I generated a character video of Cai Xukun singing, dancing and rapping, the link is:
Cai Xukun singing and dancing RAP character video
convert a video to character video using python
Detailed video tutorial, 8 minutes in total.
A detailed tutorial on making character videos with python (video version)
See text tutorial below:
First you need to install python
Then install the required packages: cv2, PIL
Specific installation method:
install cv2
pip install opencv-pythonInstall PIL
pip install pillowDuring the installation process, we often encounter problems such as long installation time and installation failure.
You can download the corresponding version of the whl file from the official website and use the local installation method.
Opencv-python official download address
Pillow official download address
When downloading, you need to pay attention to whether your computer is 32-bit or 64-bit.
You need to pay attention to the version of python installed on your computer.
When choosing opencv_python and pillow, it must be corresponding.
In addition, you need to download the ffmpeg package , which is used to extract the sound in the video and add the sound to the generated character video.
If ffmpeg is not configured, the generated video will have no sound.
When downloading, also pay attention to whether your computer is 32-bit or 64-bit.
ffmpeg32-bit and 64-bit download address
After downloading the ffmpeg package, you need to configure environment variables.
After everything is ready, you can run the python code to generate the character video. The python code is as follows.
import argparse
import os
import cv2
import subprocess
from cv2 import VideoWriter_fourcc
from PIL import Image, ImageFont, ImageDraw
class Video2CodeVideo:
def __init__(self):
self.config_dict = {
# 原视频文件
"input_file": "test.mp4",
# 中间文件存放目录
"cache_dir": "cache",
# 是否保留过程文件。True--保留，False--不保留
"save_cache_flag": True,
# 使用使用的字符集
"ascii_char_list": list("01B%8&WM#*oahkbdpqwmZO0QLCJUYXzcvunxrjft/\|()1{}[]?-_+~<>i!lI;:oa+>!:+. "),
}
# 第一步从函数，将像素转换为字符
# 调用栈：video_2_txt_jpg -> txt_2_image -> rgb_2_char
def rgb_2_char(self, r, g, b, alpha=256):
if alpha == 0:
return ''
length = len(self.config_dict["ascii_char_list"])
gray = int(0.2126 * r + 0.7152 * g + 0.0722 * b)
unit = (256.0 + 1) / length
return self.config_dict["ascii_char_list"][int(gray / unit)]
# 第一步从函数，将txt转换为图片
# 调用栈：video_2_txt_jpg -> txt_2_image -> rgb_2_char
def txt_2_image(self, file_name):
im = Image.open(file_name).convert('RGB')
# gif拆分后的图像，需要转换，否则报错，由于gif分割后保存的是索引颜色
raw_width = im.width
raw_height = im.height
width = int(raw_width / 6)
height = int(raw_height / 15)
im = im.resize((width, height), Image.NEAREST)
txt = ""
colors = []
for i in range(height):
for j in range(width):
pixel = im.getpixel((j, i))
colors.append((pixel[0], pixel[1], pixel[2]))
if (len(pixel) == 4):
txt += self.rgb_2_char(pixel[0], pixel[1], pixel[2], pixel[3])
else:
txt += self.rgb_2_char(pixel[0], pixel[1], pixel[2])
txt += '\n'
colors.append((255, 255, 255))
im_txt = Image.new("RGB", (raw_width, raw_height), (255, 255, 255))
dr = ImageDraw.Draw(im_txt)
font = ImageFont.load_default().font
x = y = 0
# 获取字体的宽高
font_w, font_h = font.getsize(txt[1])
font_h *= 1.37 # 调整后更佳
# ImageDraw为每个ascii码进行上色
for i in range(len(txt)):
if (txt[i] == '\n'):
x += font_h
y = -font_w
dr.text((y, x), txt[i], fill=colors[i])
y += font_w
name = file_name
im_txt.save(name)
# 第一步，将原视频转成字符图片
# 调用栈：video_2_txt_jpg -> txt_2_image -> rgb_2_char
def video_2_txt_jpg(self, file_name):
vc = cv2.VideoCapture(file_name)
c = 1
if vc.isOpened():
r, frame = vc.read()
if not os.path.exists(self.config_dict["cache_dir"]):
os.mkdir(self.config_dict["cache_dir"])
os.chdir(self.config_dict["cache_dir"])
else:
r = False
while r:
cv2.imwrite(str(c) + '.jpg', frame)
self.txt_2_image(str(c) + '.jpg') # 同时转换为ascii图
r, frame = vc.read()
c += 1
os.chdir('..')
return vc
# 第二步，将字符图片合成新视频
def txt_jpg_2_video(self, outfile_name, fps):
fourcc = VideoWriter_fourcc(*"MJPG")
images = os.listdir(self.config_dict["cache_dir"])
im = Image.open(self.config_dict["cache_dir"] + '/' + images[0])
vw = cv2.VideoWriter(outfile_name + '.avi', fourcc, fps, im.size)
os.chdir(self.config_dict["cache_dir"])
for image in range(len(images)):
frame = cv2.imread(str(image + 1) + '.jpg')
vw.write(frame)
os.chdir('..')
vw.release()
# 第三步，从原视频中提取出背景音乐
def video_extract_mp3(self, file_name):
outfile_name = file_name.split('.')[0] + '.mp3'
subprocess.call('ffmpeg -i ' + file_name + ' -f mp3 -y ' + outfile_name, shell=True)
# 第四步，将背景音乐添加到新视频中
def video_add_mp3(self, file_name, mp3_file):
outfile_name = file_name.split('.')[0] + '-txt.mp4'
subprocess.call('ffmpeg -i ' + file_name + ' -i ' + mp3_file + ' -strict -2 -f mp4 -y ' + outfile_name, shell=True)
# 第五步，如果没配置保留则清除过程文件
def clean_cache_while_need(self):
# 为了清晰+代码比较短，直接写成内部函数
def remove_cache_dir(path):
if os.path.exists(path):
if os.path.isdir(path):
dirs = os.listdir(path)
for d in dirs:
if os.path.isdir(path + '/' + d):
remove_cache_dir(path + '/' + d)
elif os.path.isfile(path + '/' + d):
os.remove(path + '/' + d)
os.rmdir(path)
return
elif os.path.isfile(path):
os.remove(path)
return
# 为了清晰+代码比较短，直接写成内部函数
def delete_middle_media_file():
os.remove(self.config_dict["input_file"].split('.')[0] + '.mp3')
os.remove(self.config_dict["input_file"].split('.')[0] + '.avi')
# 如果没配置保留则清除过程文件
if not self.config_dict["save_cache_flag"]:
remove_cache_dir(self.config_dict["cache_dir"])
delete_middle_media_file()
# 程序主要逻辑
def main_logic(self):
# 第一步，将原视频转成字符图片
print("第一步,正在将原视频转成字符图片")
vc = self.video_2_txt_jpg(self.config_dict["input_file"])
# 获取原视频帧率
fps = vc.get(cv2.CAP_PROP_FPS)
print("获取原视频帧率:")
print(fps)
vc.release()
print("已将原视频转成字符图片\n")
# 第二步，将字符图片合成新视频
print("第二步,正在将字符图片合成新视频")
self.txt_jpg_2_video(self.config_dict["input_file"].split('.')[0], fps)
print(self.config_dict["input_file"], self.config_dict["input_file"].split('.')[0] + '.mp3')
print("已将字符图片合成新视频\n")
# 第三步，从原视频中提取出背景音乐
print("第三步, 正在从原视频中提取出背景音乐")
self.video_extract_mp3(self.config_dict["input_file"])
print("已从原视频中提取出背景音乐\n")
# 第四步，将背景音乐添加到新视频中
print("第四步, 正在将背景音乐添加到新视频中")
self.video_add_mp3(self.config_dict["input_file"].split('.')[0] + '.avi', self.config_dict["input_file"].split('.')[0] + '.mp3')
print("已将背景音乐添加到新视频中\n")
# 第五步，如果没配置保留则清除过程文件
self.clean_cache_while_need()
print("字符视频制作完毕\n字符视频为test-txt.mp4\n")
print("按任意键结束")
a=input()
if __name__ == '__main__':
obj = Video2CodeVideo()
obj.main_logic()
Note that the source video and python code should be placed in the same directory, and the original video should be named test.mp4. After the operation, a test-txt.mp4 file will be generated, which is the character video we want.
If you have any questions, welcome to communicate