Which is quicker in file export?

Is it quicker to export multiple clips set to 7x speed so a 17 min video total, or export multiple clips set to 3x speed so 30 mins total video?

I was assuming 7x but just tried it and it seems to be slower so not sure

Only my guess: for the encoding process the machine has to go through all the videos and do the calculation - in both cases the same videos but with different output settings to generate a shorter (7xspeed) or longer (3xspeed) video. The file sizes will be different but the calculation effort will be nearly the same - so there might be no big difference in encoding time. It depends more on the quality settings of the final output than the speed, i guess.

1 Like

We can theorize all we want, but our theories can’t compete with reality. If you tried it and 7x was slower, then that’s the final answer. :smile:

Theory time…

To get 17 minutes of video when the source is playing at 7x, that means the source has to be 17 * 7 = 119 minutes long.

Similarly, the source video needs to be 30 * 3 = 90 minutes long for 3x to produce a 30 minute video.

In the 3x example, Shotcut has to process 90 minutes of source video. In the 7x example, Shotcut has to process 119 minutes of source video. That’s 33% more source video processing. “Processing” includes seeking and reconstructing frames within video formats that have very complicated and heavy compression. If the encoder is time efficient (such as hardware encoding), but the source videos are not time efficient (like HEVC/H.265 from a 4K cell phone), then that 33% difference can be a big deal.

There are many variables. But your test results have the final say.

1 Like

I just did a very simple test. Imported a 12m 20s mp4; no filters applied, then exported it at 4 different speeds:

  1x speed, it took 5m 57s (filesize: 23.75 MB)
  5x speed, it took 2m 30s (filesize: 10.37 MB)
 10x speed, it took 2m  0s (filesize:  7.50 MB)
 20x speed, it took 1m 15s (filesize:  5.40 MB)

This works out at about 4MB/minut processing time.

1 Like

Great test! This proves the results are not linear. 20x speed doesn’t mean 1/20th the export time of 1x. Otherwise, the export would be done in 18 seconds rather than 1m 15s. Something is causing diminishing returns. I’m still betting the issue is Shotcut having to uncompress the entire source video even if only 1/20th of it is used. The primary way around this issue is to have all-intra source videos.

I suspect it is I-frame and B-frame processing. It the original video had a GOP that looked something like that shown in the “GOP row below” i.e. 1 I-frame followed by 4 B-frames then in 1x speed the changes from frame to frame would be small (s), hence a small amount of processing. In 5x the changes between consecutive frames would be large (l), as you would actually be comparing the 1st I-frame with the 2nd I-frame etc. In 10x the changes between consecutive frames would be huge (h), as you would actually be comparing the 1st I-frame with the 3rd I-frame etc… (F=full frame processing)

Frame  0 1 2 3 4 5 6 7 8 9 10
GOP    I B B B B I B B B B I
 1x    F s s s s F s s s s F
 5x    F l l l l F l l l l F
10x    F h h h h F h h h h F

So even though you are throwing away 4 fifths of the frames for 5x and 9 tenths of the frames for 10x, the frames that you do process are a lot different from each other consecutively, hence a lot more compression needs to be done. I’m sure that this is the reason why the file-size, and hence processing time, does not decrease linearly with the speedup.

Which source video types are most efficient? I’m putting together multiple videos from a canon eos and exporting as a h.264 from shotcut. Then putting the exported vid back into shotcut to edit and export the final again as h.264.

Is there a more efficient type to export the pre-edited content? I’ve been using h.264 as i read it’s most compatible on all device types

H.264 can work for the pre-edited video. Device compatibility isn’t a concern at this point of the process because only your computer sees the pre-edited video format. You’re free to use other formats or fancy settings up until the final export.

If using H.264 for speed changes like you’re doing, it can be made more efficient to decode by going to the Export tab > Advanced button > Codec tab, then change GOP to 30 and B-frames to 0. This will make the exported file equally efficient to your Canon EOS clips, and probably around the same file size. If you want the pre-edited video to seek and playback and export on your final timeline even faster, then change GOP to 1 and B-frames to 0. Note this will create a noticeably larger file.

These changes probably won’t speed up the export of the pre-edited video. This only makes the pre-edited video faster to work with on the final timeline. There isn’t much we can do about this because we would have to change the format written by the Canon camera itself in order to improve decode time further. And a Canon camera doesn’t have any better formats than what you’re already using unless it’s a newer RF-mount model, or an external recorder is used.

Since the pre-edited video is going to be re-encoded again for the final export, it helps to have higher-than-usual quality settings for the pre-edited video, such as 68% quality (CRF 16). This reduces double-degradation by the time these clips hit the final export. All other settings for the pre-edited export can stay the same.

A video at 7x speed sounds a lot like a time lapse. Since you have a Canon, have you experimented with using photo mode’s time lapse timer to take a picture once per second with a shutter speed of 1/8th, then bundle the pictures into a pre-edited video? The slower shutter speed allows a significant amount of connective motion blur in the scene as objects move. So instead of people flickering from one location to the next like they’re under a strobe light, the 1/8th shutter makes people look more like streaky ghosts that are effortlessly flowing around the room. I’ve found this method to produce much more comprehensible and easy-to-watch summaries of activity, like a building project in progress. But if you’re just fast-forwarding through a long car ride, then it may not be the best option unless the camera is mounted rock-steady. Just an idea. Good luck!

1 Like

Thats all so helpful thank you so much! And I’ll definitely try the time lapse setting next time

1 Like

This topic was automatically closed after 90 days. New replies are no longer allowed.