You are phenomenal. I have watched some of your prior videos but never implemented anything other than combining audio/video files. It's time to take advantage of your hard work and establish a solid ffmpeg suite on my Windows 10. Other than the hour-long video from 2021, what videos should I watch to create the ffmpeg suite so that I can automatically split clips like you are doing? Also, are you using Linux or a Windows-based terminal in your videos? Thank you for your time and the production of such informative videos.
Hi Mate, cheers thanks a lot for the feedback I would recommend using wsl on windows if possible to install ffmpeg and the scripts , if you cant install wsl you can run the scripts with cygwin but i couldnt get the makefile to work with cygwin so if you use cygwin you have to do scene detection manually using the 3 scripts shown in the video have a look at the "Windows run shell scripts with wsl" and "Windows scene detection with cygwin" videos, that should get you started i also left a link to the whole ffmpeg playlist once you have ffmpeg and the scripts installed they are very easy to use, all the scripts can be run with the -h option to print out the help if you have a look at the ffmpeg scripts read me page it should make sense, so dont feel like you need to watch all the videos you can just referrer to them if needed im using alacritty on linux Fedora 38 but the scripts will work fine with the windows terminal github.com/alacritty/alacritty Videos Windows run shell scripts with wsl ua-cam.com/video/N8t9W3u6cxc/v-deo.html Windows scene detection with cygwin ua-cam.com/video/hu8ZnzIsEmQ/v-deo.html ffmpeg scripts playlist ua-cam.com/play/PL7hhhG5qUoXlVUkHAhgw_px_x0zB29okR.html Text ffmpeg scripts read me github.com/NapoleonWils0n/ffmpeg-scripts all my tech notes github.com/NapoleonWils0n/cerberus ffmpeg tips github.com/NapoleonWils0n/ffmpeg-scripts/blob/master/ffmpeg-tips.org more ffmpeg tips github.com/NapoleonWils0n/cerberus/tree/master/ffmpeg let me know if you need any help getting everything set up
Hello, great great video but a lot of it goes over my head. I want to be able to split videos up via time stamps (which I get from a csv file) and rename them based on info in the csv file. Is that possible with ffmpeg? My project is analysing football games, there is an app that allows me to record the timestamps for the video, i can export to a csv file. Then I think ffmpeg can cut up the video for me? Any help would be greatly appreciated
Hi Mate, you can do exactly what you want with ffmpeg have a csv file with timestamps and use ffmpeg to split the video into clips this is exactly what i designed the scripts for to chop up things like football matches heres how the scene detection works it analyse either the whole video or a portion of a video eg from 5 minutes to 10 minutes and create clips for every scene change eg every time the camera cuts you dont need to use another app, the reason is ffmpeg needs the timestamps in a certain format ffmpeg can can analyse all the action in the match and create the clips you can automate the whole process with a makefile so you only have to run one command and the video will be automatically split into clips its also easy to create a clip that contains several scene changes or camera cuts eg someone crossing the ball and then another camera shot of someone heading the ball in heres how the filename are created with my scripts for example if the video is called "newcastle-vs-liverpool.mp4" the clips will have the same filename but with start and end time of the clip appened the filename for a clip from 5 minutes to 6 minutes would look like this newcastle-vs-liverpool-[00:05:00-00:06:00].mp4 so that means you can see exactly what part of the match the clip is from timestamp format ffmpeg needs the timestamps to be in a particular format start time, duration so to create a clip with a start time 5 minutes for 5 minutes 5 minutes to 10 minutes the format is the start time and the duration 00:05:00,00:05:00 and not 00:05:00,00:10:00 that would create a clip with a start time of 5 minutes and a duration of 10 minutes also my scripts dont use a header in the csv file they should only contain the timestamps if the the csv file from the app you use creates timestamps with start time, duration then you can use the scene-cut script with that file to create the clips if the csv file from the app has timestamps with start time, end time then i have a script that will convert the timestamps into the format ffmpeg expects but you need to put each timestamp on a seperate links so you need to change the csv file from 00:00:00,00:00:11.875 00:00:15.75,00:00:23.042 to a new file with each timestamp on a seperate line 00:00:00 00:00:11.875 00:00:15.75 00:00:23.042 then you can use the clip-time script and it will create a new csv file with the timestamps in the format ffmpeg expects then you can use the scene-cut script to chop the video up with the csv file sexagesimal-time is a script to convert timestamps from start time, end time to start time, duration trim-clip trim-clip can be used to create a single clip from video using timestamps
@@NapoleonWils0n I've never seen such a detailed answer to a youtube comment - thank you for your time. My plan is to have one camera and either it will stay in one position for the entire video or I will have it ZOOM in when triggered. So for example, I could do a skills assessment - say 10 x skills and then chop the video up. My app will trigger the start and finish of each skill. I was playing around with the app and the timestamps aren't great (in the same cell it has the start time and duration I think) but I think they can be sorted out via some formulas to get the required format for ffmpeg. I suppose my question is - do you do freelance/project work?
ffmpeg may detect zooming as a scene change then you could easily create clips for evey zoom change start time, duration is what you want for ffmpeg you can also just create a csv file with the timestamps manually yes i freelance if you get stuck if you play around with the scripts you should get the hand of it
@@NapoleonWils0n can ffmpeg pick up on writing/overlays on the screen? I'm all for more automation. All the companies tell me its not possible to do what I want live/automated. The app will give timestamps. Its just I don't fully understand how to optimise it. My ultimate goal is to have a database linking back to the videos which can be filtered by skills/test and some other data based on the person in the video
ffmpeg might pick up overlays text as a scene, couldnt say for sure you can use emacs and org mode to create a text file with timestamps for video and open the video to that point with mpv emacs and mpv ua-cam.com/video/xBp5uLw-ns8/v-deo.html
You are phenomenal. I have watched some of your prior videos but never implemented anything other than combining audio/video files. It's time to take advantage of your hard work and establish a solid ffmpeg suite on my Windows 10. Other than the hour-long video from 2021, what videos should I watch to create the ffmpeg suite so that I can automatically split clips like you are doing? Also, are you using Linux or a Windows-based terminal in your videos? Thank you for your time and the production of such informative videos.
Hi Mate, cheers thanks a lot for the feedback
I would recommend using wsl on windows if possible to install ffmpeg and the scripts ,
if you cant install wsl you can run the scripts with cygwin but i couldnt get the makefile to work with cygwin
so if you use cygwin you have to do scene detection manually using the 3 scripts shown in the video
have a look at the "Windows run shell scripts with wsl" and "Windows scene detection with cygwin" videos,
that should get you started i also left a link to the whole ffmpeg playlist
once you have ffmpeg and the scripts installed they are very easy to use,
all the scripts can be run with the -h option to print out the help
if you have a look at the ffmpeg scripts read me page it should make sense,
so dont feel like you need to watch all the videos you can just referrer to them if needed
im using alacritty on linux Fedora 38 but the scripts will work fine with the windows terminal
github.com/alacritty/alacritty
Videos
Windows run shell scripts with wsl
ua-cam.com/video/N8t9W3u6cxc/v-deo.html
Windows scene detection with cygwin
ua-cam.com/video/hu8ZnzIsEmQ/v-deo.html
ffmpeg scripts playlist
ua-cam.com/play/PL7hhhG5qUoXlVUkHAhgw_px_x0zB29okR.html
Text
ffmpeg scripts read me
github.com/NapoleonWils0n/ffmpeg-scripts
all my tech notes
github.com/NapoleonWils0n/cerberus
ffmpeg tips
github.com/NapoleonWils0n/ffmpeg-scripts/blob/master/ffmpeg-tips.org
more ffmpeg tips
github.com/NapoleonWils0n/cerberus/tree/master/ffmpeg
let me know if you need any help getting everything set up
Hello, great great video but a lot of it goes over my head. I want to be able to split videos up via time stamps (which I get from a csv file) and rename them based on info in the csv file. Is that possible with ffmpeg? My project is analysing football games, there is an app that allows me to record the timestamps for the video, i can export to a csv file. Then I think ffmpeg can cut up the video for me?
Any help would be greatly appreciated
Hi Mate, you can do exactly what you want with ffmpeg
have a csv file with timestamps and use ffmpeg to split the video into clips
this is exactly what i designed the scripts for to chop up things like football matches
heres how the scene detection works
it analyse either the whole video or a portion of a video eg from 5 minutes to 10 minutes
and create clips for every scene change eg every time the camera cuts
you dont need to use another app, the reason is ffmpeg needs the timestamps in a certain format
ffmpeg can can analyse all the action in the match and create the clips
you can automate the whole process with a makefile so you only have to run one command
and the video will be automatically split into clips
its also easy to create a clip that contains several scene changes or camera cuts
eg someone crossing the ball and then another camera shot of someone heading the ball in
heres how the filename are created with my scripts
for example if the video is called "newcastle-vs-liverpool.mp4"
the clips will have the same filename
but with start and end time of the clip appened
the filename for a clip from 5 minutes to 6 minutes would look like this
newcastle-vs-liverpool-[00:05:00-00:06:00].mp4
so that means you can see exactly what part of the match the clip is from
timestamp format
ffmpeg needs the timestamps to be in a particular format
start time, duration
so to create a clip with a start time 5 minutes for 5 minutes
5 minutes to 10 minutes the format is the start time and the duration
00:05:00,00:05:00
and not
00:05:00,00:10:00
that would create a clip with a start time of 5 minutes and a duration of 10 minutes
also my scripts dont use a header in the csv file
they should only contain the timestamps
if the the csv file from the app you use creates timestamps with
start time, duration
then you can use the scene-cut script with that file to create the clips
if the csv file from the app has timestamps with
start time, end time
then i have a script that will convert the timestamps into the format ffmpeg expects
but you need to put each timestamp on a seperate links
so you need to change the csv file from
00:00:00,00:00:11.875
00:00:15.75,00:00:23.042
to a new file with each timestamp on a seperate line
00:00:00
00:00:11.875
00:00:15.75
00:00:23.042
then you can use the clip-time script and it will create a new csv file
with the timestamps in the format ffmpeg expects
then you can use the scene-cut script to chop the video up with the csv file
sexagesimal-time is a script to convert timestamps from
start time, end time to start time, duration
trim-clip
trim-clip can be used to create a single clip from video using timestamps
@@NapoleonWils0n I've never seen such a detailed answer to a youtube comment - thank you for your time. My plan is to have one camera and either it will stay in one position for the entire video or I will have it ZOOM in when triggered. So for example, I could do a skills assessment - say 10 x skills and then chop the video up. My app will trigger the start and finish of each skill. I was playing around with the app and the timestamps aren't great (in the same cell it has the start time and duration I think) but I think they can be sorted out via some formulas to get the required format for ffmpeg.
I suppose my question is - do you do freelance/project work?
ffmpeg may detect zooming as a scene change
then you could easily create clips for evey zoom change
start time, duration is what you want for ffmpeg
you can also just create a csv file with the timestamps manually
yes i freelance if you get stuck
if you play around with the scripts you should get the hand of it
@@NapoleonWils0n can ffmpeg pick up on writing/overlays on the screen?
I'm all for more automation. All the companies tell me its not possible to do what I want live/automated.
The app will give timestamps. Its just I don't fully understand how to optimise it.
My ultimate goal is to have a database linking back to the videos which can be filtered by skills/test and some other data based on the person in the video
ffmpeg might pick up overlays text as a scene, couldnt say for sure
you can use emacs and org mode to create a text file with timestamps for video
and open the video to that point with mpv
emacs and mpv
ua-cam.com/video/xBp5uLw-ns8/v-deo.html