How the video was created

Set up the Raspberry Pi

During the demolition process, I had the RasPi take one picture every minute. To do so, I set up a Cron job to run a customized call to the raspistill program. This script created a time-stamped file name for each image and saved it a folder I created in the home directory (/home/pi/timelapse). The file name follows the convention "YYYY-MM-DD_hhmm.jpg"

#!/bin/bash
 
DATE=$(date +"%Y-%m-%d_%H%M")
 
raspistill -hf -vf -o /home/pi/timelapse/$DATE.jpg

*/1 6-19 * * * /home/pi/bin/make_raspistill.sh

Use Python Imaging Library to place a time stamp on each image

The Python Imaging Library (PIL) makes it very easy to overlay text on images. (I'm using the PILLOW fork of PIL.) I wrote a quick script that first resizes and crops each image, and then places the date and time stamp in the corner. The script needs all images to be placed into a folder, and then it needs to be run from within that folder. The script also needs a subdirectory called "resized" in which it will place the processed images with the timestamp overlay. The vertical crop borders (180 px), as well as text size and placement, were found by trial and error.

# script to draw image date/time over the image itself
 
from PIL import Image
from PIL import ImageFont
from PIL import ImageDraw
import os
 
font = ImageFont.truetype("/System/Library/Fonts/HelveticaNeue.dfont", 72)
fontsmall = ImageFont.truetype("/System/Library/Fonts/HelveticaNeue.dfont", 32)
fontcolor = (238,161,6)
counter = 0
# Go through each file in current directory
for i in os.listdir(os.getcwd()):
	if i.endswith(".jpg"):
		counter += 1
		# For debugging: limit how many images are processed:
		# if counter>10:
		# 	break
		print("Image {0}: {1}".format(counter, i))
		# Files are named like YYYY-MM-DD_hhmm.jpg
		# * Date is the first chunk before the underscore
		# * Time is always the first 4 characters after the "_"
		splitup = i.split("_")
		date = splitup[0]
		t = splitup[1][0:4]
		# Add colon to time
		tformatted = t[0:2] + ":" + t[2:4]
 
		# Open the image and resize it to HD format
		# 720p (1280×720 px)
		# 1080p (1920x1080 px)
		widthtarget = 1920
		heighttarget = 1080
		img = Image.open(i)
		downSampleRatio = float(widthtarget) / float(img.width)
		imgDownSampled = img.resize( (widthtarget,round(img.height*downSampleRatio) ), resample=Image.LANCZOS)
		imgDownSampled = imgDownSampled.crop((0,180,widthtarget, heighttarget+180))
 
		# get a drawing context
		draw = ImageDraw.Draw(imgDownSampled)
		draw.text((imgDownSampled.width-220,imgDownSampled.height-150), date, fontcolor, font=fontsmall)
		draw.text((imgDownSampled.width-220,imgDownSampled.height-120), tformatted, fontcolor, font=font)
		filename = "resized/" + i[0:-4] + "-resized.jpg"
		imgDownSampled.save(filename)

Use FFMPEG to make the movie from the image sequence

My workflow for converting image sequences into a movie is as follows:

1. Rename images so file names are numbered sequentially
2. Run ffmpeg on the sequentially named images

ffmpeg won't accept the time-stamped image names as input. Luckily, there's an easy way to rename the files to end with a zero-padded numbered sequence. It uses a modified "rename" program that can be found here: http://www.volkerschatz.com/unix/uware/rename.html. (I called it "renum" on my system to keep it separate from the standard Linux/Mac "rename" shell command.) Here's an example:

Change "2016-08-09_0631-resized.jpg" into "image_0001.jpg", "2016-08-09_0632-resized.jpg" into "image_0002.jpg", etc. The "sprintf("image_%04d",${i})" part in the command below means that we'll get a file name that's zero-padded to four digits.

renum -n 's/(\d+)-(\d+)-(\d+)_(\d+)-resized/sprintf("image_%04d",${i})/e' *.jpg

The "-n" option above means you can test whether it works. The output will show whether the renaming works, i.e. whether your regular expression was correct. Change it to "-v" once you see it works.

Once we have our sequentially numbered images, we can call ffmpeg:

ffmpeg -framerate 30 -i image_%04d.jpg -c:v libx264 -r 30 -pix_fmt yuv420p output_filename.mp4