fancy captions with moviepy

fancy_captions.py

import PySide6.QtSvg
import PySide6.QtGui
import PySide6.QtCore
import moviepy.video.VideoClip
import moviepy.editor
import numpy as np
from moviepy.editor import *
from vectortween.PointAnimation import PointAnimation
from vectortween.SequentialAnimation import SequentialAnimation

# set up some properties of the clip we are going to generate
fps = 25
duration = 10
W = 1000
H = 500

# some helper code to convert SVG to png
def toNumpy(incomingImage):
    '''  Converts a QImage into numpy format  '''

    incomingImage = incomingImage.convertToFormat(PySide6.QtGui.QImage.Format.Format_RGB32)

    width = incomingImage.width()
    height = incomingImage.height()

    ptr = incomingImage.bits()
    arr = np.array(ptr).reshape(height, width, 4)  # Copies the data
    return arr[:, :, :3] # remove alpha channel

# need to create dummy application to access font database
dummyapp = PySide6.QtGui.QGuiApplication()

# dynamically build a function that generates frames based on animation parameters and an svg text template
def make_frame_with_animation_and_text(xyanimation, anim_fraction, total_duration, text):
    """
    xyanimation: a PointAnimation to vary the x, y position of the text. (0, 0) is centered in the middle of the screen.
    anim_fraction: how long should the animation take to complete relative to the total_duration of the clip (e.g. 0.9 is 90%)
    total_duration: total duration of the clip in seconds
    text: svg text template (see example code below)
    """
    def make_frame(t):
        df = total_duration * fps
        tf = t * fps
        x, y = xyanimation.make_frame(tf, 0, 0, anim_fraction * df, df, None)
        svg = f"""<?xml version="1.0" encoding="UTF-8"?>
        <svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"
             width="{W}" height="{H}" viewBox="{-W / 2} {-H / 2} {W} {H}">
             {text.format(x=x, y=y)}
        </svg>    
            """
        r = PySide6.QtSvg.QSvgRenderer(bytes(svg, "utf-8"))
        i = PySide6.QtGui.QImage(PySide6.QtCore.QSize(W, H), PySide6.QtGui.QImage.Format_RGB32)
        i.fill(PySide6.QtGui.QColor(0, 0, 0))
        p = PySide6.QtGui.QPainter(i)
        r.render(p)
        p.end()
        return toNumpy(i.rgbSwapped())

    return make_frame

# define some fancy styles we can use while defining texts
my_styles = {
    "normal": """fill="black" stroke="white" stroke-width="2px" font-size="55" font-family="sans-serif" text-anchor="middle" """,
    "h1": """fill="blue" stroke="gold" stroke-width="4px" font-size="80" font-family="serif" text-anchor="middle" """,
    "h2": """fill="gold" stroke="maroon" stroke-width="2px" font-size="90" font-family="serif" font-style="italic" text-anchor="middle" """,
    "h3": """fill="green" stroke="white" stroke-width="6px" font-size="200" font-family="serif" text-anchor="middle" """,
    "small": """fill="red" stroke="white" stroke-width="5px" font-size="30" font-weight="bold" font-family="serif" text-anchor="middle" """
}

# define some fancy animations we can use to move our text around the screen
anim = {
    "top_to_bottom": PointAnimation((-W, -H / 2), (0, H / 3), tween=['easeOutBounce'], ytween=['easeOutBounce']),
    "topright_to_left": PointAnimation((W + 10, -H), (0, 0), tween=['easeOutQuad'], ytween=['easeOutCirc']),
    "multi": SequentialAnimation([
        PointAnimation((-W / 5, -H / 3), (W / 3, H / 2), tween=['easeOutBounce'], ytween=['easeOutQuad']),
        PointAnimation((W / 3, H / 2), (-W / 5, H / 5), tween=['easeOutQuad'], ytween=['easeOutBounce']),
        PointAnimation((-W / 5, H / 5), (W / 5, -H / 3), tween=['easeOutQuad'], ytween=['easeOutQuad']),
        PointAnimation((W / 5, -H / 3), (-W / 5, -H / 3), tween=['easeOutQuad'], ytween=['easeOutBounce']),
    ], timeweight=[1, 0.5, 0.5, 1], repeats=3)
}

# define some texts
# each line of text must be defined separately, and can be styled and animated separately

# text svg template must take include x="{{x}}" y="{{y}}" properties for an animation to have effect
my_text = """<text {normal} x="{{x}}" y="{{y}}">Outline text <tspan {h1}>with</tspan> <tspan {h2}>highlights</tspan>!</text>""".format(
    **my_styles)
# render clip with text line
txt_clip = moviepy.video.VideoClip.VideoClip(
    make_frame=make_frame_with_animation_and_text(anim['top_to_bottom'], 0.25, duration, my_text), \
    duration=duration / 2)
# make black (= background) transparent so we can overlay text lines later on
txt_clip = moviepy.editor.vfx.mask_color(txt_clip, color=[0, 0, 0])  

my_2nd_text = """<text {normal} x="{{x}}" y="{{y}}">Follow-up <tspan {h3}>to</tspan> <tspan {h1}>first part?!</tspan></text>""".format(
    **my_styles)
_2nd_txt_clip = moviepy.video.VideoClip.VideoClip(
    make_frame=make_frame_with_animation_and_text(anim['topright_to_left'], 0.1, duration, my_2nd_text), \
    duration=duration / 2)
_2nd_txt_clip = moviepy.editor.vfx.mask_color(_2nd_txt_clip,
                                              color=[0, 0, 0])  # make black transparent so we can overlay clips

my_3rd_text = """<text {small} x="{{x}}" y="{{y}}">MOVIEPY ROCKS!</text>""".format(**my_styles)
_3rd_txt_clip = moviepy.video.VideoClip.VideoClip(
    make_frame=make_frame_with_animation_and_text(anim['multi'], 1.0, duration, my_3rd_text), \
    duration=duration
)
_3rd_txt_clip = moviepy.editor.vfx.mask_color(_3rd_txt_clip,
                                              color=[0, 0, 0])  # make black transparent so we can overlay clips

# Overlay all text clips for some fancy animation
video = CompositeVideoClip([txt_clip, _2nd_txt_clip, _3rd_txt_clip])

# Write video file with the result
video.write_gif("test_with_svg.gif", fps=fps)

see: Example of what this produces.

if you extend the script with font size animation, you can do stuff like: this example