Given a text with annotations, it searches and downloads Google Images and generates a text-on-image slide-show.
Annotation to video making instructions is implemented through a custom grammar, an ANTRL lexer and parser and a custom visitor. Video Instructions to video is implemented using the python libraries cv2 and Pillow.
There are two types of annotations, coupling slides and slide modifiers.
(word1 word2 ...) make a multi-word slide
Word1, word2... will appear in the same slide.
Modifiers come at the end of the word/sentence.
{word} changes default background search to word
The slide (Hello all!){man waving} will have a man waving background.
[N] pauses N tenths of a second
The slide (Hi!)[10] will have an extra 10 tenths of second screen time added to its default time. Default times depend on the number of letters of the slide and its puntuations symbols.
* doesn't change background for that slide
The slide (Hello world!)* will have the previos slide's background. If it is the first slide, it will have a black background.
You can find the complete Grammar in lazyvideo.g4.
(Instead of) {arrow} [10]
("Have a nice day") {sunny day} [5]
(think I'll start saying) {start} [5]
("Have the day you deserve") {angry man drawing} [5]
(You know,) {karma} [5]
(let Karma sort that shit out.) * [10]
To execute LazyVideo, use
python3 lazyvideo.py --input <input-file-path>If you want to modify the grammar, build the ANTRL files with
antlr4 -Dlanguage=Python3 -no-listener -visitor lazyvideo.g4