After you have trained your data-set, and are pretty confident with your loss values, you might want to proceed to generating the final video with the machine learning from your app using your model data.
It is better to have a very small video of 3-10 seconds that you will use as a final video (for now), just to test the results.
It is even better if you can squeeze in these 10 seconds more scenes from the final video.
There is no reason to generate the full video (which might be between 5 to 10, even more minutes), because that will take a very long time to process. You can use that time to better to further train your model.