Art and Science With Six Million Timelapse Images and Python
summary
A window camera plus a Raspberry Pi and gphoto2 collects six and a half million images (about 4.2 TB storage) from a Canon A510; images hit a Postgres DB via the Django ORM with EXIF pulled by piexif and stats from Pillow. Code normalizes uneven 10s captures, shells out to mencoder for movies, uses Celery for parallel jobs, stacks nights by lightest pixel to reveal lightning, extracts one-line daily scans to visualize season changes, and runs simple OpenCV/pythonsimplecv face/feature detection; cameras and SD cards sometimes fail but the pipeline works.
AI-written summary · about
chapters (26)
- 0:00Art and Science With Six Million Timelapse Images and Python
- 0:54Camera Setup
- 1:12Timelapse Equipment
- 6:49Image Info Model
- 8:03Inserting Image Data
- 10:31Normalization
- 12:23Normalized Model
- 12:37Creating Normals
- 17:04Using Pyephem
- 17:40Using Generators to Get Sunsets
- 22:56Reduce our Workload
- 24:47Using PIL's ImageChops
- 25:12Lightning!
- 26:19More Lightning (Multiple Images)
- 31:08Daystrip Scanner
- 32:25Shortest and Longest Day
- 33:00Scanning to Make a Picture
- 35:51Time is no longer on the Z axis
- 37:48Let's Use All The Days!
- 38:13A Visual Representation
- 38:49Using Median Color
- 39:27Using Max Color
- 41:08Sunset Mosaic
- 44:10Faces in Clouds
- 44:43Open Source Makes it possible!
- 45:42Thank You!
related talks
51mFun Observational Science with Python and a Webcam
42mMaking Sense Out of Flight Test Data with Python
31mOpenDataCube - Petabyte Scale Satellite Imagery Processing in Python
39mOpening a treasure trove: The Historical Aerial Photography project
27mExploring Art with Python: Building an Italian Art Bot
33mMixing a Raspberry Pi with Python into a 5th Grade Science Fair Project
40mDigitizing 125 Years of Images
43mProcessing Data From The James Webb Space Telescope