python如何为视频制作一个简单的上限。
Ideia:为9个偶数(在时间轴上)拍摄9张快照,然后在JGP中取代
我该怎么做?
PIL是唯一的方法吗? (做我想做的事情并不容易,对吧?)
是不是有任何模块(python seens有大型模型用于任何事情,或者我运气不好)
示例:
Movie lenght = 10 min
T1= snapshot of 1 min
T2= snapshot of 2 min
......
T9= snapshot of 9 min
| |
T1| T2| T3
---+---+---
T4| T5| T6
---+---+---
T7| T8| T9
| |
答案 0 :(得分:7)
这是对我前段时间做过的快速更改,使用ffmpeg进行帧提取,使用PIL创建完整的拇指图像。
import os, sys
from PIL import Image
# Use "ffmpeg -i <videofile>" to get total length by parsing the error message
chout, chin, cherr = os.popen3("ffmpeg -i %s" % sys.argv[1])
out = cherr.read()
dp = out.index("Duration: ")
duration = out[dp+10:dp+out[dp:].index(",")]
hh, mm, ss = map(float, duration.split(":"))
total = (hh*60 + mm)*60 + ss
# Use "ffmpeg -i <videofile> -ss <start> frame<nn>.png" to extract 9 frames
for i in xrange(9):
t = (i + 1) * total / 10
os.system("ffmpeg -i %s -ss %0.3fs frame%i.png" % (sys.argv[1], t, i))
# Make a full 3x3 image by pasting the snapshots
full = None
for y in xrange(3):
for x in xrange(3):
img = Image.open("frame%i.png" % (y*3+x))
w, h = img.size
if full is None:
full = Image.new("RGB", (w*3, h*3))
full.paste(img, (x*w, y*h))
# Save result
full.save("thumbs.png")