Python functions used for Swain healing video: https://streamable.com/7041mg

swain.py

"""
Code used to make Swain video wit healing annotation. Nothing special, just manual frame annotation (30 mins).
CV would have been over-engineering IMO.

pip install streamlit
pip install numpy
pip install opencv-python==4.5.5.62

Read the bottom of the code for run instructions.
"""

import cv2
import streamlit as st
from streamlit import session_state as state
import os
import glob
import numpy as np


def write_video_as_pngs(video_file, output_dir):
    """Step 1, split video into frames"""
    if not os.path.isdir(output_dir):
        os.mkdir(output_dir)
    cap = cv2.VideoCapture(video_file)
    frame_count = 0
    total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
    while cap.isOpened():
        ret, frame = cap.read()
        if not ret:
            break
        output_path = os.path.join(output_dir, f"{frame_count}.png")
        if not os.path.exists(output_path):
            cv2.imwrite(output_path, frame)
        frame_count += 1
        if frame_count % 100 == 0:
            print(f"Processed {frame_count}/{total_frames} frames")

    print("Done")


def save_annotation(file_name):
    """
    Helper used in step 2.
    Callback function called when the Streamlit frame annotation form is submitted.
    """
    text_input = state.text_input
    print(file_name, text_input)
    if text_input is not None and text_input != "":
        with open("healing.txt", 'a') as f:
            f.write(f"{os.path.basename(file_name)}:{text_input}\n")
    state.index += 1


def streamlit_annotation():
    """
    Step 2. Creates a Streamlit app that allows you to annotate frames. Just type the healing amount on the first
    frame that you see it and hit the Submit button. The healing amount will be written to healing.txt and the
    next frame will be shown. Repeat until the end of the video.
    """
    if "index" not in state:
        state['index'] = 0 # can change to another starting frame if you were interrupted while running the script
    dir = os.path.join(os.getcwd(), "output")
    files = sorted(glob.glob(f'{dir}/*.png'), key=len)

    form = st.form("hi")
    file_name = files[state.index]
    img = cv2.imread(file_name)
    form.image(img, channels="BGR")
    form.text_input(f"Healing for frame {state.index}", key="text_input")

    form.form_submit_button("Submit", on_click=save_annotation, args=[file_name])


def generate_video_with_text(original, healingtxt, output_name):
    """Step 3. Uses healing.txt to write healing text to a video. This video was overlaid onto the original in After Effects. """
    with open(healingtxt, 'r') as f:
        lines = f.readlines()
    lines = [line.strip() for line in lines]
    lines = [line.split(':') for line in lines]
    lines = {line[0]: line[1] for line in lines}

    cap = cv2.VideoCapture(original)
    fps = cap.get(cv2.CAP_PROP_FPS)
    frame_count = 0
    # total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
    fourcc = cv2.VideoWriter_fourcc(*'mp4v')
    out = cv2.VideoWriter(output_name, fourcc, fps, (1920, 1080))
    files = sorted(glob.glob(f'output/*.png'), key=len)
    total_healing = 0
    for file in files:
        frame_count += 1
        if frame_count % 100 == 0:
            print(f"Processed {frame_count} frames")
        white_frame = np.zeros((1080, 1920, 3), np.uint8)
        white_frame[:] = (255, 255, 255)
        # write total healing onto the frame
        cv2.putText(white_frame, f"Total Healing: {total_healing}", (50, 100), cv2.FONT_HERSHEY_SIMPLEX, 3, (0, 0, 0), 5)
        out.write(white_frame)
        if os.path.basename(file) in lines:
            total_healing += int(lines[os.path.basename(file)])


if __name__ == '__main__':
    # uncomment one step at a time to run
    # for step 2, you need to run `streamlit run main.py` instead of the normal `python main.py`

    # step 1: break video into pngs. Put your own video path ig if you're using this script. My video
    # had the green health bar from the bottom left of the replay motion tracked onto the usual health bar 
    # using After effects. This took two separate recordings of the LoL replay.
    write_video_as_pngs("swain_with_bar.mov", "output")

    # step 2: show all the frames and annotate them in streamlit
    # st.set_page_config(layout="wide")
    # streamlit_annotation()

    # step 3: generate opencv video with text from healing.txt
    # generate_video_with_text("swain_with_bar.mov", "healing.txt", "output.mp4")