Camcorder video of ice hockey - gap stitching

[p9g]

I have a Samsung SC-MX10 camcorder that I use to record ice hockey games. I typically start recording when the puck is dropped, then stop recording when there is a whistle stopping the play. This camcorder creates an indvidual avi file each time. So, lets say the game starts at [T1] 8:00:00am and I start recording, then stop at [T2] 8:00:45 (45 seconds of play), then start again at [T3] 8:01:10 and stop at [T4] 8:02:20 (70 seconds of play). I would get two avi files (F1, F2) whose date/time last modified would be [T2] and [T4], respectively. For three 15-minute periods, I get around 50 files that total about 45 minutes. The actual clock time would be about an hour.

Here is where it gets interesting... I have an audio recording (mp3) of a sports announcer. He continues to talk continuously for the whole 60 minutes; he does not stop talking when I stop recording.

Now I would like to create a single avi file that combines my video (and some of the original audio perhaps) and the complete (60-minutes) announcer's audio. So, I would need to fill in video, in the example above, for the time when I was not recording; i.e., 8:00:45 to 8:01:10, or 25 seconds for the first gap (G1) between files F1 and F2.

The avi file has a date/time last modified, which I assume is probably when I stopped recording (e.g., [T2] 8:00:45 for the first file, [T4] 8:02:20 for the next, and so on).

I don't think the avi files contain any embedded metadata that would give me the date/time that the video started or ended, so I only have the last modified date/time of the file to work with. However, I would need to look inside (parse) the avi file to find the duration of the video in that file.

Lets say I have 50 avi files at the end of the game that total about 45 minutes.

One problem is how to calculate the duration of each of the 49 gaps of time between the 50 video files.

If I can do that, then I can effectively create 49 avi files of different sizes, then stitch them together to have a continuous video, then overlay the announcer's audio so his play-by-play is in sync with the video.

If my video files are F1, F2, ... F50, and I derive "gaps" G1, G2, ... G49, the final sequence would be F1, G1, F2, G2, F3, G3 .... F49, G49, F50.

I am looking for a way to extract the file date/time last modified (e.g., [T4] for F2) and substract the time of the video in that file (e.g., 70 seconds); I would do a calculation to get [T3]. I then need to subtract [T3] from [T2] to get the time for my filler video segment (G1).

So, in summary, I have about 45 minutes of video in 50 avi files (4CC = SEDG), and announcer (mp3) audio of about 60 minutes. I need to create about 15 minutes of "gap filler" video divided among 49 segments of different durations. I need to merge/stitch the recorded video with the gap filler video to create a single avi with 60 minutes of video and 60 minutes of audio. For now I don't care if the gap filler video is just a black screen; maybe later I can replace some of the gaps with instant slow motion replays.

Any suggestions?

[guns1inger]

I would use Vegas (because I have it, and because it is ideal for anything requiring this type of audio work) and put down the commentator track first. I would load each avi file in turn and sync each to the appropriate section of audio. At the end of this process, which would not take very long, even with 50 or so segments), I would have a single video and 2 audio tracks. I would then use the volume envelope on the video audio to fade it in and out over the commentator track, and output the two as a single video file. Any gaps in the video would be output as black sections, but this can be changed if necessary.

You can probably do the same in most NLE's, but I just happen to like the way Vegas does it. There is no simple way to automate this process. It would be virtually impossible to do in something like virtualdub or avidemux unless you can create black video segments of exactly the length of your gaps as you go along.

[jagabo]

Your files probably have creation date/times in addition to last modified date/times.