simple-thumbnail
simple-thumbnail copied to clipboard
philipjscott
Performance Improvement
Having tried to generate a thumbnail of an MP4 video, I noticed that running ffmpeg took a lot of time to retrieve a thumbnail from a given time in the video. Upon inquiring, I found this comment interesting (https://superuser.com/a/821680)
Each invocation of above ffmpeg command takes a fraction of second (!) for ~1GB H.264. That is because it instantly jumps to
Currently in the creation of arguments for ffmpeg the argument -ss is placed after the definition of the source file.
https://github.com/ScottyFillups/simple-thumbnail/blob/44379e88307c1aa2c8efee518dd2c1d4861eb566/index.js#L60-L68
I have therefore rearranged the order of the arguments to put the -ss before the -i as suggested in the comment above.
return [
'-y',
`-ss ${seek}`,
`-i ${input}`,
scaleArg,
'-vframes 1',
output
]
For the following test script with an MP4 file of 23m40s and 188Mb I give you the total execution time of this same script with the original order and the one I propose.
The script:
const ffmpeg = require('ffmpeg-static')
const genThumbnail = require('simple-thumbnail')
async function download () {
await genThumbnail('test.mp4', 'test_s.jpg', '150x?', {
path: ffmpeg,
seek: '00:12:00'
})
console.log('Done!')
}
download()
The characteristics of the MP4 file (the return of ffprobe):
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'test.mp4':
Metadata:
major_brand : isom
minor_version : 512
compatible_brands: isomiso2avc1mp41
encoder : Lavf58.33.100
Duration: 00:23:40.16, start: 0.000000, bitrate: 1113 kb/s
Stream #0:0(und): Video: h264 (Main) (avc1 / 0x31637661), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 915 kb/s, 23.98 fps, 23.98 tbr, 24k tbn, 47.95 tbc (default)
Metadata:
handler_name : VideoHandler
Stream #0:1(und): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 192 kb/s (default)
Metadata:
handler_name : SoundHandler
The execution time of the script with the original order:
PS C:\Users\marti\Documents\test> Measure-Command {node .\index.js}
Days : 0
Hours : 0
Minutes : 0
Seconds : 18
Milliseconds : 188
Ticks : 181882832
TotalDays : 0,000210512537037037
TotalHours : 0,00505230088888889
TotalMinutes : 0,303138053333333
TotalSeconds : 18,1882832
TotalMilliseconds : 18188,2832
The execution time of the script with the proposed order:
PS C:\Users\marti\Documents\test> Measure-Command {node .\index.js}
Days : 0
Hours : 0
Minutes : 0
Seconds : 1
Milliseconds : 480
Ticks : 14806789
TotalDays : 1,71374872685185E-05
TotalHours : 0,000411299694444444
TotalMinutes : 0,0246779816666667
TotalSeconds : 1,4806789
TotalMilliseconds : 1480,6789
The time saving is huge for me, and I see the same time saving if I provide the MP4 file via a URL and not as a local file.
The execution time of the script with the file over the internet with the original order:
PS C:\Users\marti\Documents\test> Measure-Command {node .\index.js}
Days : 0
Hours : 0
Minutes : 0
Seconds : 26
Milliseconds : 18
Ticks : 260180286
TotalDays : 0,000301134590277778
TotalHours : 0,00722723016666667
TotalMinutes : 0,43363381
TotalSeconds : 26,0180286
TotalMilliseconds : 26018,0286
The execution time of the script with the file over the internet with the proposed order:
PS C:\Users\marti\Documents\test> Measure-Command {node .\index.js}
Days : 0
Hours : 0
Minutes : 0
Seconds : 2
Milliseconds : 265
Ticks : 22658406
TotalDays : 2,62250069444444E-05
TotalHours : 0,000629400166666667
TotalMinutes : 0,03776401
TotalSeconds : 2,2658406
TotalMilliseconds : 2265,8406
My tests were performed on a Windows 10 Pro 64bits with an Intel Core i7-10510U.
The order of the parameters definitely seems to have an importance on the performance of ffmpeg, unfortunately I don't have time to do more tests but I'll let you try on your side too.
This is my first issue and I hope I have respected the rules as much as possible 😢
Have a nice day !
