- 该项目是使用 python 对 FFmpeg 的类库做了一次封装使用,因最近不搞视频,所以很长时间不维护了,大家可以参照写法,使用 python 构建自己的视频处理服务(Dont forget GPU)。
ffmpeg 3.0 及以上 ffprobe 3.0 及以上 python 3.0 及以上
mac,linux,windows 相应的 gpu 显卡驱动 (使用硬编码时需支持)
pip install ffmpeg
from ffmpeg import video
# 输入视频
input_file = "demo.mp4"
# 输出视频
out_file = "demo_out.mp4"
# 图片列表
img_data = [
{
"img": "demo1.png",
"x": "",
"y": "",
"str_time": "5",
"end_time": "15",
},
{
"img": "demo2.png",
"x": "",
"y": "",
"str_time": "20",
"end_time": "25.5"
}
]
video.ins_img(input_file, img_data, out_file)
from ffmpeg import video
input_file = "demo.mp4"
out_file = "demo_out.mp4"
img_data = {
"img": "img.apng",
"x": "20",
"y": "20",
"str_time": "2",
"end_time": "10"
}
video.ins_dynamic_img(input_file, img_data, out_file)
from ffmpeg import image
image.img_trans_video("png/text_%02d.jpg", "5", "out.mp4")
from ffmpeg import stream
stream = Stream()
# 输入文件
stream.input(input_file)
# 图片
stream.img("t1.png")
stream.img("t2.png", "10", y=10, str_time=5, end_time=10)
# 动图
stream.img_dynamic("t1.apng", "10", "10", "5", "10")
stream.img_dynamic("t2.gif", x=10, y=10, str_time=5, end_time=9)
# 文字水印
stream.word_water_mark("test1", x="10", y="10", str_time="0", end_time="20", font="ttf.ttf", color="blue")
stream.word_water_mark("test2", x="10", y="10", str_time="0", end_time="20", font="ttf.ttf", color="blue")
# 字幕
stream.subbtitle("tt.srt")
# 显卡加速,该方式目前只适用于 mac 和 navida, 参数size是视频大小,该值越大,越慢
stream.fast("5M")
# 输出文件
stream.out(out_file)
stream.run()