg_file to Just Work.

Notes

• errors: libfreeimage, how to fix them: https://www.scivision.dev/writing-multipage-tiff-with-python/